The Research Of Fault Diagnosis Based On Information Of Multi-sensors

With the continuous development of science and technology, the system's application ability and increasing the level of modernization, and the complexity of the system also will be getting higher and higher, the possibility of system failure is increasing, and to bring about system analysis and control of bad influence. Therefore, the need of the method to improve the overall reliability and maintainability to avoid the problem is becoming increasing popular. In this context, resulting in the fault diagnosis is happened. On the other hand, because of advances in computer technology, multi-sensor measurement technology has the continuous development. In actual projects, using multiple sensors to measure the state can get a better estimate, based on multi-sensor measurement has been widely used.In this context, this paper a comprehensive study of information-based multi-sensor fault diagnosis method. Primarily done research in the following two aspects:(1) To realize the multi-sensor system state estimation, this paper adopts a centralized multi-sensor information fusion method for fusion, through the introduction of generalized measurement vector, design the Kalman filter, the system state estimation. On this basis, fault diagnosis based on Kalman filtering method had been studied. Simulation examples show that the design of multi-sensor information based on Kalman filter application to obtain good results.(2) Study based on multi-sensor information of the H∞fault detection filter design problem had been considered. First of all, centralized multi-sensor fusion methods, and multi-sensor information fusion estimation, and then apply the H∞Filter Theory into the design of the filter optimization problem, given the appropriate performance indicators, using LMI techniques, is given and proved the existence of the filter conditions and the system gain matrix solution method. The filter is designed to meet the residual right of interference robustness and fault sensitivity. At last, example of failure of the system in the single-and multi-failure of the simulation respectively, as compared with the traditional single sensor, this means a failure occurs, you can in a shorter period of time to detect the fault occurred.