Research And Application Of Fault Tolerant Control System Based On Soft Sensor

With the development of computer technology and the improvement of the level of industrial automation.The requirements for the security and reliability of the control system are becoming higher and higher.Considering the nonlinear of the industrial system and sensors are easy to malfunction.In this paper,based on the data driven technology,the control method of a soft sensor fault detection and soft closed loop fault tolerant based on the LS-SVM was proposed.The simulation and experimental research ware carried out on 3 common fault problems of sensor,such as sensor failures,constant gain and deviation.Around the proposed scheme,the following research work has been mainly carried out.In this paper,the experimental platform for sensor fault detection and fault tolerance control was built.Based on data driven method,the output signal of LS-SVM soft sensor prediction system is constructed by using process data of I/O.The prediction output of LS-SVM soft sensor and the output of actual sensor are compared.So the residual signal is obtained.Analyzing the history and current residual signals,and detect faults based on SPRT algorithm.If the decision sensor fails,the fault indication value is output.A fault tolerant control mechanism for soft closed loop switching was designed.When the sensor fails to detect the fault,the prediction output of model is used to replace the actual output of the physical sensor.After the system is restored to normal,the actual sensor will be reproduced.The output is sent to the controller as a feedback signal.Finally,the simulation and experimental verification showed that the combination of predictive output of LS-SVM soft sensor and SPRT can reliably detect sensor fault and the safety of fault tolerance of sensor fault can be realized by soft closed loop switching during the fault.