Improved algorithm of the Back Propagation neural network and its application in fault diagnosis of air-cooling condenser

Abstract

This paper addresses the application of neural network to air-cooling condenser faults diagnosis. For traditional Back Propagation (BP) neural network algorithm, the learning rate selection is depended on experience and trial. In this paper, an improved BP neural network algorithm with self adaptive learning rate is proposed using the fundamental equation. Unlike existing algorithm, self adaptive learning rate depends on only network topology, training samples, average quadratic error and error curve surface gradient but not artificial selection. The train results show iteration times is less than that of traditional algorithm with constant learning rate and it is a feasible method to diagnose air-cooling condenser faults.