Derivative free filtering in hydraulic systems for fault identification

Abstract

Diagnosis of incipient faults in hydraulic systems is of prime importance due to the performance and reliability demands. This paper outlines the application of derivative free filtering in hydraulic systems for the purpose of real-time fault identification. A flexible experimental setup is constructed in order to simulate different types of faults. The method in this paper deals with internal leakage faults. A detailed non-linear model of the hydraulic actuator experimental setup is developed and validated. Robust control strategies typically hide the presence of faults during incipient stages, making identification of the fault difficult. A partial feedback linearization based robust position control strategy is also presented and faults are identified in the presence of robust control. Faults in the hydraulic systems are modeled as parametric faults and second order divided difference filtering (DDF) is used to estimate the states and the parameters. The efficacy of this estimation algorithm is demonstrated using different fault levels as well as different fault growth profiles. The accuracy and the reliability of the methodology are also demonstrated by identifying faults as small as .01 l/s.