Sensor fault diagnosis for electro-hydraulic actuator based on QPSO-LSSVR

Abstract

In this paper, a novel fault diagnosis method based on quantum particle swarm optimization (QPSO) and least square support vector regression (LSSVR) algorithm, was proposed to detect sensor faults for electro-hydraulic actuators. Prediction model based on LSSVR algorithm is established to forecast sensor output. By calculating the residual between the forecast output of the LSSVR model and the actual output of the sensor, a fault can be indicated. Furthermore, to improve the prediction accuracy of the LSSVR model, QPSO is employed to optimize the hyper-parameters used in the LSSVR model. Simulation experiments show that, compared with PSO-LSSVR, the prediction error of QPSO-LSSVR is smaller and the convergence rate is faster. The effectiveness of the fault diagnosis method for detecting several typical sensor faults, which occurred in the actuator system, is also verified in the simulation experiments.