Application of robust l1 fault detection and isolation to an industrial benchmark using a genetic algorithm

To aid in the transition of fault detection and isolation (FDI) theory to practice, a realistic, nonlinear industrial diesel engine benchmark was defined by Blanke et al. This paper applies a robust l1 fault detection and isolation (FDI) technique to this benchmark. Using a linear model and assuming appropriate parametric uncertainty, a bank of robust linear estimators is developed using mixed structured singular value (MSSV) and l1 theories. To obtain the estimator parameters a real-coded genetic algorithm is used to solve the optimization problem. These estimators are then used to perform FDI of the industrial diesel engine actuator. The results illustrate the power of hybrid evolutionary algebraic techniques for solving important problems in estimation and control.