Search

Due to the complex and variable marine environment and weak target echo signals, the difficulty of extract underwater target features increases. On the basis of analyze the nonlinear and non-stationary characteristics of underwater target signals, in order to improve the detection performance of active sonar, the hierarchical refined multi-scale fluctuation-based dispersion entropy (HRCMFDE) method is proposed and used to extract the features of underwater target signals. Firstly, the HRCMFDE method is used to extract features from underwater target signals, and the extracted HRCMFDE feature vectors are used to form the feature set. Secondly, genetic algorithm (GA) was used to optimize the input weights and hidden layer bias of extreme learning machine (ELM), and an improved ELM algorithm was proposed to establish the recognition model for underwater target signals. Finally, a sample library of five types of targets was constructed based on the measured underwater target signals, and feature extraction was performed. The performance of the extracted features was tested through improved ELM classification, and the classification results obtained verified the effectiveness of the proposed feature extraction method. [ABSTRACT FROM AUTHOR]

Crop yield prediction (CYP) is a critical challenge for decision-makers of any kind of levels, including regional and national decision-making. Farmers might use effective CYP model to help them determine what to cultivate or when to grow it. The primary aim of precision agriculture (PA) is to increase crop growth and yield thereby decreasing production costs and emissions. Various developmental factors influence potential yield, including soil properties, irrigation, weather, fertilizer maintenance, and topography. CYP is critical to worldwide agricultural production. To improve the global food security, policymakers depend on reliable forecasts to make timely exports and imports decision. The main aim of this paper is to predict the crop yield using machine-learning technique. The collected data are pre-processed using Kalman filter algorithm and then certain features are extracted using the Linear Discriminant Analysis (LDA). The CYP is done using the improved extreme learning machine (IELM). The output weight of the extreme learning machine is improved using the chimp optimization algorithm (COA). The implementation tool used in this method is PYTHON and the dataset used for the CYP is farm yield prediction. The experimental results showed an accuracy of 99.99% which is better when compared to existing methods. [ABSTRACT FROM AUTHOR]

Accurate modeling and prediction of dam deformation contribute to the analysis of dam safety. Research on data-driven modeling of dam deformation has received increasing attention. However, most established models cannot comprehensively consider the physical characteristics (such as irreversible trend changes, periodic changes, and random changes) and hysteresis characteristics of the deformation sequence. Therefore, an improved extreme learning machine-based dam deformation prediction model was proposed. The parameters of the extreme learning machine were optimized using the improved salp swarm algorithm. For identifying physical characteristics, seasonal and Trend decomposition using Loess was used to decompose the deformation sequence into trend, periodic, and residual items. For the identification of hysteresis characteristics, the phase space reconstruction theory was adopted to solve the problem of selecting the time lag of the deformation subsequence with chaotic characteristics. The proposed model was used to predict the deformation of a concrete dam in China. Furthermore, the model outperformed other alternatives, thus providing a new solution for dam deformation prediction. [ABSTRACT FROM AUTHOR]

• A novel hybrid system is proposed with deterministic prediction and interval prediction. • Outlier detection and correction method are used to reduce the complexity of the series. • Data decomposition strategy is applied to decompose the original series to improve the prediction accuracy. • A multi-objective optimization algorithm is selected to determine the original weights and biases of the forecasting model. • Applied interval prediction to quantify the range of concentration change of air pollutants caused by uncertainties. A novel system regarding deterministic and interval predictions of pollutant concentration is constructed in this study, which can not only obtain higher prediction accuracy in deterministic prediction and also provide effective interval prediction of air pollutant concentration. In the deterministic prediction stage, the improved extreme learning machine combines outlier detection and correction algorithm, data decomposition strategy, and a multi-objective optimization algorithm to form a hybrid model for predicting pollutant concentration. Moreover, the applicability of the optimization algorithm was verified from theoretical and experimental analysis. In the interval prediction stage, three distributions are compared to mine, the traits of deterministic prediction errors are analyzed, and interval prediction is designed to quantify the uncertainties associated with pollutant concentration. To investigate the prediction performance of the proposed system, comparison experiments have been executed using the PM 2.5 concentration series from three cities. The results indicate that the system proposed in this paper outperforms comparison models in forecasting accuracy and has advantages for pollution prediction. [ABSTRACT FROM AUTHOR]

Summary: The traditional statistical models for concrete dam deformation do not consider the effective contribution to residual sequence and thus results in the low accuracy of deformation prediction of the monitoring model and in the insufficient prediction of dam deformation behavior. This study proposes a combined model for concrete dam deformation, considering residual correction by frequency division. The horizontal displacement of concrete dams is driven by water pressure, temperature, and aging factors. In this study, an accurate value of water pressure deformation is obtained by finite element calculation, and then the hybrid model is constructed. However, the temperature and aging components hardly obtain accurate information values from the traditional model. Therefore, a combined model is constructed to extract additional detailed information that characterizes the dam deformation from the residual sequence. Ensemble empirical mode decomposition is selected to analyze and denoise the residual sequence adaptively, and the consecutive mean square error is used to determine the boundary points of high‐ and low‐frequency components. Then, the combined model based on the extreme learning machine method optimized by particle swarm optimization and seasonal autoregressive integrated moving average is constructed for high‐ and low‐frequency sequences. Lastly, the prediction results of different subitems are organically integrated. Results of a case study in engineering demonstrate that the accuracy of the combined model for concrete dam deformation is higher than that of the classical numerical model, and the model can effectively overcome noise interference in the monitoring sequence. The proposed combined model for deformation is reliable for analyzing and judging the dam monitoring time series, which can provide new ideas and reference for the data processing of dam deformation monitoring. [ABSTRACT FROM AUTHOR]

Big data mining has emerged as an active field of interest, and traditional data mining approaches frequently fail to handle the complexities associated with massive datasets. One of the most extensively used strategies for big data classification is MapReduce, which combines the map and reduce processes. For filtering and sorting, the mapping approach is employed, and the reduction technique is used to combine the final classification results. A novel Archerfish Hunter Spotted Hyena Optimization-based Improved Extreme Learning Machine (AHSHO-IELM) classifier-based MapReduce framework is proposed in this paper for big data classification. The IELM algorithm is formed by integrating the ELM technique with Principal Component Analysis to overcome the multicollinear problem and enhance the training and testing time. The AHSHO method combines the archerfish hunter optimization and Spotted Hyena Optimization algorithms to improve the optimal parameter selection of the IELM classifier, which increases classification accuracy and reduces the error rate. The performance of the proposed AHSHO-IELM classifier-based MapReduce framework is evaluated using different performance metrics such as Accuracy, Sensitivity, Specificity, Computational time, F1-Score, Mathews correlation coefficient, and scale-up factor. For the rotten tomatoes movie review dataset and the Dermatology dataset, the proposed approach yields accuracy, specificity, and sensitivity values of 99%, 99%, 98.3%, and 99.3%, 99%, 98%, respectively for a mapper value (X) of 5. The proposed big data classifier is effective for both single-class and multi-class classification, according to the results of the analysis. [ABSTRACT FROM AUTHOR]

Air pollution forecasting plays a pivotal role in environmental governance, so a large number of scholars have devoted themselves to the study of air pollution forecasting models. Although numerous studies have focused on this field, they failed to consider fully the linear feature, non-linear feature, and fuzzy features contained in the original series. To fill this gap, a new combined system is built to consider features in the original series and accurately forecast PM2.5 concentration, which incorporates an efficient data decomposition strategy to extract the primary features of the PM2.5 concentration series and remove the noise component, and five forecasting models selected from three types of models to obtain the preliminary forecasting results, and a multi-objective optimization algorithm to combine the prediction results to produce the final prediction values. Empirical studies results indicated that in terms of RMSE the developed combined system achieves 0.652 6%, 0.810 1%, and 0.775 0% in three study cities, respectively. Compared to other prediction models, the RMSE improved by 60% on average in the study cities.

Abstract This paper develops a Legendre neural network method (LNN) for solving linear and nonlinear ordinary differential equations (ODEs), system of ordinary differential equations (SODEs), as well as classic Emden–Fowler equations. The Legendre polynomial is chosen as a basis function of hidden neurons. A single hidden layer Legendre neural network is used to eliminate the hidden layer by expanding the input pattern using Legendre polynomials. The improved extreme learning machine (IELM) algorithm is used for network weights training when solving algebraic equation systems, and several algorithm steps are summed up. Convergence was analyzed theoretically to support the proposed method. In order to demonstrate the performance of the method, various testing problems are solved by the proposed approach. A comparative study with other approaches such as conventional methods and latest research work reported in the literature are described in detail to validate the superiority of the method. Experimental results show that the proposed Legendre network with IELM algorithm requires fewer neurons to outperform the numerical algorithm in the latest literature in terms of accuracy and execution time.

In machining process, tool wear is an inevitable consequence which progresses rapidly leading to a catastrophic failure of the system and accidents. Moreover, machinery failure has become more costly and has undesirable consequences on the availability and the productivity. Consequently, developing a robust approach for monitoring tool wear condition is needed to get accurate product dimensions with high quality surface and reduced stopping time of machines. Prognostics and health management has become one of the most challenging aspects for monitoring the wear condition of cutting tools. This study focuses on the evaluation of the current health condition of cutting tools and the prediction of its remaining useful life. Indeed, the proposed method consists of the integration of complex continuous wavelet transform (CCWT) and improved extreme learning machine (IELM). In the proposed IELM, the hidden layer output matrix is given by inverting the Moore–Penrose generalized inverse. After the decomposition of the acoustic emission signals using CCWT, the nodes energy of coefficients have been taken as relevant features which are then used as inputs in IELM. The principal idea is that a non-linear regression in a feature space of high dimension is involved by the extreme learning machine to map the input data via a non-linear function for generating the degradation model. Then, the health indicator is obtained through the exploitation of the derived model which is in turn used to estimate the remaining useful life. The method was carried out on data of the real world collected during various cuts of a computer numerical controlled tool. [ABSTRACT FROM AUTHOR]

Traffic sign detection (TSD) in real-time environment holds great importance for applications such as automated-driven vehicles. Large variety of traffic signs, different appearances, and spatial representations causes a huge intraclass variation. In this article, an extreme learning machine (ELM), convolutional neural network (CNN), and scale transformation (ST)-based model, called improved extreme learning machine network, are proposed to detect traffic signs in real-time environment. The proposed model has a custom DenseNet-based novel CNN architecture, improved version of region proposal networks called accurate anchor prediction model (A2PM), ST, and ELM module. CNN architecture makes use of handcrafted features such as scale-invariant feature transform and Gabor to improvise the edges of traffic signs. The A2PM minimizes the redundancy among extracted features to make the model efficient and ST enables the model to detect traffic signs of different sizes. ELM module enhances the efficiency by reshaping the features. The proposed model is tested on three publicly available data sets, challenging unreal and real environments for traffic sign recognition, Tsinghua-Tencent 100K, and German traffic sign detection benchmark and achieves average precisions of 93.31%, 95.22%, and 99.45%, respectively. These results prove that the proposed model is more efficient than state-of-the-art sign detection techniques.

A novel fault diagnosis approach based on a combination of discrete wavelet transform, phase space reconstruction, singular value decomposition, and improved extreme learning machine is presented in rolling bearing fault identification and classification. The proposed method provides proper solutions for improving the accuracy of faults classification. To achieve this goal, initial signals are divided into sub-band wavelet coefficients using discrete wavelet transform. Then, each of sub-band is mapped into three-dimensional space using the phase space reconstruction method to completely describe characteristics in the high dimension. Thereafter, singular values are calculated by singular value decomposition method, which demonstrate crucial variances in original vibration signal. Lastly, an improved extreme learning machine is adopted as a classifier for fault classification. The proposed method is applied to the rolling bearing fault diagnosis with non-linear and non-stationary characteristics. Based on outputs of the improved extreme learning machine, the working condition and fault location could be determined accurately and quickly. Achieved results, compared with other schemes, show that the proposed scheme in this article can be regarded as an effective and reliable method for rolling bearing fault diagnosis. [ABSTRACT FROM AUTHOR]

• The proposed method adopts K-means, FFT, and coefficient of kurtosis to improve the signal preprocessing ability of SGMD, which makes different types of disturbance features easy to be extracted. • It adopts an optimization algorithm to optimize the feature extraction of multiple methods and the disturbance identification of ELM, which improves the accuracy of feature extraction and disturbance identification. • It adopts tent chaotic mapping and adaptive parameters to further improve the optimization performance of MPA. Power quality is the quality of electrical energy in the power system. And the power quality disturbances (PQDs) in the power system lead to severe consequences for the equipment's life and the operator's safety. The accurate disturbance identification is a crucial prerequisite to managing power quality. Currently, the identification accuracy of PQDs needs to be improved, especially in high-intensity noise environments. In this paper, a novel power quality identification method based on adaptive symplectic geometric mode decomposition (ASGMD) and tent chaotic map and adaptive parameter method improved marine predators algorithm (IMPA) optimization is proposed. Firstly, the PQDs are pre-processed by adaptive symplectic geometric mode decomposition with K-means, fast Fourier transform, coefficient of kurtosis, and wavelet thresholding to reduce the noise component. Secondly, twenty-three PQDs features of the wavelet transform, fast Fourier transform, and mathematical statistics are optimally selected by IMPA to improve the identification accuracy and efficiency. Finally, the PQDs with selected features are recognized by the extreme learning machine with optimized weights and biases. IMPA improves the uniformity of the initial population distribution and the optimization process of three iterative phases, thus promoting the search accuracy of the optimal features, input weights, and biases. The proposed method has more accurate identification accuracy and fast identification speed by simulating different noise intensity environments than other methods. Remarkably, the identification accuracy reaches 91% in high-intensity noise environments of 10dB This method has practical significance for PQDs identification of large areas, long periods, and high noise-resistant. [ABSTRACT FROM AUTHOR]

Carbon emissions and environmental protection issues have become the pressure from the international community during the current transitional stage of China's energy transformation. China has set a macro carbon emission target, which will reduce carbon emissions per unit of Gross Domestic Product (GDP) by 40% in 2020 and 60–65% in 2030 than that in 2005. To achieve the emission reduction target, the industrial structure must be adjusted and upgraded. Furthermore, it must start from a high-pollution and high-emission industry. Therefore, it is of practical significance to construct a low-carbon sustainability and green operation benefits of power generation enterprises to save energy and reduce emissions. In this paper, an intuitionistic fuzzy comprehensive analytic hierarchy process based on improved dynamic hesitation degree (D-IFAHP) and an improved extreme learning machine algorithm optimized by RBF kernel function (RELM) are proposed. Firstly, we construct the evaluation indicator system of low-carbon sustainability and green operation benefits of power generation enterprises. Moreover, during the non-dimensional processing, the evaluation index system is determined. Secondly, we apply the evaluation indicator system by an empirical analysis. It is proved that the D-IFAHP evaluation model proposed in this paper has higher accuracy performance. Finally, the RELM is applied to D-IFAHP to construct a combined evaluation model named D-IFAHP-RELM evaluation model. The D-IFAHP evaluation results are used as the input of the training sets of the RELM algorithm, which simplifies the comprehensive evaluation process and can be directly applied to similar projects. [ABSTRACT FROM AUTHOR]