Parity vector approach for detecting failures in dynamic systems

Abstract

An especially simple parity vector method is denned; in this method, the redundant measurement relations are derived using a prediction of the state and the output of the system. The parity vector is a linear combination of the redundant measurement; in the absence of failure, it is only a function of redundant measurement noise, but in the presence of failure, it deviates from the normal value. The generalized likelihood ratio test is used to detect the character of the parity vector and to declare the failure component.