A novel self-adaptive control framework via wavelet neural network

Most of existent controllers are model-based which need the knowledge about the controlled object. In fact, most industrial processes are featured with no precise mathematical model of the process. In this paper, we present a novel idea and algorithm on model free adaptive controller. First, we describe a new self-adaptive control framework based on the wavelet neural network. The identifier can identify nonlinear dynamic character of the system more precisely, and the controller can produce more complex control strategies. Generally, the initial parameters about the network we can obtain randomly, in this paper, we integrate the setting of initial parameters with the wavelet type, time frequency parameters of the wavelet and the training samples to avoid the sharp vibration at the beginning of the training course. Finally, we represent the iteration equations about the weight of the network, the scale factor and displacement factor based on the conception of information entropy. The simulation results show that the novel control system has high approximation accuracy, excellent control effect and strong anti-jamming ability.