A novel collective health monitoring of a wind park

Abstract

Compared to a time-based maintenance schedule, condition-based maintenance provides better diagnostic information on the health condition of the different wind turbine components and subsystems. Rather than using an offline condition monitoring technique, which require the WT to be taken out of service, online condition monitoring does not require any interruption on the WT operation. The online condition monitoring system uses different types of sensors such as vibration, acoustic, temperature, current/voltage etc. Using a machine learning approach, we aim to establish a data driven fault prognosis framework. Instead of traditional wired communications, wireless communication systems such as Wireless Sensor Network have the advantages of easier installation and lower capital cost. We propose the use of WSN for collecting and transmitting the condition monitoring data to enhance the reliability of Wind Parks. Using data driven approach the collective health of the WP can be represented based on the condition of the individual wind turbines, which can be used for predicting the Remaining Useful Life of the system.