Search

User activities generate a significant number of poor-quality or irrelevant images and data vectors that cannot be processed in the main data processing pipeline or included in the training dataset. Such samples can be found with manual analysis by an expert or with anomalous detection algorithms. There are several formal definitions for the anomaly samples. For neural networks, the anomalous is usually defined as out-of-distribution samples. This work proposes methods for supervised and semi-supervised detection of out-of-distribution samples in image datasets. Our approach extends a typical neural network that solves the image classification problem. Thus, one neural network after extension can solve image classification and anomalous detection problems simultaneously. Proposed methods are based on the center loss and its effect on a deep feature distribution in a last hidden layer of the neural network. This paper provides an analysis of the proposed methods for the LeNet and EfficientNet-B0 on the MNIST and ImageNet-30 datasets.

Developing metrology solutions is of major importance when building data-measurement systems. One of the key issues is evaluating the characteristics of measuring channels built on the basis of digital sensors of physical values. These sensors include a primary analog transducer and an analog-digital converter. The study evaluates the occurrence of errors in measuring channels of a multi-channel data-measurement system. This study deals with the measuring channel built on the basis of digital temperature sensor employing contact measurement. The digital temperature sensor is an integral body that combines both the primary temperature transducer and the primary data processing units. Drawing on the analysis of various errors the study develops the abstract metrology model of the measuring channel. This model takes into account the errors related to inherent instrument errors as well as the errors induced by the structure of the measuring subsystem. The work presents calculations of multiplicative and additive errors of the measuring channel. It also suggests measures for minimizing the additive errors of the measuring channel in the system.

This paper considers the process of light propagation in a multimode optical waveguide being used in fiber-optic pressure sensors (FOPS). The Maxwell’s equations for the simplified geometry of bending are considered (for example, to represent it as sections of torus segments and "stitch" the solutions in different segments). Statistical modeling methods are used, for example, the Monte Carlo method to calculate a set of trajectories based on the solution of purely geometrical problems for finding intersection points and using Fresnel conditions in case of a refracting reflection point, or reflection from a surface with two radii of curvature in case of a tunneling reflection point. The research results will allow us to design and simulate a sensitive element of the sensor.

It is difficult to imagine modern airspace without unmanned aerial vehicles (UAV), or drones. The use drones both for state purposes and for the private ones is rapidly is rapidly expanded. A widespread use of UAV is due to a significant range of tasks to be solved. The drones are not only gradually displacing piloted aircrafts (used for searching seats of fire, checking the integrity of pipelines, detecting accidents and congestion on roads and rivers, etc.), but also find their application in the areas that were not included into the sphere of aviation earlier (for example, for delivery of smaller loads, or as an air taxi in future). The active development of the UAV market is inextricably linked with increased accidents caused by the rapid growth of the number of operated drones.

This gauge was designed on the basis of a double-sided printed circuit board, which was used for forming its sensitive elements, capacitive reference gauge blocks, as well as all communication lines by the photolithography method with a high accuracy. Each meter of this level gauge is divided into eight continuous measurement sub-bands, ending with a reference gauge block, due to which an auto-calibration of each measurement sub-band with an absolute error of Δ = ±0.3 mm is carried out. An accuracy control with the help of reference gauge blocks is carried out in each measurement cycle evenly throughout the whole range of level measurement of the controlled liquid. A technique has been developed for designing a discrete-continuous capacitive level gauge for a wide class of dielectric liquids with complex manufacturing of sensitive elements and gauge communication lines based on a one-meter printed circuit board. A simple design of the level gauge and manufacturing of sensitive elements in the form of printed circuit boards, and the presence of embedded auto-calibration provide the gauge with initially high metrological characteristics along with low cost of its manufacture.

The purpose of the work is to construct estimates close in their effectiveness to effective estimates of the binomial test plan, but devoid of their shortcomings in failure-free tests. Reliability function and mean time to failure are selected as indicators of reliability. Methods. To find an effective estimate, integral numerical characteristics of the estimation accuracy were used, namely: the total squared of the offset (deviation) of the expected implementation of a certain variant of the estimate from all possible parameters of the binomial test plan. Results and Conclusions. The estimate of failure probability ${\hat P_3}$ should be recognized more effective in comparison with the proposed ones, as having a minimum displacement and close to the minimum variation of its values. The constructed composite estimate of failure probability ${\hat P_3}$ allows escaping from the incident, when we assess the situation as a reliable event in failure-free tests. At the same time, when failures occur, the ${\hat P_3}$ estimate fully coincides with the absolutely effective estimate $\hat P = 1 - R/N$, which, in this case, allows using it for any results (R>0) of the binomial tests instead of the estimate $\hat P$. The MTTF estimate ${\hat T_4}$ should be recognized more effective in comparison with the proposed ones, as having a minimum displacement and a minimum variation. The constructed estimate ${\hat T_4}$ allows estimating the MTTF with a finite value in case of failure-free tests conducted according to the binomial plan. However, the estimate ${\hat T_4}$ is constructed on the assumption that time to failure has an exponential distribution. If the distribution of time to failure is not known, the non-parametric estimate ${\hat T_0}$ has an advantage in its efficiency

The paper scrutinizes the method of artificial neural networks on the basis of radial-basic functions using CUDA architecture which considerably allows accelerating calculations during parameter optimization of connectionist structures. Most resource-intensive calculating operations in the process of neural network synthesis are singled out. The algorithm of quick calculations of neural network output while analyzing the value variants of its parameters is proposed. The algorithm is based on the storage of interim calculations, and is characterized by the calculation speed independence from the dimension of data input vector. The software implementation of the examined quick calculations algorithm on the basis of graphic accelerators is described. The comparative analysis of calculation efficiency for various variants of synthesis algorithm implementation, including the use of CPU and GPU, is presented. The research shows that the efficiency of the examined algorithm grows at the increase of the training sample size. The paper presents the results achieved during the calculating experiment for various amount of neurons and sample elements. These results prove the efficiency of the described synthesis method. The proposed algorithm can be used in the systems that require the synthesis of connectionist models in the real time mode.

To solve problems of high-frequency signal propagation in directional structures at constant exposure to elevated temperature is an urgent task. The main objective of the proposed paper is to study the effect of dielectric, based on epoxy resins, on signal propagation in the gigahertz range at elevated temperatures. This is possible using mathematical modeling that adequately reflects an actual physical process, taking into account not only the geometric dimensions of the signal transmission line, but also such parameters, as: edge effects; effective dielectric permittivity and glass transition temperature of the dielectric propagation medium; the effect of elevated temperature on the dielectric. Issues of epoxy resin-based dielectric effect on the speed of signal propagation via flat transmission lines under the influence of high temperature are investigated by means of mathematical modeling. The research results allow assessing temperature influence on the speed and delay of signal propagation via transmission lines based on dielectrics, which affects the characteristics and reliability of the operated information and communication devices and systems.

The article describes an adaptive technique for vibration testing to determine dynamic characteristics of structural elements of new generation rocket-space and aviation technology. A technique for experimental determination of centers of natural mode antinodes of the object studied is proposed. An algorithm for the formation of the test signal has been developed. A structure of an adaptive intelligent vibration testing system has been developed.

Research rationale and aims. The technical condition of radio-electronic systems depends on a multitude of parameters that change in course of operation, which can lead to the loss of system operation capacity, additional maintenance and repair costs. Reduction of running costs of radio-electronic systems while maintaining the required quality is an important task that calls for improving the methods for predicting their technical condition. The aim of this work is to develop the model for predicting the parameters of the radio-electronic system during operation. Methods. The prediction model was built on the basis of Markov chain theory and Markov branching process in particular. The model's stochastic characteristics were determined using the mathematical methods of generating functions. Result. The obtained stochastic model for parameter prediction of radio-electronic systems reliability possesses, as different from other approaches, analytical solutions for the system of differential equations. Conclusions. The suggested model can be used for predicting the RES parameter reliability and technical condition depending on the time of operation.

The way to increase the vibration resistance of the aircraft on-board systems is considered. It is proposed to use circuit design methods slightly covered in the literature known to the authors, in relation to vibration effects for solving the problem of conservation of their functional reliability. On the basis of the radar receiver chosen as the object of the analysis, its cascades, making the greatest contribution to reducing the vibration resistance, namely, the antenna assembly, and to a lesser extent, the mixer and the heterodyne, are determined. The fluctuation parameters of antennas, mixers, and heterodynes are defined. It is proposed to consider for the analysis the parts of a receiver in the form of linear parametric two-port networks. The method for using the effects of modulated interference in dealing with fluctuation noise is determined. The schematic method to improve the vibration resistance by changing the transfer function of the optimum filter and the threshold value is considered. The described results allow drawing the attention to such an important component of protection methods against mechanical impacts, as the vibration response of the instrument in the schematic and system design.

The necessity of creation of a mathematical research apparatus of the thermoblow influence for a printed unit is justified. Using the physical bases of heat and mass transfer and the Fourier equation, a mathematical model, suitable for the simulation of thermoblow for printed units, having a rectangular shape is created. Using the developed model, the numerical solution of the thermoblow task to a rectangular printed unit with installed fuel elements is conducted. The adequacy of the received decision of thermoblow directed to simulation is proven.

The possibility for using a priori information in creation of an intellectual system for assessing the reliability of radio-electronic means (REM) is considered. It is proposed to use the calculated values of reliability indicators, the results of design, and other types of tests as a priori information in the decision support system. Such an approach makes it possible to increase the assessment validity of quantitative indicators of reliability of radio-electronic means with a small volume of tests, which positively affects the cost parameters of equipment.

A technique for application of autonomous blocks to study dynamic processes in heterogeneous structures of devices under applied shock is considered. A mathematical modeling of the casted electronic block is performed. An analysis of spectral density for shock acceleration and the natural frequency spectrum for various casting materials is carried out.

To improve the efficiency of test for determining dynamic characteristics of the structure of electronic tools developed by specifying multi-channel generator. The generator is part of the information-measuring system intended for the dynamic vibration. The description of the structural scheme of the device and principle of operation of information-measuring system. The master clock consists of four independent channels. Each channel is built on the basis of independent modules implementing digital synthesis of frequencies in the range from 1 to 12000 Hz. The chosen architecture defines a multi-channel generator allows you to expand the possibilities of vibration test equipment with separate control parameters for each channel. The proposed structural scheme allows, for each channel, to specify individually the initial phase, the amplitude and frequency of the output harmonic signal. It is also possible to choose the form of the output signal and changes the frequency of the signal at a speed not less than 2 octaves per minute. The latter achieved through the using of modern DDS modules that are optimized to the digital interface, and the application of each DDS module registers two separate frequencies. To optimize the parameters switch between frequency registers in the modules of the applied method, Ramped-FSK.

Background. Investigation of dynamic characteristics of circuit boards in design of instrument devices with the purpose of revealing their vibration relief, determining resonance frequencies and oscillation forms, and also analyzing their vibration resistance under external operational influences, is carried out. In this paper, a technique for determining the locations for installation of measuring devices on circuit boards, including the stage of mathematical modeling of their dynamic state, providing maximum information in the conduct of vibration tests, is proposed. Materials and methods. In the article, the circuit boards with hinged electronic components of instrument devices subject to harmonic vibration loading are considered. A technique for performing experimental and theoretical studies, based on preliminary mathematical modeling of the stress-strain state (SSS) of the circuit board under operational dynamic loads is developed. Modeling of the circuit board SSS is performed using the finite element method (FEM) implemented in ANSYS package, taking into account that the circuit board has a layered heterogeneous structure. Results. A technique that allows the detection of the most effective locations for measuring devices on circuit boards has been developed to obtain maximum information about SSS of the circuit board and its elements during the vibration testing. Conclusions. The conducted numerical studies have shown that the application of the proposed technique at the early stages of designing instrument devices suggests a significant increase in their vibration resistance, helps to reduce time and amount of the performed experimental work, and significantly reduce the economic costs of designing new and upgrading the existing instrument devices.

In this paper, we describe the design of a fiber-optic pressure sensor and the principle of operation. The principle of operation of fiber-optic pressure sensors is as follows. From the radiation source, the light flux along the supply optical fibers is directed to the side of the membrane at an angle, the value of which is selected from the condition of ensuring the maximum sensitivity of the conversion and the depth of modulation of the optical signal. Under the influence of controlled pressure, the membrane deflects, and in the central part the gap between the membrane and the bevelled ends of the optical fibers will be less than the initial value X0. As a result, the reflectivity for electromagnetic waves in the region of the chamfered ends of the optical fibers changes, and accordingly the intensity of the light flux reflected at an angle from the given region, which carries information about the measured pressure and arrives along the diverting optical fibers to the radiation receiver, changes. The optical system of a fiber-optic pressure sensor is calculated, which consists in determining a number of design parameters for a fiber-optic micro-displacement transducer. Mathematical modeling in the MathCAD environment was carried out. Based on the results of mathematical modeling, a new design solution has been proposed for a fiber-optic micro-displacement transducer, in which the incoming and outgoing optical fibers are cut off and located at a calculated angle. The function of converting a fiber-optic micro-displacement transducer is defined. The selected parameters ensure the fulfillment of the conditions for reducing the temperature and dynamic errors, increasing the sensitivity of the optical signal conversion while reducing the mass-dimensional characteristics of the fiber-optic pressure sensor.

The article outlines the publications on measuring the field of vibration movement of the surface of the controlled object. The authors have conducted the modelling of the trace of vibration-induced image blurring with the aim of determining structural parameters that are the input data for calculating the vector components of the vibration movement of the studied material point. The article describes and justifies the use of the new methodology of measuring the modulus and components of the vector of reciprocating vibration movement of the studied material point on the basis of the analysis of geometrical parameters of the trace of vibration-induced image blur of the round mark. The article determines the informative parameters of the geometrical structure of the vibration-induced image blur. The authors suggest and provide reasons for the use of the methodology of measuring these parameters with the following calculation of the modulus and the components of the amplitude vector of vibration movement of a studied material point.

Examines the use of artificial feedforward neural networks and GRNN type networks in solving prediction problems. Exemplified by student research, the authors conducted the study on the choice of the type of neural network that is most suitable for the solution of such problems. The article desctibes the set-up and the training of generalized regression neural network of and feedforward neural network with the use of MATLAB. Both types have been compared. The authors conducted the experiment of predicting the major parameters of student research activities. The obtained data have been compared with actual university data.

Problems of identification and quality management, based on the general principles of systems analysis, are considered. The methodological principles of synthesis of complex systems, based on the developed hierarchical quality criteria structures, are offered. An example of implementation of systems approach is given.

Radio-electronic devices and complexes typically have a multi-level hierarchical structure with complexly interconnected elements within the system and are often used in difficult operating conditions, including the conditions of information conflict. These devices and complexes have to meet strict requirements in reliability, efficiency, and quality. Traditionally, the problems of the analysis of operational temperature modes, electromagnetic compatibility, dynamic stress of radio-electronic devices for strategic purposes were solved at final stages of design by means of conducting bench tests. However, this type of tests demands the presence of a special model or end product and they have to be proceeded by the stages of model design and production, which account for the greater part of production costs and considerable time resources. It is necessary in the process of creating complex radio-electronic complexes and devices to choose, at the early stages of design, the optimal type of construction, the type of cooling system, the elements base, and the operating modes of radio-electronic equipment, that will provide necessary reliability. This choice has to eliminate the errors at the later stages of design. Thus, it is advised to perform structural and parameter optimization simultaneously. The method of analogy may be used for this, where the analysis of the temperature field of radio-electronic equipment is calculated using the heat exchange model after gradual change of parameters of the whole system of heat mode support. As an optimization criterion for the block it is suggested to use the minimum of goal function for η heat mode means and m controlled temperatures on elements of radio-electronic devices in units of the heat model. The suggested criterion reflects the efficiency in the form of interconnected application costs — total of the cost of system elements, power demand, size of blocks and modules of radio-electronic devices, and the resultant effect (quality evaluation) — lower temperature deviation on cooled elements from set values, which provide the required reliability.

Modern electronic media are complex systems with a hierarchical structure, one of which elements are printed circuit boards. Their reliability is laid both at the design stage, and the stage of technological control. The aim of the work is to improve the information-measuring PCB control system with the assessment of the impact of technological variations and external factors on the development of operational defects found. To solve this problem morphological image processing techniques, computer simulation, and, in particular, the finite element method and the methods of fuzzy logic are used. For the numerical solution software package ANSYS is applied. The concept of latent defect PCBs is offered. Improved facet classification of PCBs' defects is up-grated. A technique of detection of latent defects in printed circuit boards is described. A structural scheme of the information-measuring PCB control system is proposed in this article. Introduction of the concept latent defect allows you to extend the range of tolerance deviations of the printed track sizes. Methods of detection of defects, including analysis, defect type classification unit and mathematical modeling allows to determine the further development of a latent defect. The proposed structural scheme of information measurement system implements the proposed method.

Radio-electronic means for various purposes are classified as complex systems and are characterized by a multilevel hierarchical structure in the form of separate units, blocks, and modules. As the system is influenced by a complex of factors in the process of operation, reliability of complex systems functioning depends on the reliability of the system elements and optimal distribution of reliability indicators between the elements of the system. The purpose of this article is to define the reliability requirements for structural elements, taking into account failure rates, and possible gradual changes in parameters of the designed system elements. It is required to ensure the minimum cost for a given level of the system reliability. Systems analysis method, reliability of complex systems theory and probability theory are used in the article with reference to failure rates and changing parameters in the process of operation. This approach provides an opportunity to complement the known methods and to obtain analytical expressions for solving design problems of ensuring the reliability with reference to possible gradual and sudden failures, as well as possible gradual changing in parameters of the designed system elements. It is proposed to apply the developed methodology for defining the reliability requirements for the elements of complex technical systems at the early design stages, when the lack of information makes it impossible to find the optimal distribution of reliability norms for structural elements, providing the minimum total cost of the product. In addition, the method for determining the reliability of structural elements can be used to design highly reliable onboard radio-electronic systems of long-term operation.

The article examines the problem and characteristics of contact less measurement technique of the vibration movement amplitude of the test material point along the z-axis of the object surface. In the article the error assessment of vibration measurements is determined. The authors describe overview and the results of experimental studies.

The paper substantiates the necessity of taking into account of hidden (latent) defects printed conductors. Defines the conditions of transition latent defect of the printed conductor in obvious defect, leading to inoperability of the printed circuit board. Allocated technological defect of the printed conductor in the form of cracks and is made analysis of stress-strain state. Specified conditions of overgrowth defect depending on external factors (mechanical stresses). This model allows to perform numerical simulations of the printed circuit board and to determine the conditions for the transition of latent defects (cracks) in obvious defect, depending on the external impact (acceleration). The results of computational experiments have allowed to reveal mechanical stress in the defect zone depending on the size of the printed conductor and external acceleration.

The design solutions to improve the reliability of onboard radio-electronic systems, consisting of several printed circuit units, are proposed. It is suggested to use several circuit boards, connected by nonlinear damping devices. A model of the three node system, each of which focuses a mass portion of the PCB and mounted components, and is connected by elastic links, is given. The analysis model is designed. It is shown, that the forced oscillations of the boards are extinguished with the damping device of the central node of the mounting. It is due to the fact, that some of the energy radiated by a more active oscillating board with a greater amplitude, will be repaid by the board vibrating at the same time with a lower amplitude. It will ultimately prevent the possibility of oscillations of any board at a frequency close to the resonant one. The proposed approach provides the increased reliability of onboard radio-electronic systems under the conditions of vibrational dynamic loading. A modification of the damping systems in the form of guide rollers, as well as the execution of the floating connectors on the housing of the radio-electronic unit, and the preload of the board package with elastic pad straps to the connectors, are proposed. It will ultimately reduce the mechanical impact on the radio-electronic components of the boards. Mathematical modeling of onboard radio-electronic systems structures, which confirm the validity of the theoretical calculations and design solutions, is carried out.

The article highlights the problems of the efficiency of vibration test methods and means. It gives the brief overview of existing methods. It also gives characteristics of the developed multi-channel vibration measurement machine, its component parts and the benefits of its application. The test results are given that prove that the developed vibration vibration measurement machine allows to reproduce the necessary vibration characteristics similar to real-life vibration exposure during operation life of an object.

The paper provides the overview of publications that study the task of protecting mirror antennas from external vibration influences. We suggest a new technique of contactless measurement and the analysis of the modulus and components of the vector for vibration amplitude in mirror antenna elements. We have analyzed the structural parameters of the trace of vibration-induced image blurring of the light-reflecting or fluorescent round tag with the aim of indirect contactless measuring of the modulus and components of the vector for vibration amplitude on the surface of a mirror antenna. We have provided justification for measuring half length, half width and the tilt angle of the trace left by round tag image vibratory blurring, as well as comparing these parameters with the parameters of round tag image with vibrations absent. We have given justification of the calculation technique that connects the parameters of the trace of vibration blurring with the parameters of vector of the vibration amplitude of the examined tag on the surface of a mirror antenna. The technique is based on the comparative analysis of binary image of a round tag in the conditions when the vibrations are absent and the trace of blurring when these vibrations are present. The study of geometric parameters of the trace of vibration-induced image blurring of the round tag involves the calculation of characteristic length of each pixel of this trace. The characteristic distance is understood as the distance between the pixel centre and the gravity centre of the tag image at the initial position with absent vibrations. This value is approximated to the integer of pixel units. The pixels with equal characteristic distance are formed into reference segments, which are subdivided into rings and arcs. The maximum characteristic distance of ring segment pixels constitute the half width, and the maximum characteristic distance of curve segment is the half length of the trace of vibration-induced image blurring of the round tag.

A conceptual model for determination of vibro-loaded and vibro-safe areas of the board, based on the analysis of flexural vibrations of rods and spread of elastic deformation waves from the point sources with cylindrical symmetry in the body of the plate, is proposed. A scheme for the experimental installation with the use of different types of sensors to check the model and to study the dynamic behavior of the plates and the rods, as elements of structures of radio-electronic means (REM), is offered.