Search

In this note, we rely on the theoretical foundations of Radial Basis Function Neural Networks (RBF-NNs) to form an adaptive parameter estimation problem, which we solve using the recently introduced Prescribed Performance Control methodology. Such combination results in a compact, user-congurableadaptive nonlinear system identication methodology that can be used to retrieve the open-loop nonlinear plant dynamics in any compact region of interest.

The problem of quick and accurate fault estimation in nonlinear systems is addressed in this article by combining the technique of radial basis function neural network (RBFNN) and global fast terminal sliding mode control (GFTSMC) concept. A new strategy to update the neural network weights, by using the global fast terminal sliding surface instead of conventional error back propagation method, is introduced to achieve real time, quick and accurate fault estimation which is critical for fault tolerant control system design. The combination of online learning ability of RBFNN, to approximate any nonlinear function, and finite time convergence property of GFTSMC ensures quick detection and accurate estimation of faults in real time. The effectiveness of the proposed strategy is demonstrated through simulations using a nonlinear model of a commercial aircraft and considering a wide range of sensors and actuators faults. The simulation results show that the proposed method is capable of quick and accurate online fault estimation in nonlinear systems and shows improved performance as compared to conventional RBFNN and other techniques existing in literature.

Agriculture and farming are mainly dependent on weather especially in Malaysia as it received heavy rainfall throughout the years. An efficient crop or tree management system with a weather forecast needed for suitable planning of farming operation. Radial Basis Function Neural Network (RBFNN) algorithm was used in this study to predict rainfall and the main focus of this study is to analyze the factor that affects the performance of neural model. This study found that the model works better the more hidden nodes and the optimum learning rate is 0.01 with the RMSE 49% and the percentage accuracy is 57%. Besides that, it is found that the meteorology data also affect the model performance. Future research can be conducted to improve the rainfall forecast of this study and improve the tree management system.

Phase-only nulling with low sidelobe level is a problem of interest in array synthesis which is a tedious problem without an analytical solution. In this communication, a novel framework for the phase-only nulling based on radial basis function neural network (RBFNN) is proposed to predict the phase adjustment for the array pattern nulling with sidelobe control. In the process of network training, the parameters of the RBFNN are optimized simultaneously based on the self-adaptive differential evolution (SADE) algorithm, which aims to improve the approximation ability and reduce the complexity of the network. Simulation results of the optimized RBFNN models show compact network structure and form the array pattern under desired performance with good generalization capability.

Limits in the real-time computation of micro-processors on the one hand and high-level controllers that need precise computation in addition to implementation and compiling issues, on the other hand, have caused a great gap between control science and experiments. In this work, an adaptive RBF neural network controller is proposed to control the position and attitude of an autonomous multirotor. The controller is combined with an EKF observer and simulated in real-time flight conditions. In order to check the capabilities of the system, the proposed structure has been successfully applied to a quadrotor and a hexarotor using three data types for getting similar real-time flight results. Based on the results, the proposed structure has accomplished two separate missions with different scenarios, though there exists some error, especially in position estimations which are caused by step-wise characteristics of the desired path.

The difference between the actual feeder parameters and feeder parameter data stored in a database or offered by manufacturers is significant owing to the ambient environment, temperature, and skin effect. Here, a parameter estimation method is proposed for unbalanced three‐phase distribution feeders based on the bus voltages and branch power flows measured from two terminals of the feeder. In the proposed method, a high‐precision phasor measurement unit is not required to estimate the magnitude and phase angle of the phasor quantity using a common time source for synchronisation. A radial basis function neural network with multi‐run optimisation (RBFNN‐MRO) is proposed to map the complex nonlinear relations between the distribution feeder parameters and electrical quantities. The feasibility and performance of the proposed RBFNN‐MRO method were verified using the four IEEE test systems. The comparison between the proposed RBFNN‐MRO method and the multi‐run method based on the quasi‐Newton method is implemented via the maximum absolute percentage error (MAPE) curves. The results reveal that the proposed RBFNN‐MRO method has excellent potential for improving the accuracy of feeder parameter estimation, even for bad data preparation.

Micro-Expressions (ME) has turned out to be one amongst the most vital clues for lies, as well as numerous other applications. It is the brief Facial Expressions (FE) that the people express whilst endeavouring to conceal, hide, or repress their emotions. In a real environment, recognizing facial ME is really very hard than recognizing the general-FE, which shows rich emotions. Here, to attain the most accurate classification outcomes of Micro-Expression Recognition (MER), a combination of Feature Learning (FL) algorithms is proposed. The Video Frames (VF) is taken as the input for MER. First, the inputted-VF is preprocessed and then a Viola-Jones Algorithm is carried out to attain the face images of the VF. The SB-DCNN extracts the Facial Land-Marks as of the face images. Next, for performing the effectual FL, the combination of Weight matrix-based Temporal Accumulated Optical Flow and Improved LBP features (WTAOF-ILPB) are employed. Lastly, to recognize the ME of the input VF, the features as of SB-DCNN and WTAOF-ILPB are rendered to the optimal adaptation of Radial Basis Function Neural Network (RBFNN), i.e. Levy Flight based Shuffled Shepherd Optimization Algorithm (LFSSOA)–RBFNN (LFSSOA-RBFNN). The proposed method’s performance is analyzed by executing an experiment. The outcomes exhibited that the proposed LFSSOA-RBFNN trounces other prevailing algorithms with respect to some performance metrics.

The aim of the research presented in this article was to determine the value of the friction coefficient using a simple tribological test and to build an empirical model of friction with the use of radial basis function artifi-cial neural networks. The friction tests were carried out on a specially designed friction simulator that allows a sheet metal strip to be drawn between two fixed dies. The test materials were sheets of Ti-6Al-4V titanium alloy with a thickness of 0.5 mm. The friction tests were carried out with variable contact forces of counter-samples with rounded surfaces and in various lubrication conditions. Mineral oils and bio-degradable oils with the addition of boric acid (5 wt %) were tested. Based on the results of friction investigations, neural models of friction were built using RBF artificial neural networks. The good properties of the RBF network 2:2-35-1:1 were confirmed by a high value of the determination coefficient R2 = 0.9984 and a low value of the S.D. ratio equal to 0.0557. It was found that the COF value was the highest for the average values of both the nominal pressure and kinematic viscosity. Over the entire range of nominal pressures applied, SAE10W-40 engine oil ensured the most effective reduction of the COF. The COF value was the highest for the average values of both the nominal pressure and kinematic viscosity.

This paper aims to present a comprehensive survey on water quality soft-sensing of a wastewater treatment process (WWTP) based on artificial neural networks (ANNs). We mainly present problem formulation of water quality soft-sensing, common soft-sensing models, practical soft-sensing examples and discussion on the performance of soft-sensing models. In details, problem formulation includes characteristic analysis and modeling principle of water quality soft-sensing. The common soft-sensing models mainly include a back-propagation neural network, radial basis function neural network, fuzzy neural network (FNN), echo state network (ESN), growing deep belief network and deep belief network with event-triggered learning (DBN-EL). They are compared in terms of accuracy, efficiency and computational complexity with partial-least-square-regression DBN (PLSR-DBN), growing ESN, sparse deep belief FNN, self-organizing DBN, wavelet-ANN and self-organizing cascade neural network (SCNN). In addition, this paper generally discusses and explains what factors affect the accuracy of the ANNs-based soft-sensing models. Finally, this paper points out several challenges in soft-sensing models of WWTP, which may be helpful for researchers and practitioner to explore the future solutions for their particular applications.

The main objective of this study was to produce flood susceptibility maps for Tajan watershed, Sari, Iran using three machine learning (ML) models including Self-Organization Map (SOM), Radial Basis Function Neural Network (RBFNN), and Multi-layers Perceptron (MLP). To reach such a goal, different physical-geographical factors (criteria) were integrated and mapped. 212 flood inventory map was randomly divided into training and testing datasets, where 148 flood locations (70%) were used for training and the remaining 64 locations (30%) were employed for testing. Model validation was performed using several statistical indices and the area under the curve (AUC). The results of the correlation matrix showed, three factors slope (0.277), distance from river (0.263), and altitude (0.223) were the most important factors affecting flood. The accuracy evaluation of the flood susceptibility maps through the AUC method and K-index shows that in the validation phase RBFNN (AUC = 0.90) outperform the MLP (AUC = 0.839) and SOM (AUC = 0.882) models. The highest percentage flood susceptibility of the area in MLP, SOM and RBFNN models is related to moderate (28.7%), very low (40%) and low (37%), respectively. Also, the validation results of the models using the Relative Flood Density (RFD) approach showed that very high class had the highest RFD value.

In order to improve the performance of predictive energy management strategies (PEMS), a novel neural network based vehicle velocity prediction strategy (NN-VVP) was proposed. First, an online trained radial basis function neural network (RBF-NN) with a fixed structure was adopted to build online vehicle velocity prediction (VVP) model. The influence of order and width of RBF-NN on the online prediction accuracy was studied in depth, it was found that RBF-NN with a fixed structure was not always suitable for the overall online prediction process. Then, by introducing a neural network structure determination method (SDM) with the Akaike Information Criterion (AIC), an adaptive RBF-NN which adjust structure in real time was designed to perform online VVP to further improve the prediction accuracy. Simulation results indicate that, the VVP strategy proposed in this paper predicts the future vehicle velocity with acceptable accuracy. Compared with the fixed structure, the RBF-NN with an adaptive structure significantly improve the prediction accuracy by approximately 63.2%, 70.4%, and 71.1%.

The prediction of an accurate geodetic point velocity has great importance in geosciences. The purpose of this work is to explore the predictive capacity of three artificial neural network (ANN) models in predicting geodetic point velocities. First, the multi-layer perceptron neural network (MLPNN) model was developed with two hidden layers. The generalized regression neural network (GRNN) model was then applied for the first time. Afterwards, the radial basis function neural network (RBFNN) model was trained and tested with the same data. Latitude ( $$\varphi$$ ) and longitude (λ) were utilized as inputs and the geodetic point velocities ( $${V}_{X}$$ , $${V}_{Y}$$ , $${V}_{Z}$$ ) as outputs to the MLPNN, GRNN, and RBFNN models. The performances of all ANN models were evaluated using root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination ( $${\text{R}}^{2}$$ ). The first investigation demonstrated that it was possible to predict the geodetic point velocities by using all the components as output parameters simultaneously. The other result is that all ANN models were able to predict the geodetic point velocity with satisfactory accuracy; however, the GRNN model provided better accuracy than the MLPNN and RBFNN models. For example, the RMSE and MAE values were 1.77–1.88 mm and 1.44–1.51 mm, respectively, for the GRNN model.

Purpose: The purpose of this paper was to design an intelligent controller of ship motion based on sliding mode control with a Radial Basis Function (RBF) neural network optimized by the genetic algorithm and expansion observer. First, the improved genetic algorithm based on the distributed genetic algorithm with adaptive fitness and adaptive mutation was used to automatically optimize the RBF neural network. Then, with the compensation designed by the RBF neural network, antisaturation control was realized. Additionally, the intelligent control algorithm was introduced by Sliding Mode Control (SMC) with the stability theory. A comparative study of sliding mode control integrated with the RBF neural network and proportional-integral-derivative control combined with the fuzzy optimization model showed that the stabilization time of the intelligent control system was 43.75% faster and the average overshoot was reduced by 52% compared with the previous two attempts. Background: It was known that the Proportional-Integral-Derivative (PID) control and self-adaptation control cannot really solve the problems of frequent disturbance from external wind and waves, as well as the problems with ship nonlinearity and input saturation. So, the previous ship motion controller should be transformed by advanced intelligent technology, on the basis of referring to the latest relevant patent design methods. Objective: An intelligent controller of ship motion was designed based on optimized Radial Basis Function Neural Network (RBFNN) in the presence of non-linearity, uncertainty, and limited input. Methods: The previous ship motion controller was remodeled based on Sliding Mode Control (SMC) with RBFNN optimized by improved genetic algorithm and expansion observer. The intelligent control algorithm integrated with genetic neural network solved the problem of system model uncertainty, limited control input, and external interference. Distributed genetic with adaptive fitness and adaptive mutation method guaranteed the adequacy of search and the global optimal convergence results, which enhanced the approximation ability of RBFNN. With the compensation designed by the optimized RBFNN, it was realized anti-saturation control. The chattering caused by external disturbance in SMC controller was reduced by the expansion observer. Results: A comparative study with RBFNN-SMC control and fuzzy-PID control, the stabilization time of the intelligent control system was 43.75% faster, the average overshoot was reduced by 52%, compared to the previous two attempts. Conclusion: The intelligent control algorithm succeed in dealing with the problems of nonlinearity, uncertainty, input saturation, and external interference. The intelligent control algorithm can be applied into research and development ship steering system, which would be created a new patent.

In this article, an adaptive radial basis function neural network scheme for trajectory tracking control of surface vehicles is proposed. Under complex uncertainties, the proposed controller is designed by combining radial basis function neural network and finite-time control algorithm. Using the novel controller, the stability of accurate trajectory tracking can be ensured and the robustness of control system can be improved. Theoretical proof is proposed by Lyapunov function that the radial basis function neural network controller can make surface vehicle to accurately track desire trajectory steadily. Simulation studies conducted on a prototype CyberShip II demonstrate remarkable performance of proposed control scheme.

Dissolved oxygen (DO) concentration is a key variable in the activated sludge wastewater treatment processes. In this paper, an auto control strategy based on Euler method and gradient method with radial basis function (RBF) neural networks (NNs) is proposed to solve the DO concentration control problem in an activated sludge process of wastewater treatment. The control purpose is to maintain the dissolved oxygen concentration in the aerated tank for having the substrate concentration within the standard limits established by legislation of wastewater treatment. For that reason, a new proposed control strategy based on gradient descent method and RBF neural network has been used. Compared with RBF neural network PI control, the obtained results show the effectiveness in terms of both transient and steady performances of proposed control method for dissolved oxygen control in the activated sludge wastewater treatment processes.

The three-dimensional structures and shapes of biomolecules provide essential information about their interactions and functions. Unfortunately, the computational cost of biomolecular shape representation is an active challenge which increases rapidly as the number of atoms increase. Recent developments in sparse representation and deep learning have shown significant improvements in terms of time and space. A sparse representation of molecular shape is also useful in various other applications, such as molecular structure alignment, docking, and coarse-grained molecular modeling. We have developed an ellipsoid radial basis function neural network (ERBFNN) and an algorithm for sparsely representing molecular shape. To evaluate a sparse representation model of molecular shape, the Gaussian density map of the molecule is approximated using ERBFNN with a relatively small number of neurons. The deep learning models were trained by optimizing a nonlinear loss function with L1 regularization. Experimental results reveal that our algorithm can represent the original molecular shape with a relatively higher accuracy and fewer scale of ERBFNN. Our network in principle is applicable to the multiresolution sparse representation of molecular shape and coarse-grained molecular modeling. Executable files are available at https://github.com/SGUI-LSEC/SparseGaussianMolecule. The program was implemented in PyTorch and was run on Linux.

Named data networking (NDN) corresponds to content‐centric networking, content‐based networking, and data‐oriented networking. Future internet architecture is modelled to overcome the fundamental limitations of the present internet protocol‐based internet and to provide specific strong security. Caching is a key NDN feature in the network. However, pervasive caching strengthens security issues in particular cache pollution assaults including cache poisoning (e.g. presenting malicious content in caches as false‐locality) and cache pollution (e.g. unpopular content is ruined with cache locality as locality‐disruption). In this work, a new cache replacement method based on the radial basis function neural network (RBFNN) is proposed to detect and mitigate the cache pollution attacks in NDN. RBFNN framework is constructed utilising the input associated with the cached content inherent characteristics and output data associated with the content type, i.e. locality‐disruption, false‐locality, and healthy. Experimental results show the efficiency as well as the effectiveness of the proposed method in terms of hit damage ratio and computational time.

This work presents a novel approach by considering teaching learning based optimization (TLBO) and radial basis function neural networks (RBFNs) for building a classifier for the databases with missing values and irrelevant features. The least square estimator and relief algorithm have been used for imputing the database and evaluating the relevance of features, respectively. The preprocessed dataset is used for developing a classifier based on TLBO trained RBFNs for generating a concise and meaningful description for each class that can be used to classify subsequent instances with no known class label. The method is evaluated extensively through a few bench-mark datasets obtained from UCI repository. The experimental results confirm that our approach can be a promising tool towards constructing a classifier from the databases with missing values and irrelevant attributes.

With the increasing aging level of China’s population, the probability of suffering from Alzheimer’s disease also increases. Genetic Algorithm-Back-Propagation (GA-BP) were introduced and compared, and the effectiveness of Genetic Algorithm-Radial Basis Function (GA-RBF) algorithm in diagnosis and prediction of Alzheimer’s disease was verified by simulation experiments. The results showed that GA-RBF modeling time increased by about 75%, It showed that GA-RBF algorithm can be used helpful to accurately predict the cognitive status of the elderly.

In this article, an artificial neural network is proposed to classify short audio sequences of squeak and rattle (S&R) noises. The aim of the classification is to see how accurately the trained classifier can recognize different types of S&R sounds. Having a high accuracy model that can recognize audible S&R noises could help to build an automatic tool able to identify unpleasant vehicle interior sounds in a matter of seconds from a short audio recording of the sounds. In this article, the training method of the classifier is proposed, and the results show that the trained model can identify various classes of S&R noises: simple (binary classification) and complex ones (multi class classification). © 2020 Institute of Noise Control Engineering.

To improve the computational cost and approximate optimization accuracy of solving engineering optimization problems with both expensive objective and constraints, a hybrid sequential approximate optimization (HSAO) method is proposed. First, the radial basis function neural network (RBFNN) and response surface method (RSM) approximate models fit the optimization objectives and constraints, respectively. Second, the fitted model searches for the optimal solution via global optimization algorithm, and then, the optimal solution is used as a newly added sample point for updating approximation model. The region of the optimal solution is iteratively updated until the optimal solution converges. The performance of the HSAO method is tested on numerical test functions. Compared with the traditional approximate model, adaptive RSM, and adaptive RBFNN approximate optimization, the results indicate that the proposed HSAO method is higher in search ability and efficiency. Finally, the method is applied in S-shaped tube crashworthiness optimization, and has achieved desired results.

Crosswell electromagnetic (EM) method has fundamentally improved the horizontal detection ability of well logging and will become an increasingly promising approach for the secondary exploration of hydrocarbon reservoir. We applied orthogonal least squares (OLS) radial basis function neural network (RBFNN) based on improved Gram–Schmidt (G–S) procedure to three-dimensional (3D) crosswell EM inversion problems. In the inversion process of the simplified crosswell model with single-grid conductivity anomalies and normal oil reservoir, compared the inversion results of other five neural networks, OLS-RBFNN was proved to have the best global optimization ability and the fastest sample learning speed and the average inversion error of low conductivity anomalies model (4%) and oil reservoir model (9%) can meet the inversion requirements of crosswell EM method. Only the OLS-RBFNN could achieve ideal inversion results in the most concerned central area of crosswell model, and the inversion accuracy of this algorithm will be more outstanding when the model becomes more complex. Merely using the three-component time-domain crosswell EM data of two wells, the inversion of 3D medium conductivity in the crosswell dominant exploration area can be effectively realized through the nonlinear approximation of the OLS-RBFNN.

To alleviate the nonuniform error distribution and slow convergence caused by the uncertainty of sample selection in the training process of artificial neural networks, a semisupervised radial basis function neural network (SS-RBFNN) model with a new sampling strategy is proposed for parametric modeling of microwave components in this article. After evaluating the current training performance, the new sampling strategy selects suitable training samples to ensure each subregion of the whole sampling region with the same level of training and testing accuracy. Meanwhile, the proposed SS-RBFNN simplifies the modeling process to further enhance the modeling accuracy and efficiency. Two numerical examples of a slow-wave defected ground structure dual-band bandpass filter and a microstrip-to-microstrip vertical transition are employed to verify the effectiveness of the proposed model.

In this paper, we propose an effective method of $$\hbox {PM}_{2.5}$$ prediction based on image contrast-sensitive features and weighted bagging based neural network (WBBNN). Three types of image contrast-sensitive features are first extracted from the images and fuzzified. Next, a weighted bagging strategy combining the ensemble fuzzy neural network (FNN) and ensemble radial basis function neural network (RBFNN) is established. The ensemble neural network (NN), regardless of FNN and RBFNN, is obtained by simply averaging the outputs of component neural networks. And these component neural networks are trained by the improved gradient descent algorithm and samples acquired by bootstrap sampling. Finally, the WBBNN is used to forecast $$\hbox {PM}_{2.5}$$ concentration by extracting three types of image contrast-sensitive features. Results of experiments demonstrate that our prediction method is more reliable, practical and efficient than FNN, RBFNN, the ensemble NNs, and state-of-the-art quality assessment method in terms of predicting the concentration of $$\hbox {PM}_{2.5}$$. More importantly, an improved gradient descent algorithm is developed to accelerate the convergence speed and ensure the prediction accuracy of WBBNN and the fuzzy features acquired by the feature fuzzification method can greatly improve the robustness and precision of WBBNN.

Scene classification is one of the most significant and challenging tasks in computer vision. This paper presents a new method for scene classification using bag of visual words and a particle swarm optimization (PSO)-based artificial neural network classifier. Contributions of this paper are introducing a novel feature integration method using scale invariant feature transform (SIFT) and local binary pattern (LBP) and a new framework for training radial basis function neural network, combining optimum steepest decent method with a specially designed PSO-based optimizer for center adjustment of radial basis function neural network. Our study shows that using LBP increases the performance of classification task compared to using SIFT only. In addition, our experiments on Proben1 dataset demonstrate improvements in classification performance (averagely about 6.04%) and convergence speed of the proposed radial basis function neural network. The proposed radial basis function neural network is then employed in scene classification task. Results are reported for classification of the Oliva and Torralba, Fei–Fei and Perona and Lazebnik et al. datasets. We compare the performance of the proposed classifier with a multi-way SVM classifier. Experimental results show the superiority of the proposed classifier over the state-of-the-art on the three datasets.

The objective of this work is to evaluate the efficiency of genomic selection (GS) of genome-enabled prediction by Radial Basis Function Neural Networks (RBFNN) in the prediction of genetic values considering dominance effects and diferent degrees of heritability. In addition, the results were compared with those obtained by G-BLUP. An F1 population with 500 individuals genotyped with 1000 SNP-type markers was simulated. The phenotypic traits were determined by adopting two different gene action models: additive and dominance admitting heritability levels (h2) 30 and 60%, each is controlled by 50 loci, considering two alleles per loco. The accuracy and the mean squared error root (MSER) were estimated using a five-fold cross-validation scheme. For the low heritability scenario, h2 = 0.3 in the additive scenario, the accuracy of validation was 31% for RBFNN, 58% for RR-BLUP, and in the complete dominance scenario the values were 28% e 25%, respectively. Additionally, when analyzing the MSER the difference in performance of the techniques is even greater. For additive scenario, the estimates were 97.33 RR-BLUP and 5.80 for RBFNN, in the most critical scenario, 91.31 GBLUP and 14.55 for RBFNN. Overall, a RBFNN shows accuracy lower than those obtained through G-BLUP. On the other hand, the RBFNN has low prediction bias. Finally, the adjustment of the GBLUP to dominance models despite increasing the complexity of the model also increased the predictive accuracy compared to the model without considering the dominance effect.

The modelling of a switched reluctance motor (SRM) is indispensable owing to its inherent non-linear characteristics. In this study, a general flux linkage model for SRMs developed by normalising the magnetic saturation levels and magnetic structures are described. On the basis of this general model, entire flux linkage characteristics of a specific SRM can be calculated easily with only need for the flux linkage at the aligned and unaligned rotor positions. In addition, a radial basis function neural network is adopted for high-precision modelling of the general flux linkage model. The flux linkage characteristics of five prototype motors were also analysed using both the finite element method and the proposed general modelling strategy. The results demonstrate perfect agreement, which verify the feasibility and practicality of the proposed general modelling method.

In the intelligent fault diagnosis of rolling bearings, the high recognition accuracy is hardly achieved when small training samples and strong noise happen. In this article, a novel fault diagnosis method is proposed, that is radial basis function neural network with power spectrum of Welch method. This fault diagnosis model adopts the way of end-to-end operating mode. It takes the original vibration signal (time-domain signal) as input, and Welch method transforms the data from time-domain signals to power spectrums and suppresses high strength noise. Then the results of Welch method are classified by radial basis function neural network. To test the performance of radial basis function neural network with power spectrum of Welch method, the method is compared with some advanced fault diagnosis methods, and the limit performance test for radial basis function neural network with power spectrum of Welch method is carried out to obtain its ultimate diagnosis ability. The results show that the proposed method can realize the high diagnostic precision without the complex feature extraction from the signal. At the same time, in the case of a small amount of training data, this method also can achieve the diagnosis in high precision. Moreover, the anti-noise performance of radial basis function neural network with power spectrum of Welch method is better than the performance of some fault diagnosis methods proposed in recent years.

The dynamic characteristics of the traction gear transmission system have a great influence on the safety, comfort, and reliability of EMU (electric multiple units). Combining the methods of theoretical analysis, numerical simulation, and optimization design theory, establishing a parameterized gear modification model. Meanwhile, designing reasonable shape modification schemes and parameters. The dynamic characteristics, vibration response characteristics, and acoustic response characteristics of gear meshing of CRH380A high-speed EMU under continuous traction conditions are analyzed. The corresponding relationship between gear modification parameters and gear transmission radiation noise is approximated by finite element simulation data and RBF (radial basis function) neural network. Using a multi-island genetic algorithm to optimize gear modification parameters to minimize gear transmission noise, further seeking to meet the low-noise modification design of high-speed train traction helical gear transmission system under continuous operating conditions method.

The overall objective of this study is to design the nonlinear U-model-based radial basis function neural network model reference adaptive control system, through research into a class of complex time-varying nonlinear plants. First, the ideal nonlinear plant is adopted as the reference model and transformed into the U-model representation. In the process, the authors establish the corresponding relationship between the degrees of the reference nonlinear model and the controlled nonlinear plants, and carry out research into the corresponding coefficient relationship between the reference nonlinear model and the controlled nonlinear plants. Also, the impact of the adjusting amplitude and tracking speed of the model on the system control accuracy is analyzed. Then, according to the learning error index of the neural network, the paper designs the adaptive algorithm of the radial basis function neural network, and trains the network by the error variety. With the weight coefficients and network parameters automatically updated and the adaptive controller adjusted, the output of controlled nonlinear plants can track the ideal output completely. The simulation results show that the model reference adaptive control system based on RBF neural network has better control effect than the nonlinear U-model adaptive control system based on the gradient descent method.

Machine learning models are regarded as efficient and popular models for natural disaster susceptibility prediction. However, few studies have focussed on the applications of the latest popular machine learning models in collapse susceptibility assessment (CSA). This paper proposes a 3S (RS, GPS and GIS) technology-based support vector machine (SVM) to map collapse susceptibility in the Nantian area of China. The 3S technology-based back-propagation neural network (BPNN) and radial basis function neural network (RBFNN) models are also proposed for comparison. First, 44 recorded collapses are identified through field investigation, and fourteen collapse-related causal factors are acquired using 3S technology. Second, among these recorded collapses and randomly selected ‘non-collapses’, 70% of the collapse and non-collapse grid cells are used to train the three models, while the remaining 30% are used to test the models. Third, the collapse susceptibility maps of the Nantian area are produced using the three models. Finally, the prediction accuracies of these models are evaluated. The results indicate that the SVM model has the highest prediction accuracy, while the RBFNN model has the lowest prediction accuracy for CSA. In addition, the distribution characteristics of collapse susceptibility in the Nantian area are produced well by all three models.

The connectivity restoration has significance for Industrial IoTs (IIoTs). If the connectivity is compromised, mobile data collectors can be deployed to restore the connectivity. The aggregation ratio, which is the proportion of data successfully delivered to the sink over all data, is considered as a crucial index. However, previous works only consider the travel distance, the load balance, the latency and the energy cost over the aggregation ratio. In this paper, a machine learning based connectivity restoration strategy CRrbf, that utilizes a Radial Basis Function Neural Network (RBFNN) along with an Unscented Kalman Filter (UKF), is proposed to maximize the aggregation ratio meanwhile reduce the energy cost. The theoretical analysis and simulation results indicate that CRrbf outperforms both distance based strategies and terrain based strategies in the aggregation ratio, the network latency and the network throughput. And the energy cost of CRrbf is less than that of distance based strategies.

In this paper, a novel control scheme with respect to the adaptive decoupling controller based on radial basis function neural network (ADEC-RBFNN) is developed. On one hand, in order to improve the system performance of the torque closed-loop control system (TCLCS) of the permanent magnet synchronous motor (PMSM) with the effects of the dynamic coupling and back electromotive force (EMF), we present a novel ADEC with which the TCLCS is asymptotically stable under Lyapunov stability theory. On the other hand, considering the uncertainty and time variant of both the PMSM and ADEC parameters, the RBFNN is utilized to optimize the ADEC parameters to achieve optimal system performance. Ultimately, experimental results demonstrate that the torque and current with the proposed control scheme have the good performance of small fluctuation and fast response in the whole ranges of the speed and torque, that is to say, the system with the proposed control scheme is with the good decoupling performance.

A robot should be endowed with certain collaboration experience to recognize human’s behavioral intention. This paper provides a method based on machine learning to recognize the collaborator’s intention. A radial basis function neural network model was built for offline practice of a robot to recognize intention. Some collaboration skills can be obtained by the robot by building a map between the collaborator’s intention and the system state, deducing human’s intention based on the dynamic characteristics of collaborator and robot and taking the collaborator’s intention as the feedforward information for controlling the robot so as to estimate the human’s intention online based on collaborator’s force and robot’s motion characteristics during collaboration. The proposed method can overcome the difficulties in building the human-robot collaboration model by traditional method, especially the complicated human motion model, and difficulties in estimation of impedance parameters of human body. An experiment was conducted on a motion platform with single degree of freedom. The results prove that the collaborator’s force is reduced while synchronization of human-robot collaboration is improved, so that the compliance of collaborated motion is also improved.

Study region Metropolitan Taipei city, Taiwan. Study focus The severity of climate change in Taiwan is increasing, and the average temperature in Taiwan has increased by 1.0–1.4 °C over the last century, which is higher than the global average. A state-of-the-art hybrid of Multi-dimensional Complementary Ensemble Empirical Mode Decomposition (MCEEMD) and Radial Basis Function Neural Network (RBFNN) is developed for the analysis, simulation, and forecasting of extreme temperature events. The MCEEMD is first introduced to process non-stationary, nonlinear gridded climate data from 1960 to 2017. The RBFNN model is then coupled with the MCEEMD algorithm to forecast the 7-day daily maximum temperatures in Taipei. New hydrological insights for the region Long-term time series data such as temperature exhibit non-stationarity, and non-linearity, thus making it more challenging to assessing the future trend and variability. It is shown that for daily maximum temperatures in Taipei, changes in extremes are more pronounced and impactful than changes in the mean values in response to global warming. The newly developed hybrid MCEEMD-RBFNN model is capable of analyzing the nonstationary and nonlinear daily maximum temperature data, and then forecasting the daily maximum temperature over the next 7 days. Short-term trends and variability of daily maximum temperature in Taipei can thus be identified and diagnosed for hazard prevention and risk assessment of heatwave events.