Open AccessFlexible threshold selection and fault prediction method for health monitoring of offshore wind farm
Author(s) -
Agarwal Deepshikha,
Kishor Nand,
Raghuvanshi Ajay Singh
Publication year - 2015
Publication title -
iet wireless sensor systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.433
H-Index - 27
eISSN - 2043-6394
pISSN - 2043-6386
DOI - 10.1049/iet-wss.2014.0008
Subject(s) - real time computing , fault (geology) , computer science , network packet , offshore wind power , wireless sensor network , routing protocol , grid , cluster analysis , wind power , selection (genetic algorithm) , reliability engineering , data mining , engineering , computer network , artificial intelligence , geometry , mathematics , seismology , electrical engineering , geology
The reliable operation of offshore wind farms requires suitable real‐time, automated and remote structural‐health monitoring of the wind farm using wireless sensor networks (WSNs) for ensuring consistent power supply to the grid. This study presents a novel WSN packet‐size reduction technique using flexible thresholds for real‐time fault prediction in wind farms which is an extension to the authors work proposing application‐specific network‐lifetime enhancing tri‐level clustering and routing protocol for increasing the network lifetime. The selection of flexible threshold utilises the degree of correlation between data samples collected at different times of the day. Using combination‐summation and flow direction of received data, accurate fault prediction is achieved. The method is compared with the primitive mean method. The proposed method flexible threshold selection and fault prediction (FTSFP) provides a simple means for predicting real‐time fault occurrences in the towers and helps in reducing the message size considerably, thereby increasing the network lifetime of the system nearly by ten times. The results confirm that FTSFP has better fault‐prediction accuracy over the existing method.