Adaptive Radial Basis Function Network Research Articles

Novel General Regression Neural Networks for Improving Control Accuracy of Nonlinear MIMO Discrete-Time Systems.

In this article, a novel version of the general regression neural network (Imp_GRNN) is developed to control a class of multiinput and multioutput (MIMO) nonlinear discrete-time (DT) systems. The improvements retain the features of the original GRNN along with a significant improvement of the control accuracy. The enhancements include developing a method to set the input-hidden weights of GRNN using the inputs recursive statistical means, introducing a new output layer and adaptable forward weighted connections from the inputs to the new layer, and suggesting an interval-type smoothing parameter to eradicate the need for selecting the parameter beforehand or adapting it online. Also, controller stability is studied using Lyapunov's method for DT systems. The controller performance is tested with different simulation examples and compared with the original GRNN to verify its superiority over it. Also, Imp_GRNN performance is compared with an adaptive radial basis function network controller, an adaptive feedforward neural-network (NN) controller, and a proportional-integral-derivative (PID) controller, where it demonstrated higher accuracy in comparison with them. In comparison with the formerly proposed control methods for MIMO DT systems, our controller is capable of producing high control accuracy while it is model free, does not require complex mathematics, has low computational complexity, and can be utilized for a wide range of DT dynamic systems. Also, it is one of the few methods that aims to improve the control system accuracy by improving the NN structure.

Learning from output‐feedback control of sampled‐data systems in normal form

AbstractThis paper investigates the learning and control problem of sampled‐data systems with only output measurements. A unified approach is presented by integrating the sampled‐data observer and deterministic learning. First, an adaptive radial basis function network (RBFN) learning controller with a sampled‐data observer is designed to track a recurrent reference model. Along the trajectory estimated by the observer, it is proven that the RBFN weights can exponentially converge to their ideal values with the satisfaction of a persistent excitation (PE) condition and the closed‐loop dynamics can be accurately learned during the output‐feedback process. Second, by using the learning results, a knowledge‐based output‐feedback controller is developed to improve the tracking performance. Further research shows that choosing appropriate parameters for the observer and RBFN can guarantee learning and control performance. The significance of the proposed approach is that the closed‐loop dynamics of the output‐feedback process can be accurately learned and further utilized to improve control performance. Simulation studies indicate the effectiveness and advantages of the learning control approach.

Manipulator Control Based on Adaptive RBF Network Approximation

With the popularization of intelligent manufacturing, manipulator has found ever wider application in various industries. A manipulator requires a real-time and fast control algorithm in order to improve the accuracy in all kinds of precise operations. This paper proposes an algorithm based on adaptive radial basis function (RBF) for approximating the parameters of the manipulator, and the adaptive equations are designed to automatically adjust the weight of RBF. Proportional integral (PI) robust based on dynamic error tracking is used in controller to reduce the steady state errors and enhance the anti-interference performance of the system. The global asymptotic stability of the system is demonstrated by defining an integraltype Lyapunov function. Finally, MATLAB is used to simulate the angular positions tracking and angular velocities tracking of the double joints manipulator. The results show that the manipulator can track the ideal output signal quickly and accurately and has good anti-interference performance.

Disturbance Observer based Dynamic Surface Sliding Mode Control for Manipulator Arm with Adaptive RBF Networks

For the rotation position control of automatic loading manipulator arm with uncertain problems of load change and external disturbances, an adaptive radial basis function networks (RBF) dynamic surface sliding mode control (DSMC) strategy based on disturbance observer (DOB) is proposed. The value of the external disturbance, which is estimated by the DOB on line, is added to the output of the proposed controller for feedforward compensation. The dynamic performance of the controller is ensured by introducing the adaptive RBF networks while the stability of the controller is verified by designing the Lyapunov function. Comparison between the results of the proposed control strategy and those of the conventional sliding mode control (SMC) shows that the controller has better dynamic tracking characteristics, steady-state accuracy and strong robustness under load changes.

Intelligent Control with Artificial Neural Networks for Automated Insulin Delivery Systems.

Type 1 diabetes mellitus is a disease that affects millions of people around the world. Recent progress in embedded devices has allowed the development of artificial pancreas that can pump insulin subcutaneously to automatically regulate blood glucose levels in diabetic patients. In this work, a Lyapunov-based intelligent controller using artificial neural networks is proposed for application in automated insulin delivery systems. The adoption of an adaptive radial basis function network within the control scheme allows regulation of blood glucose levels without the need for a dynamic model of the system. The proposed model-free approach does not require the patient to inform when they are going to have a meal and is able to deal with inter- and intrapatient variability. To ensure safe operating conditions, the stability of the control law is rigorously addressed through a Lyapunov-like analysis. In silico analysis using virtual patients are provided to demonstrate the effectiveness of the proposed control scheme, showing its ability to maintain normoglycemia in patients with type 1 diabetes mellitus. Three different scenarios were considered: one long- and two short-term simulation studies. In the short-term analyses, 20 virtual patients were simulated for a period of 7 days, with and without prior basal therapy, while in the long-term simulation, 1 virtual patient was assessed over 63 days. The results show that the proposed approach was able to guarantee a time in the range above 95% for the target glycemia in all scenarios studied, which is in fact well above the desirable 70%. Even in the long-term analysis, the intelligent control scheme was able to keep blood glucose metrics within clinical care standards: mean blood glucose of 119.59 mg/dL with standard deviation of 32.02 mg/dL and coefficient of variation of 26.78%, all below the respective reference values.

Human-Gait-Based Tracking Control for Lower Limb Exoskeleton Robot

Research shows that it is practical for the normal human movement mechanism to assist the patients with stroke in robot-assisted gait rehabilitation. In passive training, the effect of rehabilitation training for patients can be improved by imitating normal human walking. To make the lower limb exoskeleton robot (LLER) move like a normal human, a tracking control scheme based on human gait data is proposed in this paper. The real human gait data is obtained from healthy subjects using a three-dimensional motion capture platform (3DMCP). Furthermore, the normal human motion characteristics are adopted to enhance the scientificity and effectiveness of assistant rehabilitation training using LLER. An adaptive radial basis function network (ARBFN) controller based on feed-forward control is presented to improve the trajectory tracking accuracy and tracking performance of the control system, where the ARBFN controller is deployed to predict the uncertain model parameters. The feed-forward controller based on the tracking errors is used to compensate for the input torque of LLER. The effectiveness of the presented control scheme is confirmed by simulation results based on experimental data.

Denoising using Self Adaptive Radial Basis Function

This paper presents an adaptive form of the Radial basis function neural network to correct the noisy image in a unified way without estimating the existing noise model in the image. Proposed method needs a single noisy image to train the adaptive radial basis function network to learn the correction of the noisy image. The gaussian kernel function is applied to reconstruct the local disturbance appeared because of the noise. The proposed adaptiveness in the radial basis function network is compared with the fixed form of spreadness and the center value of kernel function. The proposed solution can correct the image suffered from different varieties of noises like speckle noise, Gaussian noise, salt & pepper noise separately or combination of noises. Various standard test images are considered for test purpose with different levels of noise density and performance of proposed algorithm is compared with adaptive wiener filter.

Denoising using Self Adaptive Radial Basis Function

This paper presents an adaptive form of the Radial basis function neural network to correct the noisy image in a unified way without estimating the existing noise model in the image. Proposed method needs a single noisy image to train the adaptive radial basis function network to learn the correction of the noisy image. The gaussian kernel function is applied to reconstruct the local disturbance appeared because of the noise. The proposed adaptiveness in the radial basis function network is compared with the fixed form of spreadness and the center value of kernel function. The proposed solution can correct the image suffered from different varieties of noises like speckle noise, Gaussian noise, salt & pepper noise separately or combination of noises. Various standard test images are considered for test purpose with different levels of noise density and performance of proposed algorithm is compared with adaptive wiener filter.

Fault Tolerant Control of Electronic Throttles with Friction Changes

To enhance the reliability of the electronic throttle and consequently the vehicles driven by the internal combustion engines, a fault tolerant control strategy is developed in this paper. The proposed method employs a full-order terminal sliding mode control in conjunction with an adaptive radial basis function network to estimate change rate of the fault. Fault tolerant control to abrupt and incipient changes in the throttle viscous friction torque coefficient and the throttle coulomb friction torque coefficient is achieved. Whilst the throttle position is driven to track the reference signal, the post-fault dynamics are guaranteed to converge to the equilibrium point in finite time, and the control is smooth without chattering. A nonlinear Simulink model of an electronic throttle is developed with real physical parameters and is used for evaluation of the developed method. A significant change of the throttle friction torque is simulated, and the fault tolerant control system keeps system stability and tracking the reference signal in the presence of the fault.

Unified Noise Reduction using Adaptive Radial Basis Function

The images captured by SAR and sonar are blurred and corrupted more by speckle noise and also other types of noise like Gaussian noise and salt & pepper noise. Denoising all types of noises to get perfect image is a vital challenge, earlier works on the same mode addressed with one filter for one noise, there is no one common or unified filter which can denoise all types of noise. Therefore in this paper, we have designed a filter which not only removes speckle noise, but also combination of other noises. Here IUNR (Intelligent Unified Noise Reduction) algorithm is proposed which is based on neural network called adaptive radial basis function acts as a unified filter for Denoising. Proposed method needs a single noisy image to train the adaptive radial basis function neural network to learn the correction of the noisy image. The Gaussian kernel function is applied to reconstruct the local disturbance appeared because of the noise. The proposed adaptive radial basis function network is compared with the fixed form which has fixed spread and the center value of kernel function. This method can correct the image suffered from different varieties of noises like speckle noise, salt & pepper noise and Gaussian noise separately or combination of noise. Various standard test images are considered for test purpose with different levels of noise density and performance of proposed algorithm is compared with adaptive wiener filter.

Adaptive neural network sliding mode control of shipboard container cranes considering actuator backlash

Abstract Offshore container crane is a highly under-actuated nonlinear system whereas only two control inputs are employed for driving six system outputs. Controlling such a system is not easy since it faces with many challenges composed of actuator backlash, geometrical nonlinearities, seawater viscoelasticity, cable flexibility, strong wave and wind disturbances, and considerable lack of actuators. This article proposes a robust adaptive system for a ship-mounted container crane with the disadvantages mentioned above. The controller structure is constructed using second-order sliding mode control (SOSMC), and a modeling estimator is designed on the basis of radial basis function network (RBFN). While other adaptive control techniques only estimates system parameters, the adaptive RBFN algorithm approximates almost all the structure of a crane model, including system parameters. Simulations and experiments are conducted to verify the superiority of the proposed control system.

Adaptive RBF Network Based Direct Voltage Control for Interior PMSM Based Vehicles

This paper presents an adaptive radial basis function (RBF) network based speed control approach for interior permanent magnet synchronous machine (IPMSM) based vehicles. The presented control strategy achieves speed tracking without current measurement or control by acting directly on the machine's voltages. As a result, the current control loop is avoided, greatly simplifying the overall control strategy when compared to most existing speed control schemes. Moreover, no parameter knowledge is required unlike other control strategies. Furthermore, the closed-loop control system's stability is guaranteed by a Lyapunov theorem. Simulation and experimental results for different scenarios along with comparison against the well-established field oriented vector control technique demonstrate the controller's satisfactory behavior in both transient and steady state. The proposed method yields 99% and 97% tracking accuracy in the presence of parametric uncertainties and friction nonlinearities, respectively.

A vehicle stability control strategy with adaptive neural network sliding mode theory based on system uncertainty approximation

ABSTRACTModelling uncertainty, parameter variation and unknown external disturbance are the major concerns in the development of an advanced controller for vehicle stability at the limits of handling. Sliding mode control (SMC) method has proved to be robust against parameter variation and unknown external disturbance with satisfactory tracking performance. But modelling uncertainty, such as errors caused in model simplification, is inevitable in model-based controller design, resulting in lowered control quality. The adaptive radial basis function network (ARBFN) can effectively improve the control performance against large system uncertainty by learning to approximate arbitrary nonlinear functions and ensure the global asymptotic stability of the closed-loop system. In this paper, a novel vehicle dynamics stability control strategy is proposed using the adaptive radial basis function network sliding mode control (ARBFN-SMC) to learn system uncertainty and eliminate its adverse effects. This strategy adopts a hierarchical control structure which consists of reference model layer, yaw moment control layer, braking torque allocation layer and executive layer. Co-simulation using MATLAB/Simulink and AMESim is conducted on a verified 15-DOF nonlinear vehicle system model with the integrated-electro-hydraulic brake system (I-EHB) actuator in a Sine With Dwell manoeuvre. The simulation results show that ARBFN-SMC scheme exhibits superior stability and tracking performance in different running conditions compared with SMC scheme.

Development of adaptive p-step RBF network model with recursive orthogonal least squares training

An adaptive p-step prediction model for nonlinear dynamic processes is developed in this paper and implemented with a radial basis function (RBF) network. The model can predict output for multi-step-ahead with no need for the unknown future process output. Therefore, the long-range prediction accuracy is significantly enhanced and consequently is especially useful as the internal model in a model predictive control framework. An improved network structure adaptation is also developed with the recursive orthogonal least squares algorithm. The developed model is online updated to adapt both its structure and parameters, so that a compact model structure and consequently a less computing cost are achieved with the developed adaptation algorithm applied. Two nonlinear dynamic systems are employed to evaluate the long-range prediction performance and minimum model structure and compared with an existing PSC model and a non-adaptive RBF model. The simulation results confirm the effectiveness of the developed model and superior over the existing models.

Reducing the complexity of an adaptive radial basis function network with a histogram algorithm

In this paper, a constructive training technique known as the dynamic decay adjustment (DDA) algorithm is combined with an information density estimation method to develop a new variant of the radial basis function (RBF) network. The RBF network trained with the DDA algorithm (i.e. RBFNDDA) is able to learn information incrementally by creating new hidden units whenever it is necessary. However, RBFNDDA exhibits a greedy insertion behaviour that absorbs both useful and non-useful information during its learning process, therefore increasing its network complexity unnecessarily. As such, we propose to integrate RBFNDDA with a histogram (HIST) algorithm to reduce the network complexity. The HIST algorithm is used to compute distribution of information in the trained RBFNDDA network. Then, hidden nodes with non-useful information are identified and pruned. The effectiveness of the proposed model, namely RBFNDDA-HIST, is evaluated using a number of benchmark data sets. A performance comparison study between RBFNDDA-HIST and other classification methods is conducted. The proposed RBFNDDA-HIST model is also applied to a real-world condition monitoring problem in a power generation plant. The results are analysed and discussed. The outcome indicates that RBFNDDA-HIST not only can reduce the number of hidden nodes significantly without requiring a long training time but also can produce promising accuracy rates.

Post-processing Technique for Improving the Odor-identification Performance based on E-Nose System

In this paper, we proposed a post-processing technique for improving classification performance of electronic nose (E-Nose) system which may be occurred drift signals from sensor array. An adaptive radial basis function network using stochastic gradient (SG) and singular value decomposition (SVD) is applied to process signals from sensor array. Due to drift from sensor's aging and poisoning problems, the final classification results may be showed bias and fluctuations. The predicted classification results with drift are quantized to determine which identification level each class is on. To mitigate sharp fluctuations moving-averaging (MA) technique is applied to quantized identification results. Finally, quantization and some edge correction process are used to decide levels of the fluctuation-smoothed identification results. The proposed technique has been indicated that E-Nose system was shown correct odor identification results even if drift occurred in sensor array. It has been confirmed throughout the experimental works. The enhancements have produced a very robust odor identification capability which can compensate for decision errors induced from drift effects with sensor array in electronic nose system.

A Fast Adaptive Tunable RBF Network For Nonstationary Systems.

This paper describes a novel on-line learning approach for radial basis function (RBF) neural network. Based on an RBF network with individually tunable nodes and a fixed small model size, the weight vector is adjusted using the multi-innovation recursive least square algorithm on-line. When the residual error of the RBF network becomes large despite of the weight adaptation, an insignificant node with little contribution to the overall system is replaced by a new node. Structural parameters of the new node are optimized by proposed fast algorithms in order to significantly improve the modeling performance. The proposed scheme describes a novel, flexible, and fast way for on-line system identification problems. Simulation results show that the proposed approach can significantly outperform existing ones for nonstationary systems in particular.

Minimizing Human-Exoskeleton Interaction Force Using Compensation for Dynamic Uncertainty Error with Adaptive RBF Network

A critical issue in the control of exoskeleton systems is unknown nonlinear dynamic properties of the system. The improper estimation of those unknown properties can cause considerable human-exoskeleton interaction force during human's movements. It is really challenging to exactly estimate the parameters of dynamic models. In this paper, we propose a novel exoskeleton control algorithm to both compensate for the dynamic uncertainty error and minimize the human-exoskeleton interaction force. We have built a virtual torque controller based on dynamic models of a lower exoskeleton and have used an approximation of a Radial Basis Function (RBF) neural network to compensate for the dynamic uncertainty error. By doing so, we avoid using complicated force sensors installed on the human-exoskeleton interface and minimize the physical Human-Robot Interaction (pHRI) force. Moreover, we introduce the prototype of our exoskeleton system, called `PRMI' exoskeleton system. Finally, we validated the proposed algorithm on this system, and the experimental results show that the proposed control algorithm provides a good control quality for the `PRMI' exoskeleton system by compensating for dynamic uncertainty error.

One-step models for soft computing techniques. Industrial application to image processing in quality assurance process

The authors present an approach based on a one-step method using soft computing techniques for a quality assurance process in the form of dimensional checking parameters via industrial image processing. This method offers a high grade of precision in processes to solve the evaluation problems on-line applications. As well known to the researchers, the approaches in a single iteration of these techniques for artificial neural networks (ANN) such as adaptive neuro fuzzy inference systems (ANFIS) and radial basis function network (RBFN) are not documented in the literature. This work provides the simplification to one-step that provides the chance of creation and implementation of these models for online applications without loss of time in the iterations needed to adjust the model (training) to generate a fast response. The main objective of this paper is to provide a model capable of approximating the solution of a function that represents the system, that function is based on historical data of the process but the operators that compound the function are unknown. The relations between the inputs and outputs are known, but the interactions between variables are unknown. Based on the literature, the soft computing techniques are trained by trial and error because they do not have a stop criterion; also, the functions that provide an approximation are unknown in most cases. To solve the problem mentioned above, this paper proposes the one-step method without training. It is necessary to approximate the solution avoiding the overshoot and damping produced by classic approaches. The deviation generated is in the order of one standard deviation whose magnitude is in the order of common approaches for image processing as it is documented in literature for the best case RBFN.

Adaptive trajectory tracking neural network control with robust compensator for robot manipulators

This paper presents an adaptive trajectory tracking neural network control using radial basis function (RBF) for an n-link robot manipulator with robust compensator to achieve the high-precision position tracking. One of the difficulties in designing a suitable control scheme which can achieve accurate trajectory tracking and good control performance is to guarantee the stability and robustness of control system, due to friction forces, external disturbances error, and parameter variations. To deal with this problem, the RBF network is investigated to the joint position control of an n-link robot manipulator. The RBF network is one approach which has shown a great promise in this sort of problems because of its fast learning algorithm and better approximation capabilities. The adaptive RBF network can effectively improve the control performance against large uncertainty of the system. The adaptive turning laws of network parameters are derived using the back-propagation algorithm and the Lyapunov stability theorem, so that the stability of the entire system and the convergence of the weight adaptation are guaranteed. In this control scheme, a robust compensator plays as an auxiliary controller to guarantee the stability and robustness under various environments such as the mass variation, the external disturbances, and modeling uncertainties. Finally, the simulation and experimental results in comparison with adaptive fuzzy and wavelet network control method are provided to verify the effectiveness of the proposed control methodology.