Automated fault detection in power distribution networks using a hybrid fuzzy–genetic algorithm approach

• Published in: Engineering applications of artificial intelligence 2000-08, Vol.13 (4), p.407-418
• Main Authors: Srinivasan, Dipti, Cheu, Ruey Long, Poh, Young Peng, Ng, Albert Kim Chwee
• Format: Article
• Language: English
• Subjects: Fault detection; Fuzzy logic; Genetic algorithms; Hybrid AI systems; Power distribution networks
• ISSN: 0952-1976; 1873-6769
• DOI: 10.1016/S0952-1976(00)00012-9

This paper describes the development of an intelligent technique based on artificial intelligence for automatically detecting incidents on power distribution networks. A hybrid combination of fuzzy logic and genetic algorithms (GAs) has been applied to detect faults in these networks. The robust nature of a fuzzy controller allows it to model functions of arbitrary complexity, while the maximising capabilities of GAs allow optimisation of the fuzzy design parameters to achieve optimal performance. The hybrid approach used in this paper builds on these individual strengths and seeks to blend fuzzy set and GAs techniques to compensate for their inadequacies. The technique for fault detection is described and verified with experiments on a 33 kV test system containing 12 busbars, eight transformers and eight line sections. The results obtained from the test data file of 500 test cases contain only one undetected case (0.2%), 458 correctly detected cases (91.6%) of actual faults and 41 cases (8.2%) where the protection system components either had not operated or had malfunctioned but were correctly identified by the incident detection system.