Hydraulic valve fault diagnosis based on EEWT and POA-KELM

Abstract With the wide application of hydraulic systems in the industrial and mechanical fields, the analysis and study of hydraulic valve failures have become increasingly important. This study aims to identify and analyze possible faults by analyzing the operating status of hydraulic valves under different loads. In this study, we propose using a combination of two phases for the failure analysis of hydraulic valves. The first stage is the signal processing stage, in which the pressure signals of hydraulic valves under different loads are collected, the pressure signals are decomposed using the extended empirical wavelet transform (EEWT) to obtain a series of sub-signal components, and the 17 eigenvectors of these components are computed, and the feature selection is carried out using the method of sequence backward selection. Next is the classification stage, where the features selected in the first stage are input into a kernel extreme learning machine (KELM) classifier improved by the Pelican optimization algorithm (POA) to classify and identify the anomalous signals. The experimental results show that the fault diagnosis method based on EEWT and POA-KELM can effectively detect and recognize hydraulic valve faults.