Condition Monitoring of Wind Turbine Anemometers Based on Combined Model Deep Learning

The dynamic working environment brings challenges to the condition monitoring of anemometers. To accurately grasp the actual performance status of wind turbines (WTs) and timely detect anemometer faults, a combination of Particle Swarm Optimization (PSO) and the long and short term memory network (LSTM) based anemometer status monitoring method is proposed. Firstly, utilizing the wind speed data collected by the anemometer as an input variable, select the wind turbine (WT) with high similarity through similarity analysis. Then, use the PSO to enhance the structural parameters of the LSTM network to acquire efficient anemometer state estimation. This method can monitor the abnormal state of the anemometer and reconstruct the faulty wind speed data. Finally, to demonstrate the efficiency of the approach, the condition of the WT anemometer is predicted using examples.