Radial Basis Function Neural Network Research Articles

A fixed-time nonsingular terminal sliding mode control based on radial basis function neural network for vehicle active front steering system

In our work, a novel fixed-time nonsingular terminal sliding mode controller is developed to improve the control performance for the active front steering system of electric vehicles. First, the sliding mode surface and the sliding mode convergence law are constructed so that the sliding variable can be stabilized to zero at a fixed time. Moreover, the radial basis function neural network is introduced to estimate the systematic uncertainties, which can make the system have stronger anti-disturbance ability. The Lyapunov analysis also is given to verify the fixed-time stability of the whole system. The presented control scheme possesses two appealing advantages, including the strong robustness to perturbations and the fixed-time convergence to zero. Notably, the settling time is independent of the initial value while only related to the controller parameters. Through the joint simulation of CarSim and MATLAB, the given controller is compared with the conventional sliding mode controller and nonsingular terminal sliding mode controller to demonstrate the plausibility and validity of the presented control strategy.

Trajectory tracking of binocular vision system for picking robot based on fast non-singular terminal sliding mode control

This paper proposes a non-singular fast terminal sliding mode control method for a binocular active vision platform of a picking robot with unknown dynamics. The method uses radial basis function (RBF) neural networks to achieve trajectory tracking accuracy and enhance robustness against external interference. A non-singular fast terminal sliding mode controller is designed for the system’s convergence within a limited time. An adaptive neural network approximates the unknown nonlinear function of the dynamic model. Stability and finite-time convergence of the closed-loop system are established using Lyapunov theory. Experimental verification on the binocular vision platform demonstrates position and speed errors converging to the desired trajectory within 2 and 1 second, respectively. Moreover, when subjected to external interference, the position and velocity errors converge within 0.1 seconds. Simulation experiments confirm the method’s effectiveness in improving convergence speed, trajectory tracking accuracy, and robustness against external interference, while reducing system chattering.

Dynamic modeling and optimization of an eight bar stamping mechanism based on RBF neural network PID control

Introduction: Modern industrial manufacturing often requires the eight-bar stamping mechanism to have high motion accuracy and stability. To meet these stringent requirements, traditional control techniques such as proportional-integral-derivative (PID) control need to be improved.Methods: In this study, radial basis function neural network is introduced to improve the traditional proportional integral derivative control technique. The improved proportional integral derivative technique is applied to the modeling and optimization of eight kinds of bar stamping mechanisms.Results: Comparing the improved control technology, the experiment showed that the peak time and adjustment time of the improved technology were 0.516 s and 1.038 s, respectively, which are better than the comparative control technology. In addition, in the comparative analysis of the eight bar stamping mechanism, the proposed architecture scored 9.3 points in operational efficiency, which is significantly greater than the comparative architecture.Discussion: The results show that the combination of PID control strategy and radial basis function neural network provides a powerful tool for dynamic modeling and optimization of eight-bar stamping mechanism. It not only provides enhanced motion accuracy and stability, but also brings significant practicality to industrial manufacturing. This integration opens up new possibilities for improving the performance of complex mechanical systems to meet the evolving needs of modern manufacturing.

Near-Optimal Tracking Control of Partially Unknown Discrete-Time Nonlinear Systems Based on Radial Basis Function Neural Network

This paper proposes an optimal tracking control scheme through adaptive dynamic programming (ADP) for a class of partially unknown discrete-time (DT) nonlinear systems based on a radial basis function neural network (RBF-NN). In order to acquire the unknown system dynamics, we use two RBF-NNs; the first one is used to construct the identifier, and the other one is used to directly approximate the steady-state control input, where a novel adaptive law is proposed to update neural network weights. The optimal feedback control and the cost function are derived via feedforward neural network approximation, and a means of regulating the tracking error is proposed. The critic network and the actor network were trained online to obtain the solution of the associated Hamilton–Jacobi–Bellman (HJB) equation within the ADP framework. Simulations were carried out to verify the effectiveness of the optimal tracking control technique using the neural networks.

Robust control with protected feedback information for switched systems under injection attacks

In this paper, the robust security protection control for switched systems is studied. A robust anti-disturbance mechanism using the Radial Basis Function Neural Network (RBFNN) is proposed for switched systems, which can approximate and compensate for the impact of unknown disturbance on the system state. Then, a network security protection mechanism based on encoder and decoder is presented, which has the ability to resist the dual impact on the feedback information caused by the network privacy snooping and data injection attacks. Accordingly, stability analysis and state-feedback controller design are given for the switched systems under unknown disturbance, network privacy snooping and injection attacks. Finally, simulation results illustrate the effectiveness of the proposed method.

Adaptive neural network fault‐tolerant control for uncertain non‐strict feedback nonlinear system with actuator faults and state constraints

SummaryThis article studies a neural network (NN)‐based adaptive fault‐tolerant control (FTC) scheme for uncertain non‐strict feedback systems with time‐varying state constraints and actuator faults. The introduction of asymmetric Barrier‐Lyapunov function (BLF) makes controller design more difficult due to the emergence of actuator faults and state constraints. Therefore, this article designs a fault‐tolerant controller with constraint compensation information under the backstepping control design framework to solve the state constraint asymmetry problem caused by actuator failure. By designing an improved asymmetric time‐varying BLF, the design of the state‐constrained controller will become more realistic and the constraints will be weakened. In the design process, the characteristics of the radial basis function neural network are used to avoid the algebraic ring problem. Actuator failure in this article considers deviation failure and loss of effectiveness. Based on the properties of the exponential function, the improved BLF can make the bounds of the state constraints smaller and smaller, and the bounds of the constraints can change with the desired trajectory. Simulation verified the feasibility of this control method.

Adaptive Prosthetic Trajectory Estimation Based on Key Points Constraints

Lower limb knee–ankle prostheses can effectively assist above-knee amputees in completing their basic daily activities. This study explored methods for estimating the joint kinematics of intelligent lower limb prostheses to better adapt them to the walking requirements of amputees. A method of estimating the knee and ankle joint trajectories under different speeds and slopes was raised. The joints of a prosthetic need to have a movement trajectory similar to that of the joints of healthy individuals, taking into account the person’s motion intentions and complying with the law of movement. In this study, a prosthetic kinematics estimation method was studied to realize continuous speed and slope adaptation through key points. The iterative Douglas–Peucker algorithm automatically found the key points of the kinematic trajectory curve, and an invariant basis function-fitting model with key points constraints was constructed. Finally, the radial basis function neural network was used to estimate the adaptive task function of the parts affected by the speed and slope, resulting in an overall kinematics estimation model. Our experiments verified the effectiveness of the proposed method, which accurately estimated the kinematic trajectory; the knee and ankle joint estimation errors were greatly reduced compared to those obtained with previous methods, facilitating further research on individual algorithms.

Prognostication of Weather Patterns using Meteorological Data and ML Techniques

In the field of modern weather prediction, the accurate classification is essential, impacting critical sectors such as agriculture, aviation, and water resource management. This research presents a weather forecasting model employing two influential classifiers random forest and technique based on gradient boosting, both implemented using the Scikit-learn library. Evaluation is based on key metrics including F1 score, accuracy, recall, and precision, with Gradient Boosting emerging as the superior choice for precipitation prediction. The study examines the performance of Random Forest Regression, Gradient Boosting Regression, and Radial Basis Function Neural Network in forecasting precipitation, drawing on prior research that demonstrated the superiority of the Random Forest algorithm in terms of accuracy and speed. Ensemble methods, particularly the Voting Classifier, a fusion of Random Forest and Gradient Boosting, outperform individual models, offering a promising avenue for advancing weather classification.

New insights into estimating the cementation exponent of the tight and deep carbonate pore systems via rigorous numerical strategies

AbstractOne of the main constituents of any reservoir characterization is an accurate forecast of water saturation, which is highly dependent upon the cementation exponent. Even though there have been a lot of studies, the most common correlations depend on total porosity. This means that they do not work as well in heterogeneous carbonate reservoirs, especially tight formations with total porosities less than 10%. This study aims to develop accurate and universal models for estimating the cementation exponent in deep and tight carbonate pore systems located in West Asia. Two heuristic algorithms, including the radial basis function neural network optimized by ant colony optimization (RBFNN-ACO) and gene expression programming (GEP), were employed to calculate the cementation exponent. To do this, we prepared a databank incorporating cementation exponents, total porosity, and various pore types. Then, the databank is classified into the test subset (for model prediction checking) and the train subset (for model construction). The reliability of the new recommended models is inspected by applying several statistical quality measures associated with graphical analyses. So, the consequences of the modeling disclose the large precision of the above-mentioned RBFNN-ACO, GEP Model-I, and GEP Model-II by average absolute percentage relative deviations (AAPRD%) of 6.28%, 6.39%, and 7.45%, respectively. Based on the outliers analysis, nearly 95% of the databank and model estimations are, respectively, valid and reliable. Additionally, the three input variables, including moldic porosity (with a + 70% impact value), non-fabric-selective dissolution (connected) porosity (with a -30% impact value), and interparticle porosity (with a -23% impact value), exhibit the main affecting parameters on the cementation exponent. Comparing current results with traditional literature correlations demonstrates the supremacy of the RBFNN-ACO model (AAPRD = 6.28% and root mean squared error (RMSE) = 0.17) over the examined literature correlations such as Borai’s equation (AAPRD = 12.30% and RMSE = 0.41). In addition, RBFNN-ACO can give better results than Borai’s Eqn. for tight (porosity less than 10%) and deep carbonate samples.

Hybrid hyperplane gradient learning algorithm for RBF neural network

Gradient-based algorithms are still popularly used for training radial basis function neural network (RBFNN). However, these algorithms may lead to the vanishing gradient (saddle point or local minimum) problem, which is one of the common problems that limit learning performance. To cope with this, a hybrid hyperplane gradient learning algorithm (HHGLA) is proposed to improve the learning performance of the RBFNN in this paper. First, a hyperplane gradient (HPG), based on a hyperspace constructed by a solution population (SP), is introduced. Thus, the search process can cross the surface of the cost function to avoid the appearance of vanishing gradient. Second, an adaptive learning rate is designed to restrict the search process into this hyperspace. Then, during the learning process, the hyperspace is constantly contracted to approximate the global optimal solution. Third, the convergence property of HHGLA-based RBFNN is thoroughly analyzed to ensure its application success. Finally, empirical comparisons of the proposed HHGLA-RBFNN are given to illustrate its superiority. The simulation results show the feasibility of HHGLA-RBFNN of accuracy.

Design and Control Performance Optimization of Variable Structure Hydrostatic Drive Systems for Wheel Loaders

The traditional loader drive system is based on the hydraulic torque converter as the key component, and its gear is shifted through the coordination of the clutch and gearbox, which greatly increases power loss and operator fatigue. To address the above problem, a variable-structure hydrostatic drive system is proposed. This system adopts a closed-loop design and adjusts the displacement of the pump and dual motors to follow the throttle opening of the vehicle, achieving automatic gear shifting and smooth speed regulation. It can also automatically change the system structure according to the vehicle’s speed, meeting the vehicle’s demand for rapid switching output of high torque and high speed. At the same time, in the displacement matching control process, an adaptive sliding mode control scheme based on radial basis function neural network compensation is proposed. This scheme designs an adaptive hyperparameter update strategy according to the characteristics of the system to effectively compensate for changes in uncertain factors. Experimental results show that, compared to traditional drive systems, this system has the characteristics of simple operation, smooth speed regulation, and high fuel economy.

Event-triggered adaptive control for quadrotor unmanned aerial vehicle with prescribed performance

This paper proposes an event-triggered adaptive control algorithm with prescribed performance for quadrotor unmanned aerial vehicles (UAVs). The dynamic uncertainty and unknown external interference of the system are compensated by the property of approximation in a radial basis function (RBF) neural network. Based on this, introduce a performance constraint function to ensure that the trajectory tracking error of quadrotor UAV converges to a preset area, thereby guaranteeing transient and steady-state performance. In addition, design an event-triggered adaptive controller with the combination of the event-triggered mechanism, reducing the controller’s update frequency. Simulation results demonstrate that this algorithm exhibits strong anti-interference properties and effectively resolves the flight control problem of quadrotor UAV.

Response of Gaussian color noise excited oscillators with inertia nonlinearity based on the Radial Basis Function Neural Network method

Given the widespread utilization of oscillators featuring inertial nonlinear terms across engineering disciplines and the omnipresence of noise excitations, comprehending their response to noise excitation holds paramount importance. In contrast to Gaussian white noise, colored noise stands out as a more adept mathematical model for simulating ambient noise scenarios. The widely used stochastic averaging method is somewhat ineffective in dealing with the response of oscillators with inertial nonlinearity under Gaussian colored noise excitation, as it cannot display the saddle shaped characteristics of the steady-state probability density function (PDF) caused by inertial nonlinearity. To overcome the shortcomings, this paper adopts a semi-analytical method called a radial basis function neural network method (RBFNN) to address this problem. This method represents the solution of the Fokker-Plank-Kolmogorov (FPK) equation of the system as the sum of a series of weighted radial basis functions, and obtains the optimal weight coefficients through the Lagrange multiplier method. This article investigates two examples, one of which has inertial nonlinearity and the other has multiple potential wells. The effects of inertial nonlinearity coefficients and the delay coefficients of colored noise on response are explored. The mean square errors between Monte Carlo (MC) and RBF are presented, which indicates the RBF results perfectly agree with the MC predictions.

Distributed Adaptive Forwarding Finite-Time Output Consensus of High-Order Multiagent Systems via Immersion and Invariance-Based Approximator.

A finite-time output consensus control problem is investigated in this article for an uncertain nonlinear high-order multiagent systems (MASs). For this class of MASs, the order of individual follower is reduced gradually by implementing the immersion and invariance (I&I) control theory repeatedly, and a requirement of solving partial differential equations (PDEs) in I&I control theory is obviated. Furthermore, an I&I-based radial basis function neural network (RBFNN) approximator is developed, where an extra cross term is added in the approximation mechanism, and the form of an update law for weights is transformed into a proportional and integral one. This I&I-based RBFNN approximator does not rely on a cancellation of the perturbation term, and these uncertainties are reconstructed by the I&I manifold adaptively, which is for improvement of approximation behaviors of traditional RBFNNs. On this basis, a distributed adaptive forwarding finite-time output consensus control strategy is proposed by combining a sign function, and the convergence time of the MAS can be adjusted with appropriate finite-time parameters. Finally, two illustrative examples verify the effectiveness of the theoretical claims.

A novel additive regression model for streamflow forecasting in German rivers

Forecasting streamflows, essential for flood mitigation and the efficient management of water resources for drinking, agriculture and hydroelectric power generation, presents a formidable challenge in most real-world scenarios. In this study, two models, the first based on the Additive Regression of Radial Basis Function Neural Networks (AR-RBF) and the second based on the stacking with the Pace Regression of the Multilayer Perceptron and Random Forest (MLP-RF-PR), were compared for the prediction of short-term (1–3 days ahead) and medium-term (7 days ahead) daily streamflow rates of three different rivers in Germany: the Elbe River at Wittenberge, the Leine River at Herrenhausen, and the Saale River at Hof The lagged values of streamflow rate, precipitation and temperature were considered for the modeling. Moreover, the Bayesian Optimization (BO) algorithm was used to assess the optimal number of lagged values and hyperparameters. Both models showed accurate predictions for short-term forecasting, with R2 for 1-day ahead predictions ranging from 0.939 to 0.998 for AR-RBF and from 0.930 to 0.996 for MLP-RF-PR, while MAPE ranged from 2.02 % to 8.99 % for AR-RBF and from 2.14 % to 9.68 % for MLP-RF-PR, when exogeneous variables were included. As the forecast horizon increased, a reduction in forecasting accuracy was observed. However, both models could still predict the overall flow pattern, even for 7-day-ahead predictions, with R2 ranging from 0.772 to 0.871 for AR-RBF and from 0.703 to 0.840 for MLP-RF-PR, while MAPE ranged from 10.60 % to 20.45 % for AR-RBF and from 10.44 % to 19.65 % for MLP-RF-PR. Overall, the outcomes of this study suggest that both AR-RBF and MLP-RF-PR can be reliable tools for the short- and medium-term streamflow rate prediction, requiring a short number of parameters to be optimized, making them easy to implement while reducing the calculation time required.

Observer-Based Neural Control of N-Link Flexible-Joint Robots.

This article concentrates on the adaptive neural control approach of n -link flexible-joint electrically driven robots. The presented control method only needs to know the position and armature current information of the flexible-joint manipulator. An adaptive observer is designed to estimate the velocities of links and motors, and radial basis function neural networks are applied to approximate the unknown nonlinearities. Based on the backstepping technique and the Lyapunov stability theory, the observer-based neural control issue is addressed by relying on uplink-event-triggered states only. It is demonstrated that all signals are semi-globally ultimately uniformly bounded and the tracking errors can converge to a small neighborhood of zero. Finally, simulation results are shown to validate the designed event-triggered control strategy.

Exploration on Human Resource Management and Prediction Model of Data-driven Information Security in Internet of Things

The advent of the Internet of Things (IoT) has accelerated the pace of economic development across all sectors. However, it has also brought significant challenges to traditional human resource management, revealing an increasing number of problems and making it unable to meet the needs of contemporary enterprise management. The IoT has brought numerous conveniences to human society, but it has also led to security issues in communication networks. To ensure the security of these networks, it is necessary to integrate data-driven technologies to address this issue. In response to the current state of human resource management, this paper proposes the application of IoT technology in enterprise human resource management and combines it with radial basis function neural networks to construct a model for predicting enterprise human resource needs. The model was also experimentally analyzed. The results show that under this algorithm, the average prediction accuracy for the number of employees over five years is 90.2%, and the average prediction accuracy for sales revenue is 93.9%. These data indicate that the prediction accuracy of the model under this study's algorithm has significantly improved. This paper also conducted evaluation experiments on a wireless communication network security risk prediction model. The average prediction accuracy of four tests is 91.21%, indicating that the model has high prediction accuracy. By introducing data-driven technology and IoT applications, this study provides new solutions for human resource management and communication network security, promoting technological innovation in the fields of traditional human resource management and information security management. The research not only improves the accuracy of the prediction models but also provides strong support for decision-making and risk management in related fields, demonstrating the great potential of big data and artificial intelligence technology in the future of enterprise management and security.

A new paradigm for smart packaging: A dual-channel freshness monitoring platform based on aerogels of sodium alginate-anthocyanin complex with high colorimetric sensitivity and stability

Global seafood consumption is estimated at 156 million tons annually, with an economic loss of >25 billion euros annually due to marine fish spoilage. In contrast to traditional smart packaging which can only roughly estimate food freshness, an intelligent platform integrating machine learning and smart aerogel can accurately predict remaining shelf life in food products, reducing economic losses and food waste. In this study, we prepared aerogels based on anthocyanin complexes that exhibited excellent environmental responsiveness, high porosity, high color-rendering properties, high biocompatibility, high stability, and irreversibility. The aerogel showed excellent indication properties for rainbow trout and proved suitable for fish storage environments. Among the four machine learning models, the radial basis function neural network and backpropagation network optimized by genetic algorithm demonstrated excellent monitoring performance. Also, the two-channel dataset provided more comprehensive information and superior descriptive capability. The three-layer structure of the monitoring platform provided a new paradigm for intelligent and sophisticated food packaging. The results of the study might be of great significance to the food industry and sustainable development.

Error Loss Networks.

A novel model called error loss network (ELN) is proposed to build an error loss function for supervised learning. The ELN is similar in structure to a radial basis function (RBF) neural network, but its input is an error sample and output is a loss corresponding to that error sample. That means the nonlinear input-output mapper of the ELN creates an error loss function. The proposed ELN provides a unified model for a large class of error loss functions, which includes some information-theoretic learning (ITL) loss functions as special cases. The activation function, weight parameters, and network size of the ELN can be predetermined or learned from the error samples. On this basis, we propose a new machine learning paradigm where the learning process is divided into two stages: first, learning a loss function using an ELN; second, using the learned loss function to continue to perform the learning. Experimental results are presented to demonstrate the desirable performance of the new method.

Event-triggered simultaneous tracking control and stabilization of underactuated USVs

This study addresses the control problem of simultaneous tracking control and stabilization for underactuated unmanned surface vessels (USVs) with only two available inputs in the surge and yaw directions. Firstly, two velocity modifications are introduced to transform the error dynamics into two decoupled subsystems. Then, a novel error transforming scheme is presented to reduce the complexity of controller expressions and overcome the difficulties caused by underactuated dynamics. Based on the above transforming scheme and the event-triggered control (ETC) mechanism, a unified control method is designed to guarantee that the error states converge to a bounded neighborhood of the origin while excluding the Zeno behavior of the controllers. In addition, two radial basis function neural networks (RBFNNs) are applied to approximate the unknown dynamics and additional control errors caused by the ETC mechanism. Finally, comparative simulation experiments are performed to validate the effectiveness of the proposed controller.