A Tabu-Search based Neuro-Fuzzy Inference System for fault diagnosis

This paper presents a novel hybrid Tabu Search (TS) Subtractive Clustering (SC) based NeuroFuzzy Inference System (ANFIS) design for fault detection. The proposed model uses the TS algorithm to find optimal parameters for Subtractive Clustering (SC) based ANFIS. The developed TS-SC-ANFIS scheme provides critical information about the presence or absence of a fault. The TS being an efficient local search technique, shows remarkable success in finding optimal cluster parameters which proves instrumental in ANFIS training, making it efficient in fault detection. The proposed scheme is evaluated on a laboratory scale coupled-tank system. Fault detection results presented at the end of the paper using fresh set of data show successful diagnosis of most incipient leakage faults in the coupled-tank system.