Fault Detection and Classification of an Electrohydrostatic Actuator Using a Neural Network Trained by the Smooth Variable Structure Filter

Abstract

A multilayered neural network is a multi-input, multioutput (MIMO) nonlinear system in which training can be regarded as a nonlinear parameter estimation problem by estimating the network weights. In this paper, the relatively new smooth variable structure filter (SVSF) is used for the training of a nonlinear multilayered feed forward network. The SVSF is a recursive sliding mode parameter and state estimator that has a predictor-corrector form. Using a switching gain, a corrective term is calculated to force the network weights to converge to within a neighbourhood of the optimal weight values. SVSF-based trained neural networks are used to classify faults on the input and output data of an electrohydrostatic actuator (EHA). Two faults are induced in the system: friction and leakage. Furthermore, a comparative study between the popular back propagation method, the extended Kalman filter (EKF), and the SVSF is presented.