Puncochar, Ivo (University of West Bohemia), Straka, Ondrej (University of West Bohemia)

Parity-Space and Multiple-Model based Approaches to Measurement Fault Estimation

Scheduled for presentation during the Regular Session "Observers and Estimation" (WeA1), Wednesday, November 16, 2022, 17:20−17:40, MAIN ROOM - E406

16th European Workshop on Advanced Control and Diagnosis, November 16-18, 2022, Nancy, France

This information is tentative and subject to change. Compiled on April 23, 2024

Abstract

The paper compares two approaches to measurement fault estimation for linear discrete-time stochastic systems. The first fault estimator utilizes the parity-space approach. It assumes that at most a limited number of components of the measurement fault can be non-zero. A fault is detected using the 𝜒2 test applied to the parity-space based residuals and then the indices of non-zero components and fault itself are estimated. The second fault estimator is based on the multiple-model approach. The space of measurement faults is quantized to construct a set of models and a variable structure interacting multiple-model estimator is employed to estimate the state and the measurement fault. Both approaches are compared in a numerical example.