A Fuzzy Inference-Based Fault Detection Scheme Using Adaptive Thresholds for Health Monitoring of Offshore Wind-Farms

Abstract

The proposed flexible threshold selection and fuzzy inference system-based fault detection system (FTSFFDS) is a simple and efficient method of fault localization which forms the basis for the structural health monitoring of offshore wind farms. Based on the actual parameter readings obtained by the sensors attached to the towers, one can remotely observe and monitor the health condition of wind farm. The simulation considers day time and night time collected data for wave-height and sea-surface temperatures. The correlation between the data sets indicates the choice of adaptive (dynamic) thresholds. The method involves quantization for extracting meaningful information from the collected samples. A fuzzy inferencing system is proposed which uses combination-summation and flow-direction of received data to accurately predict faults in the wind farm. The method is compared with primitive mean method. The proposed method provides a simple approach for detecting real-time fault occurrences and in-effect helps in reducing the message size considerably to increase the network lifetime of the wireless sensor network by nearly ten times. The results confirm that the proposed method FTSFFDS has better fault-prediction accuracy over the existing method.