Adaptive Neural Network Structure Based on Sensitivity Analysis
Artificial neural networks have been widely used for system identification and adaptive control in recent years.Unlike traditional methods,the neural network based approach does not require a priori knowledge on the model of the system under control and also has some other significant advantages,such as adaptive learning ability as well as nonlinear mapping ability.In general,a larger neural network(i.e.,more hidden nodes and weights)may yield faster rate of convergence and is more powerful for solving problems.On the other hand,a smaller neural network requires less computation time that is advantageous in real-time environment where speed is crucial.Therefore,choosing the optimal network dimension becomes a critical issue in the design of artificial neural networks.In this paper,a neural network controller with adaptive size/structure is investigated and applied to regulate a class of DC power supplies.Reducing the size of the neural network can increase its speed of response and thus improve the performance of the overall system.An adaptive pruning algorithm based on sensitivity analysis will be discussed and computer simulation results will be presented.
Artificial Neural Networks Sensitivity Analysis Pruning Algorithm
Xiao-Hua Yu
Department of Electrical Engineering California Polytechnic State University San Luis Obispo,USA
国际会议
The World Forum on Smart Materials and Smart Structures Technology(SMSST07)(2007年世界智能材料与智能结构技术论坛)
重庆·南京
英文
2007-05-01（万方平台首次上网日期，不代表论文的发表时间）