Early Fault Warning For Wind Turbine Group Main Transmission System Based On Multi-source Data Fusion

The rapid development of wind power technology has provided a huge opportunity for the development of China’s wind power industry.However,with the continuous expansion of wind turbine equipment capacity and the increase of equipment complexity,problems such as frequent equipment failures,poor reliability,and high operation and maintenance costs of wind turbines have become increasingly prominent.The main transmission system of the wind turbine,as an important path for energy and load transmission,is affected by factors such as inertia force and aerodynamic load due to the harsh working environment and complex and variable operating conditions,resulting in increased failure rate,which seriously affects the economic operation of the wind turbine.Therefore,it is the key to ensure the safe and reliable operation of wind turbines to carry out early fault warning of the main transmission system and grasp the real-time operation status of the main transmission system.This research is supported by the National Natural Science Foundation of China project "SCADA data-based dynamic behavior and state assessment of main transmission system of wind turbine under multiple working condition ".The research is aimed at the urgent need for status monitoring and early fault warning of the main transmission system of wind turbines in wind farms.Considering the complex and changeable operation status of the main transmission system induced by the coupling effects of inertia forces and system aerodynamic loads,this research takes the main transmission system as the research object,integrates SCADA and CMS monitoring data,explores the correlation mechanism between characteristic parameters and the operation status of the main transmission system,and studies the early fault warning method of the main transmission system of wind turbines.The research solves the problem of difficult monitoring of the operation status of the main transmission system under complex operating conditions.Furthermore,the research carries out integration and division studies of wind farm clusters,introduces transfer learning to reduce the influence of operational parameter differences among different turbines on the warning results,establishes a fault warning model of the main transmission system for wind farm clusters,and realizes the early fault warning of the main transmission system of multiple wind turbines in wind farms.The main research work is as follows:(1)An exposition of the wind turbine main transmission system,SCADA,and CMS systems is provided.Common causes of failures in the main components of the main transmission system are analyzed,revealing the correlation mechanism between the characteristic quantities and the operating characteristics of the main transmission system.State evaluation parameters that can characterize the operating status of wind turbines are selected.Preprocessing is applied to the SCADA and CMS data,and the feature parameters that are highly correlated with the state evaluation parameters are determined using the maximum information coefficient,which lays the foundation for establishing an early warning model for wind turbine main transmission system failures.(2)In response to the limitations of early fault warning methods based on a single data source,a method for early fault warning of the main transmission system of wind turbines based on multi-source data fusion is proposed.The SSAE algorithm is used to fuse the feature parameters of SCADA and CMS,deeply exploring the correlation characteristics between parameters,and a prediction model for state evaluation parameters is established based on PSO-LSTM.Different prediction models and the prediction performance when using a single data source are discussed.A quantification algorithm for evaluating the operating status of the main transmission system is proposed for residual analysis,and an early fault warning model for the main transmission system is established.Actual wind farm data validation shows that early fault warning based on multi-source data fusion can detect abnormal main transmission system operation in advance,which can provide a warning function earlier than the existing SCADA system.(3)An early warning method for wind farm cluster main transmission system failures is proposed to simplify research on wind turbine main transmission system failure warning in wind farms.Based on the characteristics of wind farms,the impact of external environmental factors and internal factors on operational conditions is explored,and the clustering method for wind farm clusters is investigated.A wind farm cluster division model based on spectral clustering is established.Based on this,the characteristics of operational parameters between wind turbines in the cluster are analyzed,and the differences in parameters between wind turbines in the cluster are revealed.Transfer learning is introduced to reduce the impact of parameter differences between wind turbines in the cluster on the early warning results,and an early warning model for wind farm clusters is established.Actual data analysis shows that this method can effectively achieve early warning of the main transmission systems of other wind turbines in the cluster,and compared with existing methods,the model performs better in terms of universality and monitoring accuracy.