Detection of Parameter Variations by Continuous-Time Parity Equations

Abstract As many faults induce parameter variations, it is important for detection and diagnosis to observe their timely development. One possibility is to estimate the parameters directly by well-known least squares methods for dynamic systems, but this requires some computational effort at each sampling step. The proposed method uses the easy to calculate parity space equations with special, parameter-sensitive features and state variable filters to generate signal derivatives. Different to previous approaches here parity equations are used in the continuous time domain. Then the connection between physical and model parameters is direct and eases a fault diagnosis. A parameter classification table can be derived from the according pattern of the obtained set of residuals. This table enables to localize which physical parameter has been affected. Furthermore well known parity space design methods can be included for the detection of sensor faults.