Cascade Steepest Descendant Learning Algorithm for Multilayer Feedforward Neural Network

概要

論文の詳細を見る
In this article, a new and efficient multilayer neural networks learning algorithm is presented. The key concept of this new algorithm is the two-stage implementation of the steepest descendant method. At the first stage, the steepest descendant method is used to search the optimal learning constant η and momentum term a for each weights updating process. At the second stage, the Delta learning rule is then employed to modify the connecting Weights in terms of the optimal η and α. Computer simulations show that the proposed new algorithm outmatches other learning algorithms both in converging speed and success rate. On real industrial application, we first apply the new neural network learning algorithm to the identification of a highly nonlinear injection molding barrel system. Experimental results demonstrate that the new algorithm can precisely identify the complicate injection molding barrel system. Further more, a self-tuning PID controller for precise temperature control based on the trained neural network barrel model is developed. Real experiments show that the self-tuning PID controller can precisely control the barrel temperature within ±0.5℃.
一般社団法人日本機械学会の論文
2000-06-15