A Self-reconstructing Algorithm for Single and Multiple-sensor Fault Isolation Based on Auto-associative Neural Networks

Recently different approaches have been developed in the field of sensor fault diagnostics based onauto-associative neural network (AANN). In this paper, we present a novel algorithm called selfreconstructingauto-associative neural network (S-AANN) which is able to detect and isolate singlefaulty sensor via reconstruction. We have also extended the algorithm to be applicable to multiplefault conditions. The algorithm uses a calibration model based on AANN. AANN can reconstruct thefaulty sensor using non-faulty sensors due to correlation between the process variables, and the meanof the difference between the reconstructed and original data determines which sensors are faulty. Thealgorithms are tested on a dimerization process. The simulation results show that the S-AANN canisolate multiple faulty sensors at a low computational time, which makes the algorithm appropriatecandidate for online applications.

Keywords:

Sensor Fault , Fault Isolation , Reconstruction Algorithm , Auto-associative Neural Networks , Multiple Faults