A Neural Network-Based Approach to Sensor and Actuator Fault-Tolerant Control

Abstract

The paper is devoted to the problem of design of robust estimator and controller on the basis of the neural-network model represented in a linear parameter-varying form. In particular the fault-tolerant controller for multiple sensor and actuator faults is developed. The proposed approach is able to minimise the influence of the multiple faults of sensor as well as actuator on the controlled system. The robust estimator and robust controller procedure boil down to solving a set of linear matrix inequalities. The illustrative part of the paper is devoted to the application of the proposed approach to fault tolerant control of the laboratory multi-tank system.