An adaptive fuzzy self-learning technique for prediction of abnormal operation of electrical systems

This paper introduces the details of an intelligent adaptive fuzzy system-with self-learning functions-that could be installed to monitor electrical equipment or systems, and self-learn the trend of events leading to the failure of the monitored system. If any of the learned trends is repeated, the intelligent fuzzy system will predict the eventual failure of the monitored system in case of either human or automatic nonintervention. This paper presents the design details of the new intelligent fuzzy predictor. The self-learning process is accomplished using adaptive neuro-fuzzy techniques. Full details of the development of the new tool and the results of several test cases based on various practical applications and real-site data are included. Wavelet denoising is also used as the filtering technique for pre-preparation of the data before introducing it to the fuzzy predictor.