Inverse fault detection and diagnosis problem in discrete dynamic systems

This paper investigates an inverse fault detection and diagnosis problem in discrete dynamic systems. The problem is how to adjust the system parameters according to observation value of inputs and outputs so that the system is concordant. First we formulate the problem as a least square problem with interval coefficients. Then two algorithms for this problem are presented. The first algorithm based on the expected value of observation value of inputs and outputs. We are only required to solve a classical least square problem in this algorithm and the algorithm is robust. The second algorithm by using linear programming approach can deal with large scale systems and suit for on line adjustment.