Research of Sensor Fault-Tolerant Control Based on CSA-SVR Algorithm for EMB Actuator System

Abstract

The electro mechanical brake (EMB) system is a efficient pure electric vehicle braking system in where technologies in the fields of electronics, mechanics, and vehicle communication network were considered at the same time. Because that offered information to the electronic control of the EMB system is taken from several detectors, during developments of the pure electric vehicle, unsolved difficulties in EMB system are the accurate fault detection as well as the timely process of a fault tolerant control. In this paper, the fault detection and the fault tolerant control on the mathematical models of a current, a speed, and a pressure detector loaded on the EMB actuator system will be studied. Based on the three-loop control architecture model of the EMB actuator, these models of detectors are constructed by the method of the dynamics analysis of the actuating agency respectively. On the purpose of improving the accuracy of fault detection, the clonal selection - support vector regression algorithm (CSA-SVR) is proposed, which is a combination of CSA and SVR calculation methods. Through CSA-SVR algorithm, optimized parameters of support vector machine (SVM) and improved accuracy of fault detection are obtained. The adaptive fault tolerant control architectural model mentioned which is designed using CSA-SVR algorithm in this paper shows effective function in fault detection, isolation from fault, and fault estimation.