MANFIS Observer Based Sensor Fault Detection and Identification in Interacting Level Process with NN Based Threshold Generator
This study presents the design of Adaptive Neuro-Fuzzy Observer based sensor fault detection in a three-tank interacting level process. Three pairs of observers estimate the three system states. These Observers are designed with Multiple Adaptive Neuro-Fuzzy Inference System (MANFIS) that uses a neural network to fix optimal shape and parameters for the membership functions and effective rule base for the fuzzy system. Fault detection is performed by estimating the states of the level process and comparing them with measured values. A fault is signaled when the difference between the estimated and measured values crosses a threshold value. Decision functions are built from estimation errors to detect the fault. If any failure is identified, the control law is modified accordingly using the estimated value replacing the failed sensor output. In this research, MANFIS observer based fault detection is designed and simulated. Since, the threshold value can be different for different set point, a Neural Network (NN) based threshold generator is designed to give best threshold values for fault detection. The individual failures of three level sensors are considered for various set points and the results are discussed. The results show that the system is able to detect any sensor failure for any set point and to control the level in interacting tanks perfectly under failure situations.