A flexible threshold selection and fault detection method for health monitoring of offshore wind farms

Now-a-days the power generation sector is focussing more attention towards using renewable energy resource e.g. wind power. However the reliable operation of off-shore located wind farms is one of the major requirements to ensure consistent power supply to the grid. This can be ensured by using suitable real-time automated and remote structural health monitoring of the wind farm. This is one of the major applications of wireless sensor networks (WSN). This paper is an extension to our work in which we proposed an application-specific clustering and routing algorithm (NETCRP) for increasing the network-lifetime. This paper presents a novel loss-less packet-size reduction algorithm using flexible-thresholds for real-time fault detection in wind farms. The method utilizes the correlation between data samples collected at different times of the day for selecting the suitable thresholds. The method is compared with the primitive Mean-method which computes only one threshold for all data samples independent of the day and time of sample collection. This method then uses suitable fault detection methods using combination-summation and flow-direction of received data. The proposed method FTSFD (Flexible Threshold Selection and Fault Detection) provides two benefits - firstly, it is a simple method by which a remote observer can easily understand and predict real-time fault occurrences in the tower, and, secondly, it also helps in reducing the message size considerably and helps in increasing the overall network-lifetime of the system. The simulation results confirm that the proposed method has better fault-detection accuracy over the existing method and the overall network lifetime of the WSN in wind farm is increased by nearly ten times as compared to previous methods.