Hydraulic System Fault Diagnosis Method Based on HPSO AND WP-EE

Abstract

A fault diagnosis approach of hydraulic system based on Hybrid Particle Swarm Optimization (HPSO) algorithm and Wavelet Packet Energy Entropy (WP-EE) is presented. A heuristic algorithm is adopted to give a transition from particle swarm search to gradient descending search. A HPSO algorithm is formed with the heuristic algorithm, which is used to optimize BP neural network weights and threshold. Then an application of fault diagnosis with HPSO and WP-EE based on a valve-controlled hydraulic motor system is presented. After wavelet packet decomposition of the acquired hydraulic pressure signal, the WP-EE is extracted; the eigenvector of wavelet packet is constructed. Taking this eigenvector as fault sample, the neural network based on HPSO algorithm is trained to implement the fault diagnosis for the hydraulic system. The test result shows that the fault recognition accuracy and the efficiency in the hydraulic system are improved by this method.