Fuzzy Wavelet Neural Network Controller Research Articles

Variable structure fuzzy wavelet neural network controller for complex nonlinear systems

Controlling a complex nonlinear system has always been an important problem. A neuro fuzzy controller may be an appropriate controller for such systems. There have been always problems with large number of neurons which cause heavy complex computations in neuro fuzzy controllers. In this paper, a variable structure neuro fuzzy controller is introduced to prevent number of neurons from rising. The proposed variable structure controller, adapts itself on-line to the system. Therefore, because of the adaptive structure, there is no need to have a large number of neurons. As a result, calculations and controller complexity will decrease in this structure. The proposed controller also uses an improved gradient descent method to update different varying parameters. In this improved method, adaptive learning rates are used to prevent parameters getting stuck in local optima. Improved gradient descent method also takes the advantages of Lyapunov stability method in updating the parameters. The use of Lyapunov method in updating the network parameters will guarantee the stability of the controller, too. A complex nonlinear time delay system is introduced in this paper to evaluate the proposed variable structure fuzzy wavelet neural network controller. Simulation results illustrate the efficacy of this new controller.

A Lyapunov stable type-2 fuzzy wavelet network controller design for a bilateral teleoperation system

A bilateral teleoperation controller should make the system stable and achieve optimal performance in the presence of time delays, plant disturbances, measurement noise, modeling errors, and unknown environments. This paper proposes a new method of controller design based on a type-2 fuzzy wavelet neural network structure. A new algorithm based on gradient descent is used. The controller has the advantage of fuzzy controllers and wavelet neural network controllers to allow design of a bilateral teleoperation controller. The salient characteristics of the T2FWNN controller are that the system is constructed on the basis of type-2 membership functions to handle uncertainties associated with information and data in the knowledge base; it can handle unknown environmental interactions; it will remove most noise added to data transmitted through a communication channel. The algorithm of the proposed controller is Lyapunov stable and tracking error is negligible. The proposed controller is compared with a type-1 fuzzy wavelet neural network controller and a conventional fuzzy controller and simulation results show the efficacy of the proposed method.

Study on Genetic Fuzzy Wavelet Neural Network Controller of Robotic Manipulator

This paper deals with the tracking controller design of robotic manipulator using genetic algorithm (GA). A genetic fuzzy wavelet neural network (GFWNN) controller is designed and implemented based on MATLAB in this paper, whose parameters are optimized by GA. The structure and algorithm of fuzzy wavelet neural network (FWNN) are described at first. Then the key content of GA used in this paper and the steps for using GA to optimize FWNN are demonstrated. Finally, a numerical simulation of tracking control for 2-link robotic manipulator is given to verify the effectiveness of the proposed method.

Adaptive Control of Machining Process Based on Fuzzy Wavelet Neural Network and Generalized Entropy Square Error

In order to improve the slow response and weak robustness of fuzzy control for machining process, combining qualitative knowledge expressiveness of fuzzy control with excellent local property of wavelet analysis and quantitative learning ability of neural network, a new kind of fuzzy wavelet neural network controller (FWNNC) is presented and a generalized entropy square error (GESE) function is also defined. The FWNNC is then applied to the on-line control of the cutting force under variable cutting conditions. Simulation results show that the proposed controller is superior to the fuzzy control or the neural network control for machining process and it has better static, dynamic performance. Experimental examples are also given to demonstrate the effectiveness of the proposed controller.

Study on a miniature robotic system for active monitoring in the human respiratory tract

A three degrees-of-freedom inchworm-type robot is presented for miniature endoscope inspection in the human respiratory system. The robot is actuated with pneumatic artificial muscle and clamped by two air chambers. The robot's extra resistance in the human trachea is analysed in detail and the dynamic model of the robot is built. In order to solve the problems of low stiffness and the nonlinear movement characteristic of the pneumatic robotic system, a fuzzy wavelet neural network controller is proposed to control the robot position servo system using the principle of pulse width modulation. Simulation and experimental results both demonstrate that the training of the fuzzy wavelet neural network controller is faster and the control quality is effectively improved. The monitoring experiments had been done in a swine trachea during mechanical ventilation, and the miniature system can dynamically measure respiratory parameters.

GAIT SYNTHESIS BASED ON FWN FOR A FIVE-LINK BIPED ROBOT

This paper presents a new reference trajectory and a fuzzy wavelet neural network controller to synthesize the gait of a five-link biped robot when walking on the level ground. Both the single support phase (SSP) and the double support phase (DSP) are considered. The gait of the biped can be determined when the trajectories of the hip and the swing limb are designed. The trajectories of the hip and the swing limb are approximated with the time polynomial functions. The coefficients of the functions are determined by the constraint equations cast in terms of coherent physical characteristics, such as repeatability, continuity, stability, and minimization of the effect of impact. The fuzzy wavelet neural network controller is trained by error back-propagation algorithm. Given the certain gait parameters such as the step length, maximum step height, walking speed, and so on, the control scheme can generate the smooth gait profiles. The simulation results show that the designed controller can follow the reference trajectories well.