MLP, PNN and Fuzzy Logic Improved by Genetic Algorithms in Fault Detection and Isolation

S.B. Bouabdallah and M. Tagina (Tunisia)


Fault Detection and Isolation, Analytical Redundancy, Artificial Neural Networks, Fuzzy Logic, Genetic Algorithms, and Uncertain Parameter Systems.


In this paper a multilayer Perceptron (MLP), a Probabilistic Neural Network (PNN) and a fuzzy approach are proposed in on-line sensors and actuators fault detection and isolation for systems with parameter uncertainties. The residuals obtained by analytical redundancy relations (ARRs) are used as inputs of the three systems. The MLP is trained to present for each output signal 1 in the occurrence of a fault at the associated variable and 0 otherwise, PNN has one output which is trained to present the fault index (1 in case of a fault-free context, 2 in case of a fault affecting variable 1…). These previous approaches are improved by the use of genetic algorithms (GA) to optimize the initial weights and bias in case of MLP, the spread in case of PNN and the membership functions parameters in case of the fuzzy approach. MLP, PNN and fuzzy approach are compared through a hydraulic example.

Important Links:

Go Back