Detection and Diagnosis of Multiple Faults With Uncertain Modeling Parameters

In this brief, a fault-detection and diagnosis method is proposed for the stochastic hybrid system with the consideration of model parameter uncertainty. To negate the effect of model uncertainty, a compensation step is introduced, which uses a compensation parameter to adjust the degree of dependence of the filtering on the model or the measurements. To determine the compensator parameter governing the degree of dependence, an orthogonality principle between the estimation error and the residual is applied. Numerical simulations with a second-order tracking system and a pilot-scale experiment study on a quadruple water tank system are conducted to demonstrate the effectiveness of the proposed approach.