Reducing expert dependency in dynamic risk analysis through intelligent algorithms

Abstract

Risk management is a crucial tool for facilities to manage decision-making and operations. Risk management has become essential in the energy sector because of the recent global energy supply crisis around the world. Risk analysis, a key tool in risk management, was performed using different techniques. As in many other fields, machine learning has been used for risk analysis along with technological developments. This study aims to decrease the dependency on experts associated with the use of machine learning techniques for wind turbine risk analysis. Therefore, a SCADA-related database consisting of operational and alarm data of a wind turbine was examined. System failure and economic loss-based risk models have also been investigated. Features that affect risk levels are automatically included in the risk model and are aimed at reducing expert opinions. The study culminated in highly consistent real-time-based and future-based risk predictions.