Search

The aim of this article is a critical analysis of the logocentric representation of brain as a conceptual basis for artificial neural networks (ANNs). Neuro-logocentrism turns possible mostly because of some similarities of brain activity to logical reasoning. They are discrete mode of action, semblance of realization of logical connectives and inferences, presence of two “values” (“all-or-none” principle). It is demonstrated that they are only loose analogies. The idea that neurons are net nodes that realize logical and computational operations as quite simple automata is based on the concept of connectionism that appears a new special type of hylemorphism. On the contrary, brain activity demonstrates the primacy of the whole and its irreducibility to the aggregative sum of elements and parts. It appears continual totality, where everything is reciprocally interconnected, so that it is impossible to establish any unified formalism for it. Brain essentially is a part of the living, not “bio-logic” but fully biological reality, which has an emergent mode of existence with its unique properties. Such processes as neuroplasticity and synaptic pruning are responsible for adaptive and teleonomic re‑shaping of the brain. Usual conventional logic is suitable for computing machines and their nets, but it is incapable of reproducing such autopoietic phenomena. As a result, there is a fatal ontological gap between natural networks of neurons and ANNs. The second ones are not able to implement complex cognitive functions as they are product of the logocentric hypertrophy of algorithms which is fully devoid of flexibility of the living brain.

The purpose of this work is development and research of an algorithm for determining the structure of couplings of an ensemble of chaotic self-oscillating systems. The method is based on the determination of causality by Granger and the use of direct propagation artificial neural networks trained with regularization. Results. We have considered a method for recognition structure of couplings of a network of chaotic maps based on the Granger causality principle and artificial neural networks approach. The algorithm demonstrates its efficiency on the example of small ensembles of maps with diffusion couplings. In addition to determining the network topology, it can be used to estimate the magnitue of the couplings. Accuracy of the method essencially depends on the observed oscillatory regime. It effectively works only in the case of homogeneous space-time chaos. Discussion. Although the method has shown its effectiveness for simple mathematical models, its applicability for real systems depends on a number of factors, such as sensitivity to noise, to possible distortion of the waveforms, the presence of crosstalks and external noise etc. These questions require additional research.

In recent decades, neural networks have been widely applied in many fields of science and medicine. Accurate and early diagnosis of malignancies is a key challenge in oncology. Neural networks can analyse a wide range of medical data and identify relationships between qualitative and quantitative features. This allows for more precise and timely diagnoses. Moreover, they can be used to predict tumour progression, evaluate treatment effectiveness, and optimise treatment plans for each patientIn oncourology, the use of neural networks offers new perspectives for the diagnosis, prognosis, and treatment of various cancer conditions related to the urinary tract and male reproductive system. This review article explores how neural networks are being used in this field and present research into the use of neural networks for diagnosing, predicting the course and treating urological oncological diseases. The advantages and limitations of using neural networks in this field are demonstrated, and possible directions for future research are suggested. The application of neural networks in oncourology opens new horizons for the development of a personalised approach to diagnosing and treating oncological diseases. Artificial intelligence has the potential to become a powerful tool for improving the accuracy of patient outcome predictions and reducing undesirable side effects of therapy. Introducing neural networks into oncourological practice creates new opportunities for enhancing the work of healthcare organisations and improving the quality of care provided to patients. This can lead to better treatment outcomes and improved patient satisfaction.

Objective. The goal of the research is to develop an artificial neural network (ANN) to detect and neutralize network DDoS attacks.Method. The research is based on the use of the Python programming language in an environment that supports the training functions of PyCharm neural networks.Result. An analysis of existing artificial neural networks was carried out to determine their optimal structure; Existing methods for detecting network DDoS attacks have been studied; Datasets were collected and refined to improve the quality of training; The structure of the artificial neural network of the classifier was created and its training was carried out, a demonstration software was created that illustrates the process of classification and blocking and neutralizing DDoS attacks.Conclusion. Having systems to monitor traffic, a Web application firewall, speed limiting, a status page, and a company face to answer questions on social media will all help ensure the most effective protection against DDoS attacks.

Background. Carotid atherosclerosis is one of the urgent problems due to the high risk of developing ischemic stroke and cognitive impairment. The dynamics of clinical disorders in patients with carotid stenosis is determined by a complex of neurophysiological, angiological, tissue and biomolecular reactions, the characteristics of which can act as predictors of the course of the pathology.The aim of the work. To determine the neurophysiological parameters and predictors of cognitive dysfunction in patients who underwent carotid endarterectomy.Materials and methods. The study included 59 people with carotid atherosclerotic disease. All included patients underwent carotid endarterectomy. We assessed the degree of stenosis of the internal carotid artery and cognitive status using the FAB (Frontal Assessment Battery) scale and MoCA (Montreal Cognitive Assessment) Test and recorded electroencephalogram (EEG), P300 cognitive evoked potentials and heart rate variability in patients at various terms (before surgery, 6 months after the surgery). Patients were divided into groups based on the dynamics of cognitive tests using cluster analysis (k-means) with identification of elements included in the clusters: patients of cluster 1 had a “preserved” profile of cognitive status; patients of cluster 2 – moderate cognitive dysfunction.Results. Patients of cluster 1 had a higher power of beta oscillations in the frontal lead, a higher amplitude of the P3 component of the P300 potential, and a greater variability of R-R intervals in terms of the total indicator and high-frequency power. We proposed a model that allows us to classify patients into groups according to the dynamics of cognitive function scores. According to the data obtained, the most significant predictors of the dynamics of cognitive status were the initial characteristics of the EEG and the P300 cognitive evoked potential.Conclusions. We determined the clinical and neurophysiological correlates of cognitive dysfunction: an association with greater preservation of activating effects on the EEG, processes of recognition and decision-making in the associative zones of the cortex, and less pronounced activity of stress-implementing mechanisms. Indicators of EEG spectral analysis and characteristics of the P300 cognitive evoked potential are predictors of the cognitive status dynamics.

Significant insights have arisen from an extensive review of the current literature, highlighting the importance of Convolutional Neural Networks (CNNs) in cricket performance analysis and mapping new directions for future research. Despite difficulties such as limited availability of data, processing difficulty, and interpretability issues, incorporating CNNs into cricket statistics is a potential effort made possible by advances in machine learning and deep learning methods. Instructors, players, and data analysts can use CNNs to better comprehend the game, extract meaningful information from video data, and improve decision-making processes. Key findings show that CNNs are effective tools for a variety of cricket analysis tasks involving batting, bowling, fielding, and player tracking. The use of CNNs represents an advancement in cricket analysis, promising to open up new aspects of performance and usher in a data-driven era of cricket genius. Augmenting data, the use of parallelization, explainable AI, and concerns about ethics, provide opportunities to address current challenges can be identified as future advances in sports analysis with CNNs. Embracing technological advancements and mapping out future research directions are critical steps towards realizing this revolutionary potential.

BACKGROUND: Time in range (TIR) is a promising indicator of glycemic control used for evaluation of continuous glucose monitoring (CGM) for patients with diabetes mellitus (DM). The current problem is the assessment and prediction of TIR for patients who use self-monitoring of blood glucose (SМBG) corresponding low CGM availability for the majority of diabetic patients.AIM: To develop a predictive model of TIR for patients with T1DM based on data of the experimental mobile application.MATERIALS AND METHODS: An analysis of 1253 professional CGM profiles of patients with T1DM was performed. On the base of included records, TIR(CGM) was calculated and training models of 7-point SMBG profiles were generated. SMBG profiles’re loaded into the developed experimental mobile application that calculated standard glycemic control parameters. The dataset was divided into main and test samples (80 and 20%). For the main sample, the following methods’re used to develop predictive models: simple linear regression (SLR), multiple linear regression (MLR), artificial neural network (ANN). The effectiveness of the developed models was assessed on the test sample with the calculation of the mean absolute error (MAE), the root mean square error (RMSE).RESULTS: The 568 CGM profiles’re included in the study. TIR in the main group (n=454) — 45 [33; 65]%, in the test group (n=114) — 43 [33; 58]%. The most significant predictors of the regression models were the derived TIR (dTIR), p

Objective: to evaluate the possibility of using artificial neural networks for integration into the medical decision support system as an optimization of outpatient management of patients with COPD.Materials and methods: a dynamic followup of 150 patients with chronic obstructive pulmonary disease, registered at the dispensary for the underlying disease, who completed the outpatient stage of pulmonary rehabilitation after a moderate exacerbation was carried out. The material of the study was a universal questionnaire of 69 indicators, including anamnesis, clinic, laboratory and instrumental diagnostics. A four-layer neural network has been created: the first two layers — 69 neurons, the third layer — 34 neurons and the last layer — 3 neurons.Results: the software was used in the Java programming language using the Encog 3.4 module.Conclusion: the use of the capabilities of artificial neural networks for integration into the medical decision support system in the outpatient management of patients with chronic obstructive pulmonary disease has shown high specificity. The predictive model is implemented in the form of a computer program: "The program for predicting an unfavorable outcome, the development of cardiovascular complications and the effectiveness of rehabilitation measures in patients with chronic obstructive pulmonary disease (CardioRisk)" and was introduced into the work of outpatient polyclinic institutions in Rostov-on-Don.

Background. The article is devoted to the development of an ambulatory personal telemonitoring ECG systems (APTECG systems). The relationship of increasing the reliability of decisions made in APTECG systems with quality control of recorded ECGs is shown. The expediency of including the quality control module of registered ECGs in the APTECG systems is proved. Materials and methods. A block diagram of the APTECG system with a quality control module of registered ECGs in its composition has been developed. Its functions and connections with other modules are defined. The quality control module of registered ECGs is implemented as a neural network binary classifier based on 2D CNN. To represent the ECGs as an image, a wavelet transform was used. A database of images has been created for training and testing neural networks. Results and conclusions. Four deep learning neural networks have been developed in the Python programming language, according to the test results of which the VGGNet19 network showed the best result with an accuracy of 0.97, a logarithmic loss of 0.1 and an F-measure of 0.97. The developed block diagram of the APTECG system and the quality control module of the registered ones will increase the reliability of the diagnostic decisions made.

The research resulted in the development of software that implements various algorithms of neural networks optimization, which allowed to carry out their comparative analysis in terms of optimization quality. The article takes a detailed look at artificial neural networks and methods of their optimization: quantization, overcutting, distillation, Tucker’s dissolution. Algorithms and optimization tools of neural networks were explained, as well as comparative analysis of different methods was conducted with their advantages and disadvantages listed. Calculation values were given as well as recommendations on how to execute each method. Optimization is studied by text classification performance: peculiarities were removed, models were chosen and taught, parameters were adjusted. The set task was completed with the use of the following technologies: Python programming language, Pytorch framework and Jupyter Notebook developing environment. The results that were acquired can be used to reduce the demand on computing power while preserving the same level of detection and classification abilities.

Scientific relevance. Studies of the toxicological and pharmacokinetic properties of medicinal compounds are a crucial stage of preclinical research; unsatisfactory results may invalidate further drug development. Therefore, the development of in silico methods for a preliminary pre-experimental assessment of toxicological and pharmacokinetic properties is a relevant and crucial task.Aim. The study aimed to review current approaches to in silico prediction of the absorption, distribution, metabolism, excretion, and toxicity (ADMET) parameters of pharmacologically active compounds, in particular, the most important toxicological and pharmacokinetic parameters, and to present the results of the authors’ own research in this area.Discussion. According to the review of models for predicting the toxicological properties of chemical compounds (acute toxicity, carcinogenicity, mutagenicity, genotoxicity, endocrine toxicity, cytotoxicity, cardiotoxicity, hepatotoxicity, and immunotoxicity), the accuracy of predictions ranged from 74.0% to 98.0%. According to the review of models for predicting the pharmacokinetic properties of chemical compounds (gastrointestinal absorption; oral bioavailability; volume of distribution; total, renal, and hepatic clearance; and half-life), the coefficient of determination for the predictions ranged from 0.265 to 0.920. The literature review showed that the most widely used methods for in silico assessment of the ADMET parameters of pharmacologically active compounds included the random forest method and the support vector machines method. The authors compared the literature data with the results they obtained by modelling 12 toxicological and pharmacokinetic properties of chemical compounds using the consensus method in the IT Microcosm system and artificial neural networks. IT Microcosm outperformed the models described in the literature in terms of predicting 2 toxicological properties, including carcinogenicity and blood–brain barrier penetration (the prediction accuracy reached 93.4%). Neural network models were superior in predicting 4 toxicological properties, including acute toxicity, carcinogenicity, genotoxicity, and blood–brain barrier penetration (the prediction accuracy reached 93.8%). In addition, neural network models were better in predicting 3 pharmacokinetic properties, including gastrointestinal absorption, volume of distribution, and hepatic clearance (the coefficient of determination reached 0.825).Conclusions. The data obtained suggest that artificial neural networks are the most promising and practically significant direction for the development of in silico systems for predicting the ADMET characteristics of new medicinal products.

Introduction. Artificial neural networks are widely utilized in medical fields, such as dentistry, molecular genetics, immunology, cardiology, and others. Forensic medicine is no exception, as artificial neural networks are also beginning to find applications in this field. The aim of this study was to demonstrate the potential for predicting human anthropometric parameters using dermatoglyphic parameters, which could enhance the method of dermatoglyphic identification. Materials and methods. We analyzed dermatoglyphs of the hands and feet from 567 individuals aged 18 to 59 years, with no genetic or endocrine disorders and no musculoskeletal problems. Results and discussion. The outcome of our work resulted in the development of the "Dermatoglyphics For Prediction (DFP)" program [Author's Certificate No. 74561. Computer program "Forensic Medical Identification Program using Artificial Neural Networks" Registration date: 07.11.2017]. This software device, after appropriate training, enables the prediction of an individual's ethnic-territorial affiliation and the presence of specific anthropometric parameters using such input data as dermatoglyphs of the hands and feet. Conclusions. The increasing needs of the Ukrainian community for the identification of unknown individuals, given the geopolitical situation related to Russian invasion in Ukraine (the constant threat of ballistic missile attacks and unmanned aerial vehicles across the entire territory of Ukraine, which could lead to mass casualties), justify the relevance and the search for innovative approaches to dermatoglyphic identification expertise, utilizing state-of-the-art technologies, particularly neural network-based prediction of anthropometric parameters, sex, and ethnic-territorial affiliation of an unknown person, using input parameters such as dermatoglyphs of the hands and feet, with the aim of enhancing the evidentiary value of identification expertise. This software device, after appropriate training, allows for the prediction of ethnic-territorial affiliation and the presence of specific anthropometric parameters in the examined individual using such input data as dermatoglyphs of the hands and feet.

The article analyzes the application of various machine learning methods for solving the problem of approximating the forces of fluid film bearing lubricating layer in static formulation. The initial data on the values of lubricating layer forces for different shaft positions were obtained using a model of a rotor-bearing system based on the numerical solution of the Reynolds equation, with account for the cavitation effect. Methods for reducing the amount of calculation required to obtain the necessary data set are determined on the basis of analyzing solution approximation accuracy with artificial neural networks. After that, approximation models were constructed using a number of other machine learning methods, and the accuracy of predictions as well as the duration of the training process were analyzed. Finally, conclusions were drawn about the most effective approaches to building such models.

Objective. Monitoring and determining the state of crops in agricultural production requires the use and improvement of neural network methods of artificial intelligence.The aim of the study is to create a mathematical model of the learning process of the DeepLabV3 neural network for intelligent analysis and segmentation of agricultural fields.Method. Based on the newly formed RGB database of images of agricultural fields, marked up into four classes, a neural network of the DeepLabV3 architecture was developed and trained. Approximations of the learning curve by the modified Johnson function are obtained by the methods of least squares and least modules.Result. A statistical assessment of the quality of training and approximation of neural networks to the DeepLabV3 architecture in combination with ResNet 50 was carried out. The constructed DNN family based on DeepLabV3 with ResNet50 showed the efficiency of recognition and sufficient speed in determining the state of crops.Conclusions. Approximation of the neural network learning diagram to the DeepLabV3 architecture, using a modified Johnson function, allows us to estimate the value of the “saturation” of the simulated dependence and predict the maximum value of the neural network metric without taking into account its possible retraining.

Objectives. Specifications of models and tools for the development of artificial neural networks (ANNs) and their integration into knowledge bases (KBs) of intelligent systems are being developed. The relevance is determined by the necessity of implementing the possibility to solve complex problems by intelligent systems, which algorithms and methods of solving are not available in the knowledge base of the intelligent system.Methods. Four levels of integration of artificial neural networks into knowledge bases are formulated and analyzed. During the analysis the requirements and specifications for required models and tools for the development and integration are formulated. Specified at each level the models and tools include the models and tools of previous level. The application of the tools is considered by the example of solving the problem of classifying the knowledge base entities using a graph neural network.Results. The specifications of the ANN representation model in the knowledge base, the agent-based model for the development and interpretation of the ANN, which ensures the integration of the ANN into knowledge bases at all selected levels, as well as the method for classifying knowledge base entities using a graph neural network, have been developed.Conclusion. The developed models and tools allow integrating any trained ANNs into the knowledge base of the intelligent system and using them to solve complex problems within the framework of OSTIS technology. It also becomes possible to design and train ANNs both on the basis of external data and on the basis of fragments of the knowledge base. Automation of ANNs development process in the knowledge base becomes available.

The purpose of the study is to develop the architecture and working algorithm of intelligent relay protection of an electrical network with an isolated neutral. “Intellectualization” of relay protection is necessary to increase its sensitivity and reliability. This problem is relevant due to the accident rate of overhead power lines and incorrect operation of traditional relays. This is due to the fact that the set point is not adjusted according to the deviation of the line parameters with changing environmental and load conditions. The purpose of the study is achieved by solving the following problems: the concept of supplementing an electronic relay with a mechanism for correcting the operation set point is proposed. The use of the digital shadow of the electrical network is substantiated. The digital shadow interrogates the sensor system. The advantage of using a digital shadow as part of relay protection has been proven. This allows emulating faults and estimating currents to correct the setpoint. The most significant result of the study is an example of the implementation of intelligent relay protection using a digital shadow in the form of an artificial neural network. The application of a neural network for approximating the current function of a two-phase short circuit is demonstrated. The significance of the results lies in the improvement of the sensitivity and selectivity of current protection due to the digital shadow for working out emergency modes. The convenience of operating and reproducing a digital shadow in the form of a neural network is noted.

Purpose: to summarize the experience of the development and application of artificial neural networks (ANW) in early diagnosis of primary open-angle glaucoma (POAG).Material and methods. A total of 690 patients (918 eyes) were tested. The training clinical group consisted of 459 clinical examples (459 eyes), of which 369 eyes had an initial stage of POAG and 90 eyes had no glaucoma. The testing clinical group was represented by 131 examples (131 eyes), of which 110 eyes belonged to patients with POAG and 21 eyes were without glaucoma. The final diagnostic testing using ANW was conducted on 328 eyes with the diagnosis unknown to the researchers, which belonged to people with suspected POAG. The diagnostic complex included an optimally necessary set of research techniques.Results. ANW identified glaucoma in 198 eyes out of those with suspected glaucoma (60.4 %) with 100 % certainty. 76 eyes (23.2 %) were classified as non-glaucoma, or “healthy”; 54 eyes of the suspected glaucoma patients were identified as “doubtful”, whereupon they were retested by a neural network pool consisting of 5 neural networks. According to the results of the retesting, 28 eyes, or 51.9 % of the “doubtful” ones were identified as having glaucoma, whereas 26 eyes (48.1 %) were identified as non-glaucomatous, i. e. healthy.Conclusion. Our experience suggests that artificial neural networks pose no danger to the doctor or the patient and can be viewed as a very convenient tool for early POAG diagnostics.

В обучении искусственной нейронной сети одной из центральных проблем является начальная инициализация и настройка весовых коэффициентов, связанная с псевдослучайной инициализацией весовых коэффициентов. В статье описан базовый генетический алгоритм, а также метод определения весовых коэффициентов с использованием данного алгоритма. Также представлен комбинированный метод определения весовых коэффициентов, предусматривающий на первом этапе начальную инициализацию с помощью генетического алгоритма и использование стохастического градиентного спуска на втором этапе обучения Предлагаемые методы проверены на ряде искусственных нейронных сетей прямого распространения для различных задач по бинарной классификации реальных и синтетических данных, а также для однозначной многоклассовой классификации рукописных цифр на изображениях из базы данных MNIST. Искусственные нейронные сети сконструированы на основании теоремы Колмогорова-Арнольда. В данной статье проведен сравнительный анализ двух методов определения весовых коэффициентов – с использованием генетического алгоритма и градиентного спуска. По результатам сравнительного анализа сделан вывод о возможности применения генетического алгоритма для определения весовых коэффициентов как в качестве алгоритма начальной инициализации искусственной нейронной сети, так и в качестве алгоритма настройки весовых коэффициентов

In this paper, the relevance of developing methods and algorithms for neural network incremental learning is shown. Families of incremental learning techniques are presented. A possibility of using the extreme learning machine for incremental learning is assessed. Experiments show that the extreme learning machine is suitable for incremental learning, but as the number of training examples increases, the neural network becomes unsuitable for further learning. To solve this problem, we propose a neural network incremental learning algorithm that alternately uses the extreme learning machine to correct the only output layer network weights (operation mode) and the backpropagation method (deep learning) to correct all network weights (sleep mode). During the operation mode, the neural network is assumed to produce results or learn from new tasks, optimizing its weights in the sleep mode. The proposed algorithm features the ability for real-time adaption to changing external conditions in the operation mode. The effectiveness of the proposed algorithm is shown by an example of solving the approximation problem. Approximation results after each step of the algorithm are presented. A comparison of the mean square error values when using the extreme learning machine for incremental learning and the developed algorithm of neural network alternate incremental learning is made.

Background. Artificial neural network models can be used to analyze and predict structural components within the value dimension of the main processes in an outpatient clinic as indicators of patient satisfaction.Objective — to form and test the methodology for analyzing and predicting structural components within the value dimension of the main processes in an outpatient clinic, as indicators of patient satisfaction with availability and quality of medical care, using artificial intelligence.Methods. The results of questionnaires administered to 525 patients were used to analyze their satisfaction with GP appointments. A network ensemble consisting of radial basis network and multilayer perceptron was chosen as the basis for a neural network model. The model testing involved five outpatient clinics in Kirov. The total number of respondents comprised 217 patients. Statistical processing included data description and analysis. Qualitative attributes were represented by relative values (P, %). The statistical significance of differences in qualitative data was assessed using the Chi-square test. The correlation between the observed and predicted data was assessed by means of nonparametric Spearman correlation analysis. The value of p

Background. This article presents the results of the analysis of the mechanism of artificial neural networks on Russian railways for the control of current collectors of trains. A critical assessment of the situation was carried out and a forecast of the further development of neural network technologies in railway transport was given. Materials and/or research methods. Methods of mathematical modeling, system analysis, data analysis, comparison, systems theory, as well as architecture and mathematical model of neural network and information standards were used. Results. A model of the information system for monitoring the characteristics of current collectors has been developed, the algorithm of its operation and the functionality of the software package of the model of the system for monitoring the characteristics of current collectors of rolling stock have been described. The algorithm of neural network operation for determining the characteristics of electric rolling stock current collectors is presented. The analysis of the prospects for the use of artificial neural networks in the field of railway transport has been carried out, and a simulation model of an information system for monitoring the characteristics of current collectors of electric rolling stock of railway transport has been created. Conclusion. The article examines the positive experience of JSC “Russian Railways” on the introduction of neural network technologies in technological processes and further prospects for the use of neural network technologies on Russian railways. The necessity of intensifying the development and implementation of neurotechnologies to solve a wide range of tasks, including tasks to optimize the organization of maintenance and repair of rolling stock, is noted. The effectiveness and relevance of the use of artificial neural networks in various fields of activity, especially on Russian railways, is shown, and a model of an information system in the field of monitoring the characteristics of current collectors is proposed.

The paper considers the problem of monitoring the technical state of steel ropes used in industrial and civil installations, such as elevators, cable cars, lifts, etc. To date, the assessment of the state of steel ropes and the safety of their operation is carried out mainly through visual-instrumental control, which is long in time and subject to the influence of the human factor, which determines the relevance of the task of automating this process using computer technology and artificial intelligence methods in order to improve the quality of the assessment, reducing the check interval, detecting defects in steel ropes at an early stage. The article proposes two approaches to the algorithmizing of the problem of videoscopic control of the state of steel ropes using computer vision technologies, based on software modeling of the logic of visual analysis and machine learning methods, such as deep learning artificial neural networks. The authors have developed algorithms for automatic detection of the main types of defects in images of the surface of steel ropes: increase/decrease in diameter, deformation in the form of rope, breaks in the outer wires, thermal and electric damage. The paper presents the results of preliminary testing of the developed algorithms and programs, the composition and methods for determining the main indicators for evaluating their effectiveness are proposed.

Introduction. The safety problem and the situation with accidents during the operation of elevator installations are elucidated. The role of elevator rope defects as a factor of dangerous incidents is indicated from the point of view of statistics. The malfunctions of the elevator mechanical equipment related to the defective indices of the ropes are listed. There is a difference in the documentary fixation of defective indices and rejection rates of ropes of lifting structures. Materials and Methods. The well-known approaches to the control of ropes of lifting structures were described. It was emphasized that visual inspection control (VIC) was required to identify such rejection rates of steel elevator ropes as geometry change, corrosion and wear, wire breaks, temperature exposure, etc. The rejection rate was presented in the form of a mathematical system. The technical condition of elevator ropes during the operation was integrally assessed by the totality of identified defects at a fixed length. The decision to create a software and hardware complex (PAC) for the practical implementation of visual and measuring control was validated. Results. The developed PAC VIC laboratory sample consisted of a hardware part, a video stream processing module, communicator for the server connectivity, specially designed software, and a client mobile application. PAC VIC implemented the following functions: – automatic detection and classification of the major significant rope defects based on a deep convolutional artificial neural network; – demonstration of a three-dimensional image of a rope and an image scanning algorithm with distortion compensation, according to which the metric characteristics of defects were fixed; – integral assessment of the technical condition of the rope according to the totality of detected defects; – color interpretation of the actual technical condition of the rope with subsequent transmission to the user's mobile device. Preliminary tests have shown the suitability of the PAC VIC for identifying defects. The reliability of the results for the identification and qualification of defects exceeded 80%. Work on deep learning of the system continues. Discussion and Conclusions. PAC VIC of elevator ropes provides eliminating the risks of visual control caused by the psychophysical state of a person. It works remotely and contactless. The solution proposed by the authors automatically evaluates the rejection rates according to five criteria: external wire breaks, surface wear, rope diameter change, undulation, traces of temperature exposure. An important result of the VIC of steel ropes using computer vision and artificial intelligence is an increase in reliability and safety during the operation of elevator equipment.

For several decades in the field of data mining in education (EDM), predictive learning has remained one of the most popular and internationally discussed research topics. Specifically, data mining is used to predict educational outcomes such as academic performance, retention, success, satisfaction, achievement and dropout rates. In the management practice of higher education institutions, on the basis of an operational forecast, measures are developed and implemented to support those students who fall into the risk group.Our study is aimed at substantiating a model for predicting the early departure of students using an artificial neural network and analyzing predictors that increase the accuracy of predicting successful graduation from a Russian university. This work will expand the international practice of comparative research in higher education.The paper confirms the already existing hypotheses about the influence of a number of factors on the prediction of academic performance and suggests the need to test their universality or specificity in a particular institution of higher education. We also proved that an artificial neural network model with a certain set of attributes can be applied in the context of a single higher education institution, regardless of specialization. To determine the potential risk group of students, a binary classification prediction model is used. The overall prediction accuracy of a neural network with combined data reaches 88%. For this neural network model, the basic predictors that affect the accuracy of the forecast are the cumulative average level of achievement (CGPA) and the year of admission to the university.

Introduction. The ability of runoff modeling is an essential step in the hydrologic modeling process and therefore water ba­lance studies, therefore, this study aims to verify the ability and reliability of the MIKE 11NAM program in modeling runoff, in the upper basin of Orontes River in Syria, with the use of artificial intelligence models to fill the gaps in runoff time series.Materials and methods. In this study, models of artificial neural networks and fuzzy inference models were used and they were compared with each other to determine the best model in order to fill the gaps in the surface runoff data at Al-Jawadiyah and Al-Amiri stations. Then the results were used in the modeling process using the MIKE 11 NAM program.Results. The results showed a high reliability of artificial intelligence models, whether neural networks or fuzzy inference models, with a relative preference for neural networks, and after using these results within the data required for modeling in the Mike program, it was found that there are large differences between the observed and simulated values due to the lack of existing data on the study area.Conclusions. This study recommends to continue research on the issue of hydrological modeling in case of lack of data and to compare between different models to find the best way to solve this problem.

Image segmentation plays an important role in detecting various diseases and pathologies through medical image processing. Over the years, a number of traditional approaches such as the binary threshold method (Otsu method), watershed method, and K-means clustering method have been developed to perform image segmentation, using domain-specific knowledge to efficiently solve segmentation problems in specific applications. But unfortunately, these methods were not effective. In fact, the essence of image segmentation is the classification of containing each pixel with similar attributes. And the goal of classifying a pixel is to treat it coarse, insensitive to noise, fine, and a medium star filter. Something that traditional methods cannot do. Therefore, there is a need to propose a new method that allows eliminating these disadvantages. We propose a method based on autoencoder (a special architecture of artificial neural networks that allows the application of learning without a teacher using the method of back propagation of the error), which allows compressing any large image into a small one, (with the same properties as the input). This method also saves resources. In this case, we use an image processing method called "MaxPooling" (U-Net). In the work, we created our own metric to track the result of the network training. In the practical part, we tried segmenting into classes the images from the construction site, as well as pixel by pixel the locations of the aircraft in the images.

redicting currency rates is important, for everyone who is trading and trying to build an investment portfolio from a range of crypto currencies. It is not subject to the same restrictions as fiat currencies. In this study, we seek to predict the exchange rate of BIT-COIN against the US dollar. The short-term data (365 observations) is processed using the LSTM model as one of the neural network models. Modeling is conducted by training a sample size of 67%, taking into account sharp fluctuations in the price of trade and a certain level of market efficiency. The GARCH model is used to select appropriate historical periods for how the LSTM model works and to test proficiency at the weak, semi-strong, and strong levels. The data series obtained from the website (Investing.com) have been processed. The researchers have found that the performance of the neural network improves as the EPOCH value increases with a training (research) period of 50 days before, which is consistent with the results of the proficiency test at the weak level. It agrees with the results of the sufficiency test at the weak level, which indicates that in the case under study (the Bitcoin market is effective at the weak level). It is advised that crypto-currency investors rely more on the historical trend of the price of the currency than on its current price, taking advantage of the artificial neural network model (LSTM) in dealing with little data of high volatility.

The article dwells on cause-effect relations between certain socio-hygienic factors and age-specific mortality rates due to cardiovascular diseases. New research trends in hygiene, a multidisciplinary approach to studies in the field and the current state policy make the present work topical. Our methodical approach to predicting probable age-specific mortality rates due to cardiovascular diseases relied on applying artificial neural networks. We analyzed a set of indicators that described the public healthcare system, sanitary-epidemiological welfare on a given territory, lifestyle, economic conditions, sociodemographic conditions, and primary incidence. Overall, we obtained 18 models (as per 5-year age-specific periods) of a relationship between socio-hygienic determinants and mortality rates due to cardiovascular diseases. The determination coefficients fell within 0.01–0.75 range and the greatest explanatory power occurred when the age period “30 years and older” was analyzed. We detected comparability of variational series obtained for mortality due to cardiovascular diseases among the whole population and the determination coefficients of the created models. We established predictive estimates of life expectancy at birth (LEB) in case there were changes in the analyzed socio-hygienic determinants by 2024 set within a certain scenario. Thus, changes in the whole set of determinants would result in 514 days added to LEB; lifestyle-related indicators, 205 days; indicators describing sanitary-epidemiological welfare, 126 days; economic indicators, 102 days; sociodemographic indicators, 101 days; primary incidence rates, 40 days; indicators describing the public healthcare system, 19 days. Several determinants were shown to be the most significant for reducing mortality due to cardiovascular diseases among working age population and older age groups. They are indicators describing people’s physical and motor activity, income levels, consumption of vegetables, education, and working conditions. Our research results are consistent with those obtained by other studies with their focus on establishing cause-effect relations between environmental factors and public health.

The relevance of the study is caused by the fact that differential tacks are one of the most difficult accidents in the entire technological chain of construction of oil and gas wells. A high quality and correctly selected drilling mud with optimal rheology for specific drilling conditions is one of the determining factors in preventing differential tacks. Goal: development of a neural network rheological model of drilling mud based on its component composition and the results of periodic measurements of the output parameters of the flushing fluid. With the help of a neural network, it is possible to accurately and quickly predict the values of the rheological properties of the solution, which have a significant impact on the occurrence and prevention of differential seizures. Object: neural network rheological models of drilling mud, drilling muds, which composition affects the rheological properties and the possibility of preventing differential sticking of the drill string during well construction. Methods: neural network model of drilling mud differing in the number and composition of input parameters is proposed. Results. The paper describes the learning process of three neural networks based on operational data obtained on instruments for measuring drilling mud parameters. The authors proposed six types of drilling muds that are optimal for specific geological conditions. Introduction of nanodispersed copper and potassium aluminate into the composition of a drilling mud with a high lubricity helps to reduce the friction coefficient, increase the inhibitory ability of the solution, reduce water loss and, as a result, decrease sharply the differential tack during the construction of wells on hydrocarbon raw materials.

The article presents the result produced by predicting growth potential in life expectancy at birth (LEB) of the RF Population. The predictions are based on scenario changes in social and hygienic determinants (SHD) identified by using an artificial neural network (ANN). This research is vital given the existing social strategies aimed at improving the medical and demographic situation in the Russian Federation. These strategies stipulate achieving targets set within the major national and federal projects. We identified an optimal ANN structure based on a four-layer perceptron with two inner layers containing eight and three neurons accordingly. This structure is able to produce results at the highest determination coefficient (R2= 0.78). Differences between actual LEB levels and predicted ones obtained by using the suggested model did not exceed 1.1 % (or 0.8 years). We established that average LEB in the RF would reach 75.06 years (by 2024) provided that the demographic situation in the country recovers in the nearest future, LEB level reaches its values detected in 2018–2019, and SHD values grow to their preset levels according to the target scenario. Therefore, the detected growth potential amounts to 3.0 years (1095 days) against 2018. “Lifestyle-related determinants” produce the greatest effects on the growth potential in LEB by 2024 (461 days). We also identified effects produced by such SHD groups as “Sanitary-epidemiological welfare on a given territory” (212 days), “Social and demographic indicators” (196 days), “Economic indicators” (131 days), “Indicators related to public healthcare” (70 days). An indicator that shows “A share of population doing physical exercises or sports” is the most significant determinant producing the greatest effects on potential changes in LEB. If it grows up to 55.0 %, a potential growth in LEB amounts to 243.5 days. If we do not consider COVID-related processes and rely only on the trends that are being observed now when predicting changes in the demographic situation by 2030, we can expect a possible additional growth in LEB that equals 286 days. The developed algorithm for determining growth potential in population LEB can be used as an instrument for determining and ranking priority health risk factors

The article gives information about the attempt made to select configurations, train and test artificial neural networks for predicting yields of grain crops considering of climate changes. Peculiarities of agricultural production require constant improvement of methods for analyzing crop yields, time series, and longterm climatic characteristics. Preliminary statistical evaluation of the considered time series made it possible to identify certain patterns. Time series were divided into four intervals: for building a network, its training, testing and control. During the construction of artificial neural networks, three models were used: MLP - multilayer perceptron, RBF - r adial basis functions and GRNN - g eneralized regression neural network. Based on the results of the construction, the best model was chosen. The sum of active air temperatures and the sum of precipitation for the growing season was used for artificial neural networks at the input, and the crop yield was used at the output. The use of sets of neural systems, generated automatically, contributed to the effective forecasting of crop yields based on the analysis of climate data. As a result, according to the selected model, a yield forecast was made for the coming years considering climatic characteristics.

The paper is devoted to the neural network approach, which is proposed to be used to predict the operational parameters of heterostructural nanoscale devices. The advantage of this approach is a clear methodology for evaluating the weighting coefficients as part of a trained artificial neural network, which makes it possible to solve the problem for devices with an arbitrary structure. Learning is a complex iterative process, at the end of which it is important to evaluate the functioning of the neural network model. Therefore, it is necessary to determine the achieved accuracy and to identify negative effects that may occur during the learning process, when such a model is being developed. The project presents a criterion for evaluation the training quality of the neural network model of heterostructural nanoelectronic devices for predicting their electrical parameters. The main advantage of this criterion is its sensitivity to negative effects arising in the learning process, which was demonstrated by an example with two input training parameters and confirmed by visual control of 3D surfaces. The applicability of the developed criterion in the selection of neural networks with arbitrary architecture for solving design problems in the development of semi-conductor devices has been proved.

В статье рассматривается вопросы аппроксимации и интерполяции функций с использованием искусственных нейронных сетей, приводится теоремы Колмогорова-Арнольда и Цыбенко, которые показывают возможность применения нейронных сетей для аппроксимации и интерполяции функций. Проводится серия экспериментов, в ходе которые искусственные нейронные сети аппроксимируют и интерполируют следующие функции f(x) = |x|, f(x) = sin(x), f(x) =1/(1+25x²) . Показаны проблемы в обучении нейронной сети основанная на инициализации весовых коэффициентов случайным образом. Показана возможность обучения нейронной сети для работы с многообразием.

The use of data mining and machine learning tools is becoming increasingly common. Their usefulness is mainly noticeable in the case of large datasets, when information to be found or new relationships are extracted from information noise. The development of these tools means that datasets with much fewer records are being explored, usually associated with specific phenomena. This specificity most often causes the impossibility of increasing the number of cases, and that can facilitate the search for dependences in the phenomena under study. The paper discusses the features of applying the selected tools to a small set of data. Attempts have been made to present methods of data preparation, methods for calculating the performance of tools, taking into account the specifics of databases with a small number of records. The techniques selected by the author are proposed, which helped to break the deadlock in calculations, i.e., to get results much worse than expected. The need to apply methods to improve the accuracy of forecasts and the accuracy of classification was caused by a small amount of analysed data. This paper is not a review of popular methods of machine learning and data mining; nevertheless, the collected and presented material will help the reader to shorten the path to obtaining satisfactory results when using the described computational methods

Problem and goal. The development of mass open online courses contributes to the increasing attention of students to them. At the moment, there are many large services that provide online training, but there are no clearly defined universal requirements for such courses. Also, along with this problem, there is a fairly high level of rejection of the course at various stages due to the loss of motivation to continue training. Methodology. A variant of solving these problems by using adaptive learning technologies on the example of a course on learning artificial neural network technologies was considered. Results. In the process of reviewing the issue, the topics of the online course sections were determined. As a result, a work plan was drafted and the most relevant ways to solve the identified problems were formulated. Conclusion. The developed strategy can help with further elaboration and testing of the designed course and can be applied to any mass open online course.

The authors synthesized samples of piezoelectric potassium sodium niobate ceramics of 10,25 and 40 pore percentage by volume. Capsule convolutional artificial neural network has been developed for estimation of the pore percentage in images. Using the scanning electron microscopy, f learning massive of examples was formed (photographs of surface and edges of so-synthesized samples). Development and approbation of the capsule convolutional artificial neural network was completed in a few stages. During the first stage, the network was trained using a backpropagation method. Secondly, it was tested by a testing set. At the final stage we made a comparison of the acquired results with the results of the density comparing method. The article shows that this method can be used the pore percentage estimation in sodium niobate ceramics because the acquired results are comparable with the results of other methods. It was found that the samples where the pores were not made also have some pore percentage (about 5%).

The article considers the study of methods for numerical solution of systems of differential equations using neural networks. To achieve this goal, the following interdependent tasks were solved: an overview of industries that need to solve systems of differential equations, as well as implemented a method of solving systems of differential equations using neural networks. It is shown that different types of systems of differential equations can be solved by a single method, which requires only the problem of loss function for optimization, which is directly created from differential equations and does not require solving equations for the highest derivative. The solution of differential equations’ system using a multilayer neural networks is the functions given in analytical form, which can be differentiated or integrated analytically. In the course of this work, an improved form of construction of a test solution of systems of differential equations was found, which satisfies the initial conditions for construction, but has less impact on the solution error at a distance from the initial conditions compared to the form of such solution. The way has also been found to modify the calculation of the loss function for cases when the solution process stops at the local minimum, which will be caused by the high dependence of the subsequent values of the functions on the accuracy of finding the previous values. Among the results, it can be noted that the solution of differential equations’ system using artificial neural networks may be more accurate than classical numerical methods for solving differential equations, but usually takes much longer to achieve similar results on small problems. The main advantage of using neural networks to solve differential equations` system is that the solution is in analytical form and can be found not only for individual values of parameters of equations, but also for all values of parameters in a limited range of values.

The object of research is the process of preprocessing image data, which leads to the improvement of the neural network. The subject of the research is the preprocessing of the color spectrum of the image and changes in its model. The purpose of the work is a method of improving the performance of the neural network in the task of classifying color images by changing the color model at the preprocessing stage. The proposed color model allows you to exclude color ranges with zero values that do not affect the learning and recognition process. Also, this model allows you to use mutually exclusive data processing methods. Ref. 5, pic. 5.

Sintered friction materials are widely used in friction units of automotive vehicles and special purpose vehicles. The main purpose is to transmit torque to the actuator. The development of the technology market requires the development and use of new units. At the same time, the creation of new materials is required, which also applies to sintered friction materials. This group of materials is characterized by a high service life, efficiency of torque transmission, as well as the ability to restore performance in case of violation of operating modes. One of the most significant parameters characterizing a sintered friction material is wear resistance. In most cases, it determines not only the resource of the unit itself, but the entire machine as a whole. A special place is occupied by brake units, which also use friction materials. The increased wear resistance of the friction material contributes to a decrease in the efficiency and service life of the brake system. Evaluation of the wear resistance of a friction material for the given operational parameters is a very long and costly process. The development of methodology and methods for accelerating the assessment of wear resistance is an important scientific and practical task. The paper presents the results of using artificial neural networks to predict the service life of a composite friction material based on copper on the sliding speed, pressure on the material and the amount of lubricant supplied to the friction zone. An artificial neural network has been trained using an array of experimental data for the FM-15 friction material. The training results have shown high accuracy, correctness of the proposed and implemented network architecture. The developed software has demonstrated its efficiency and the possibility of using it in calculations to determine the wear of a composite friction material.

The article presents an approach to the synthesis of combinational-logic circuits using artificial neural networks (ANNs). The presented method is focused on the use of a perspective basis using the majority function (a Boolean function of three arguments that takes the value "true" if at least two of its inputs are true). This choice is based on emerging nanotechnologies, where the majority element is most easily represented. The class of applied ANNs is deep networks with reinforcement. Such networks have been actively studied and applied in recent years. There are examples of their effective use for automatic logic circuits optimization. The original synthesis method proposed in the article with the simplification of circuits that implement the Shannon expansion in all variables of the corresponding function of the logic algebra (FLA). On large schemes, it becomes essential to use some simple but effective techniques for training deep ANN agents with reinforcement. This allows you to distribute calculations into several independent subtasks, each of which is explored and makes performing by agents quicker and easier. Two reinforcement learning algorithms for simplifying schemas are described. They provide a solution to the Exploration-Exploitation conflict, which is the contradiction between exploring the environment to find the optimal episode and using information about the episode considered optimal at the current time. The dependences of the parameters of the synthesized circuits on the number n = 3, ..., 10 of FAL variables and the number of network training episodes are presented.

Copyright of Cybernetics & Systems Analysis / Kibernetiki i Sistemnyj Analiz is the property of V.M. Glushkov Institute of Cybernetics of NAS of Ukraine and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Nowadays deep neural networks play a significant part in various fields of human activity. Especially they benefit spheres dealing with large amounts of data and lengthy operations on obtaining and processing information from the visual environment. This article deals with the development of a convolutional neural network based on the YOLO architecture, intended for real-time book recognition. The creation of an original data set and the training of the deep neural network are described. The structure of the neural network obtained is presented and the most frequently used metrics for estimating the quality of the network performance are considered. A brief review of the existing types of neural network architectures is also made. YOLO architecture possesses a number of advantages that allow it to successfully compete with other models and make it the most suitable variant for creating an object detection network since it enables some of the common disadvantages of such networks to be significantly mitigated (such as recognition of similarly looking, same-color book coves or slanted books). The results obtained in the course of training the deep neural network allow us to use it as a basis for the development of the software for book spine recognition.

The modern capabilities of computers have returned interest in artificial intelligence technologies. A particular area of application of these technologies is pattern recognition, which can be applied to the traditional forensic task – identification of signs of forgery (imitation) of a signature. The results of forgery are differentiated into three types: auto-forgery, simple and skilled forgeries. Only skilled forgeries are considered in this study. The online and offline approaches to the study of signatures and other handwriting material are described. The developed artificial intelligence system based on an artificial neural network refers to the offline type of signature recognition – that is, it is focused on working exclusively with the consequences of the signature – its graphic image. The content and principles of the formation of a hypothesis for the development of an artificial intelligence system are described with a combination of humanitarian (legal) knowledge and natural-technical knowledge. At the initial stage of the study, in order to develop an experimental-applied artificial intelligence system based on an artificial neural network focused on identifying forged signatures, 127 people were questioned in order to identify a person's ability to detect fake signatures. It was found that under experimental conditions the probability of a correct determination of the originality or forgery of the presented signature for the respondent is on average 69.29 %. Accordingly, this value can be used as a threshold for determining the effectiveness of the developed artificial intelligence system. In the process of preparing the dataset (an array for training and verification of its results) of the system in terms of fraudulent signatures, some forensically significant features were revealed, associated with the psychological and anatomical features of the person performing the forgery, both known to criminalistics and new ones. It is emphasized that the joint development of artificial intelligence systems by the methods of computer science and criminalistics can generate additional results that may be useful outside the scope of the research tasks.

Diagnostic efficiency of breast cancer screening remains one of the most important issues in oncology and radiology. Artificial intelligence technologies are widely used in clinical medicine to effectively solve a number of technological and diagnostic problems. The aim. To develop segmentation neural network model for breast plain radiographs analysis with subsequent study of its clinical effectiveness. Materials and methods. The artificial intelligence-based system was developed to analyze X-ray mammography, аnd included a segmentation neural network with the U-Net architecture, a classification neural architecture ResNet50 with outputting the result using gradient boosting. 15486 X-ray cases were used for training, estimation of diagnostic accuracy and validation of the developed segmental model. All cases were labeled in specially developed software environment LabelCMAITech. The segmentation accuracy was determined by Intersection over Union (IoU) similarity coefficient, the probability of malignancy was calculated using the binary classification metrics. Results. The developed system is represented by a segmentation model based on neural network architecture. The model allows, with high accuracy of 0.8176 and higher, at threshold values on the output neurons of the network of 0.1 and 0.15, to localize X-ray findings that have diagnostic value for determining the likelihood of the presence of breast cancer signs in an X-ray mammographic study — focus, architecture distortion, local asymmetry, calcifications. When comparing the results of machine segmentation and marking of images by a radiologist, it was found that the model is not inferior to the doctor in the accuracy of determining the formations, extra-focal calcifications and intraglandular lymph nodes. Conclusion. The results of this study allow considering the model as an intelligent assistant to a radiologist in the analysis of screening mammographic cases.

Introduction. The selected research topic urgency is due to the need for a quick assessment of the condition and reliability of materials used in various designs. The work objective was to study parameters of the influence of the defect on the response of the surface of the medium to the shock effect. The solution to the inverse problem of restoring the radius of a defect is based on the combination of a computational approach and the use of artificial neural networks (ANN). The authors have developed a technique for restoring the parameters of a defect based on the computational modeling and ANN. Materials and Methods. The problem is solved in the flat setting through the finite element method (FEM). In this paper, we used the linear equations of the elasticity theory with allowance for energy dissipation. The finite element method implemented in the ANSYS package was used as a method for solving the boundary value problem. MATLAB complex was used as a simulation of the application process (ANN). Results. A finite element model of a layered structure has been developed in a flat formulation of the problem in the ANSYS package. The problem of determining unsteady vibrations under pulsed loading for different radius variations of the defect is solved. Positional scanning of the research object is applied. Graphical dependences of the vibration amplitudes of points on the surface on the defect radius are plotted. Discussion and Conclusions. As a result of studying the dependences of vibration responses on the defect radius, the authors have developed an approach to restore this parameter in a flat structure based on a combination of the FEM and ANN. The research has shown that the amount of data used is sufficient for successful training of the constructed ANN model and identification of a hidden defect in the structure.

A manipulator is an alternative to progress profitability in mechanical computerization. The robotic controller executes forms’ assembly operations in hazardous conditions. Since computerized controllers are highly vulnerable nonlinear powerful frameworks, it is hard to provide precise unique conditions at controlling laws’ configuration. Virtual Reality (VR) is a fundamental advance at use in modern biomedical, medical procedures and different fields where a 3D object helps to comprehend complex behavior. This work proposes the interaction with 3D models in VR environment achieved by accurate sensing from feedback, and then reconstructs the instruction according to practical limitation of a real human arm movement. In this work ANFIS played a key role in finding results with optimal values because the combination of Neural Networks (NN) benefits and obscure logic systems research examined the individual defects in both of them. Use of Artificial Neural Networks (ANN) in dynamic systems has quite extensive and accurate results, when adding a training signal system to the mixed learning base implemented at combining the slope proportions technique, a Least Square Error (LSE) preparing the ANFIS organization for any framework to take care of the issue any way. This work presents a keen controller actualization with 7-DOF controller for a model designed with a VR situation that reproduces the system design by associating Matlab/Simulink to connect the VR model with some instruction to execute orders delivered by the hybrid intelligent controller based on ANFIS technique. Palatable outcomes are implemented in reproductions that improve the procedure as an essential utilization of this controller design.

The paper is devoted to the development of an effective semantic segmentation algorithm for automation of airport infrastructure labelling in RGB satellite images. This task is addressed using algorithms based on deep convolutional artificial neural networks, as they have proven themselves in a wide range of tasks, including the terrestrial imagery segmentation, where they show consistently high results. A new dataset was labelled for this particular task and a comparative analysis of different architectures and backbones was carried out. A conditional random field model (CRF) was used for postprocessing and accounting of contextual information and neighborhood of objects of different classes in order to eliminate outliers. Features of the solutions applied at all preparatory stages of the algorithm were described, including data preparation, neural network training and post-processing of the training results.

Relevance of the discussed issue is caused by the need to improve the accuracy of short-term load forecasting of delivery point cluster of the second level default provider. The system operator uses the result of forecast when forming power system dispatch load curve. Usually, prediction errors lead to increase of primary energy resources consumption for electric-power production, due to unjustified run and shutdown of generating equipment, as well as increasing of circuit losses, due to the choice of non-optimal scheme of electric grid. As the electricity consumption depends on many factors, the task of short-term load forecasting is poorly formalized. Under these conditions, traditional methods of mathematical statistics and simulation do not allow building the adequate forecast models. Until recently, the only fine method of load forecasting was the Delhi approach. Currently, tools of neural networks and deep machine learning are widely used for short-term load forecasting of the energy system of a region of the country or delivery point cluster of first level default provider. However, the developed models are not suitable for predicting hourly electricity consumption of delivery point cluster of the second level default provider. Short-term load forecasting of this object is complicated of reliability of electric grid 6–110 kV, the operating mode of electricity consumers with a capacity of 670–10000 kW, the presence of district heating and water supply, beside standard time and meteorological factors. For this forecasting object, the question of choosing the optimal architecture and configuration of the neural network model, as well as the learning algorithm, which can achieve the desired forecasting accuracy, remain open. The main aim of the research is to improve the accuracy of short-term load forecasting of delivery point cluster of the second level default provider with the help of tools of neural networks and deep machine learning. The methods: the methods of correlation and factor analysis, the theory of artificial neural networks and machine learning. Software implementation of theoretical calculations was performed with help of deep machine learning library Tensor flow Keras in the Python 3.6 programming language. Results. The authors have developed the neural network algorithm for short-term load forecasting of delivery point cluster of the second level default provider with adaptive learning and momentum rate and completed the software implementation of this algorithm in deep machine learning library Tensor flow Keras. The use of this artificial neural network let to decrease in monthly average relative forecast error by 5,14 %.

The article considers the importance of risk management in the corporate system of industrial enterprise, substantiates the choice of data mining techniques and suggests the conceptual model of risk assessing internal and external enterprise environment using fuzzy-logic hierarchical network tools.

This article is devoted to developing a mathematical model of nonlinear bearing supports for turbopump units of liquid rocket engines considering initial gaps and axial preloading. In addition to the radial stiffness of the bearing support, this model also considers the stiffness of the bearing cage, the rotational speed of the rotor, axial preloading of the rotor (due to which the inner cage shifts relative to the outer, changing the radial stiffness of the support), as well as radial gaps between contact elements of the bearings. This model makes it possible to calculate the stiffness of the bearing supports more accurately. The proposed model is realized using both the linear regression procedure and artificial neural networks. The model’s reliability is substantiated by the relatively small discrepancy of the obtained evaluation results with the experimental data. As a result, this model will allow determining the critical frequencies of the rotor with greater accuracy. The results have been implemented within the experience of designing turbopump units for State Company “Yuzhnoye Design Office”.