Fault Diagnosis of an Electrohydraulic System by Using Fuzzy C-Means Clustering

Abstract

Hydraulic systems typically operate under harsh conditions, such as in the heavy industry and military domain. Early diagnosis of single or multiple faults is very important to keep the system in safe working conditions. To generate different faults, an exhaustive simulation stage is required before validating the study with an experimental setup. Therefore, in this study, an electrohydraulic system model was first created using the Matlab-Simscape hydraulic system library. After the simulation stage, a data-based fault diagnosis method was applied through certain statistical feature calculations using time, frequency, and time-based frequency of the data collected using the Diagnostic Feature Designer Toolbox under theMatlab program, which allows the use of all signal and data-based debugging, identification, and classification methods under the same platform. Based on the features ranked by the Diagnostic Feature Designer Toolbox, the best fault model fits were investigated by clustering with Fuzzy C-Means.