An approach to sensor fault diagnosis based on fully-decoupled parity equation and parameter estimation

A new approach for multiple sensor faults diagnosis is presented. Fully-decoupled parity equations and parameters estimation are combined in this method. Residual generated from a fully-decoupled parity equation is sensitive to a special sensor fault and decoupled from all other sensor faults. The faults in sensors are represented as biases and changes in scale factor. A parameter estimator based on Kalman filter is designed to identify biases and changes in scale factor from the information contained in the residuals. The fully-decoupled parity vectors, which are designed to detect and isolate sensor faults, are generated in various conditions, and their existence condition is given. A simulation example on a linear airplane model is given for illustration.