Computational RSOM-MapReduce Intelligence for Real Time Faults Detection in Wind Energy Generator

Abstract

This paper investigates a computational intelligence approach using the Recurrent Self-Organizing Map (RSOM) and its MapReduce framework for anomaly detection in wind energy generation. Given that wind power represents approximately 25% of the global installed renewable energy capacity, its reliability is crucial. The proposed methodology leverages an intelligent dynamic unsupervised deep learning algorithm within a distributed MapReduce processing paradigm to diagnose and isolate faults in real time across various wind energy sources. The applied computational approach differs from existing methods by simultaneously analyzing, in adverse environments, multiple signals acquired in real time for the purpose of intelligent diagnosis of wind turbine systems located in remote mountainous regions. Signal acquisition and processing were carried out in an experimental setup. The findings provide insights into its potential benefits, limitations, and economic viability.