Robust fault detection filter design for stochastic bilinear systems

This paper is concerned with the problem of fault diagnosis for a class of bilinear stochastic systems with unknown disturbances. At first, the filter with restricted structure is designed so that the unknown disturbances are decoupled from the residual and the effects of process and measurement noises to the residual are minimized. The filter design with general structure is also discussed. The chi-squared distribution test is then applied to examine the residual and diagnose the faults. Finally, an example of practical application system is used to illustrate the applicability of the proposed method.