Neural network based intelligent sensor fault detection in a three tanks interacting level process

This paper presents sensor fault detection in a three interacting level system and fault detection and control under sensor failure conditions. Trained Back Propagation neural networks estimate the process states. The fault detection and isolation is done by decision logic. If any failure is identified, the control law is modified accordingly using the estimated value as the substitution for the failed process sensor output to implement the fault tolerant control. The individual failures of three level sensors are considered in this paper. The results show that the system is able to control the level in the interacting tanks perfectly in all the failure situations.