Research Of Sensor Fault Tolerant Control Based On CSA-SVM Algorithm For EMB Actuator

Pure electric vehicles as the future direction development of the automobile, its optimal performance and reliability optimization has attracted wide attention both at home and abroad. EMB as an important part of automobile security, research on fault detection and fault tolerant control for it has very important research significance and application value. As a result of the process control needs the feedback information from multi sensors, how to correctly detect and diagnose the sensor faults, and to take timely and effective fault-tolerant means necessary is one of the key problems in the development of pure electric vehicles.Many of the existing sensor fault detection methods for linear systems are unable to adapt to the complex, integration problems appears in nonlinear system. For the nonlinear system, the proposed neural network method exist problems such as network structure diversity and sample complexity. It is difficult to achieve satisfactory results. In order to solve this problem, the dissertation puts forward a kind of sensor fault detection and fault tolerant control model based on CSA-SVM algorithm in view of the EMB actuator. The main research contents are as follows:Firstly, based on the principle of EMB actuators three-loop control, according to dynamic analysis of the actuator, the mathematical model of current, speed and pressure sensors are respectively established, and the sensor fault is analyzed and put forward the corresponding design goals and design principles.Secondly, research on fault detection and diagnosis in the light of the equipped current, speed and pressure sensor for EMB actuators. Combined with clonal selection algorithm and SVM algorithm, design a model of sensor fault detection and diagnosis based on CSA-SVM algorithm. This method not only can detect sensor fault timely and effectively, but also for further identification of specific types of fault. It can achieve real-time detection and estimation for sensor fault.Finally, according to the research of fault tolerant control of sensor, CSA-SVR algorithm is used to design an adaptive fault tolerant control strategy based on model reconstruction. The model predicts fault sensor signal in real time and meet the real-time EMB actuator fault isolation sensor signal, and to reconstruct the signal instead of fault signal the completion of fault tolerant control demand based on functional redundancy of EMB actuators.The experimental results show that sensor fault tolerant control model not only can real-time detect and diagnose sensor fault in the system, but also can achieve satisfactory output fault based on CSA-SVM algorithm. At the same time, the proposed optimization model is compared with the traditional model, has higher detection accuracy and precision. It can improve the reliability and robustness of fault detection and fault tolerant control model effectively.