The Online Models based Fault Tolerant Control for Multiple Sensor Fault of Multi-variable Nonlinear Processes

The online model-based fault diagnosis and tolerant (FDT) control for nonlinear system under multiple sensor failures are investigated in the paper. In the close-loop control system, faults may be propagated and they can cause a malfunction of the loop. A control loop failure can easily cause production stoppage or malfunction at the plant level. The fault-tolerant control is introduced to increase the reliability of automated systems. A complete FDT architecture based on neural network model is proposed to efficiently handle the fault diagnosis and the accommodation of multiple sensor failures in online situations. The simulation results in a three-tank benchmark problem show the effectiveness and fastness of the proposed FDT framework.