Wind turbine generator fault prediction method based on deep learning

Abstract

The invention discloses a wind turbine generator fault prediction method based on deep learning, and the method aims at the SCADA data of the operation state of a wind turbine generator, and carries out the fault prediction based on a fault prediction model. The fault prediction method comprises the following steps: recombining SCADA data of the running state of the wind turbine generator into a running data graph; extracting fault features in the operation data graph by using a CNN; taking LSTM as a fault prediction model main body, and converting the single-frame fault feature into a time sequence fault feature; and completing the integration of time sequence fault features through a full connection layer, and obtaining a fault prediction result. For massive SCADA operation data, the CNN-LSTM framework is used for completing long-period prediction of the wind turbine generator faults, multi-dimensional high-sampling-frequency long-period time sequence SCADA data in the wind turbine generator can be processed, and accurate prediction of the long-period wind turbine generator faults can be effectively achieved.

Description

technical field [0001] The invention belongs to the technical field of fault prediction of wind power generators, and relates to a deep learning-based fault prediction method for wind turbines, in particular to a deep learning-based wind turbine fault prediction method for SCADA data of wind turbines. Background technique [0002] With the increasingly prominent problems of environmental pollution and energy crisis, wind energy, as a clean and renewable energy, has attracted more and more attention from all over the world, and has developed rapidly in recent years. While the installed capacity of the wind power industry continues to grow, it is also facing some new problems. Due to the limitation of the distribution of wind resources, the natural environment, the size and working mode of wind turbines themselves, wind turbines are mostly distributed in areas where there are few people and the environment is relatively changeable. had a great impact. According to statistics...

AI Technical Summary

Problems solved by technology

However, due to the lack of effective processing methods for massive time-series data in the current wind turbine prediction method, there are limitations in the feature learning ability of fault data, the processing ability of long-period high-real-time state data, and prediction efficiency and accuracy. Therefore, the current wind turbine There is still room for further improvement and improvement in fault prediction methods

Images