Early wind turbine alarm prediction based on machine learning—Alarm Forecasting

Abstract

Alarm data is pivotal in curbing fault behavior in Wind Turbines (WTs) and forms the backbone for advanced predictive monitoring systems. Traditionally, research cohorts have been confined to utilizing alarm data solely as a diagnostic tool—merely indicative of unhealthy status. However, this study aims to offer a transformative leap towards preempting alarms, preventing alarms from triggering altogether, and consequently averting impending failures. Our proposed Alarm Forecasting and Classification (AFC) framework is designed on two successive modules: first, the regression module based on long short-term memory (LSTM) for time-series alarm forecasting, and thereafter, the classification module to implement alarm tagging on the forecasted alarm. This way, the entire alarm taxonomy can be forecasted reliably rather than a few specific alarms. 14 Senvion MM82 turbines with an operational period of 5 years are used as a case study; the results demonstrated 82%, 52%, and 41% accurate forecasts for 10, 20, and 30 min alarm forecasts, respectively. The results substantiate anticipating and averting alarms, which is significant in curbing alarm frequency and enhancing operational efficiency through proactive intervention.