Research On Detection And Diagnosis Of Abnormal State Of Main Bearing Temperature Of Wind Turbine Based On Pattern Recognition

The main bearing is one of the main factors causing the failure of a wind turbine.In a traditional wind turbine,the main bearing plays a very important role as one of the core components of the wind turbine.However,wind turbines are usually located at high altitudes,and the working environment is relatively challenging.The harsh working environment makes it more vulnerable to natural disasters such as rain,snow,freezing,etc.,resulting in poor lubrication inside and causing various problems.What's more,it may cause the whole machine to be paralyzed.How to accurately conduct scientific and effective state detection and diagnosis of wind turbine components has become an urgent need for social production and development.The most direct way to detect and diagnose the main bearing failure is to install an online vibration monitoring system to analyze the main bearing failure.However,the early fans did not have this system.Today,the number of vibration sensors installed on site is also Rarely,but it is more common to use SCADA system data to estimate whether the main bearing will fail.The closest data to vibration is the temperature of the main bearing.There are about one or two hundred measurement points for the fan in the SCADA system.Except for the temperature of the main bearing,some are relevant and some are not very relevant;how to choose a suitable algorithm to analyze the relevant parameters Nowadays,the two typical methods that are more commonly used are classical mathematical models and intelligent algorithms.Due to the single failure index evaluation index of the classic mathematical model,it is impossible to comprehensively consider the abnormal temperature of the main bearing,and it needs to be analyzed one by one.In this case,it is difficult to establish an accurate mathematical model.For black box models such as intelligent algorithms,a large amount of data is required to train the model,but the problem also exists.A large amount of data is required to ensure that your model is accurate.Due to many conditions,it is impossible to obtain enough state data..For example,when the failure occurred and which unit occurred,are full of uncertainties.Sometimes the scope of data acquisition may need to cover all your working conditions.In the face of the above problems,this paper takes the main bearing of a direct-drive 2MW wind turbine as the main research object.The research is divided into three sections:the preprocessing of the data related to the abnormal temperature of the main bearing,the selection of the main bearing temperature detection method and the design of the detection scheme and the main Intelligent reasoning and exploration of the cause of abnormal bearingtemperature.The main contents of the research are as follows:(1)Preprocessing of data related to abnormal main bearing temperature.In the selection of data preprocessing methods,the Analytic Hierarchy Process and Evidence Theory are selected,and the relevant parameters affecting the abnormal strength of the main bearing temperature are set through the expert decision standard table and the relevant level matrix table,and the @RISK software is called and multiple expert opinions are combined.,Obtain the measurement points of the parameters related to the abnormal temperature of the main bearing.For the problem data(null values,error values,singular points,etc.)existing in the parameter measurement point data,it is planned to use the local outlier factor(LOF)algorithm to clean up the data,and then apply it as input to the main bearing temperature abnormality detection program.(2)Selection of main bearing temperature detection method and detection scheme design.After expounding and analyzing the two main methods of fault abnormality detection and diagnosis,select the BP neural network in the main bearing temperature intelligent detection method,and plan to improve it with the enhanced particle filter algorithm,and design the wind turbine main engine of the EPF-BP algorithm.State detection program for abnormal bearing temperature,build a detection model for abnormal state of main bearing temperature.The temperature of the main bearing of a direct drive 2MW wind turbine is simulated.The results show that the diagnosis scheme can more effectively and accurately identify the abnormal temperature of the main bearing of the core component of the wind turbine,so as to improve the reliability of the main bearing.(3)Intelligent reasoning and exploration of the cause of abnormal main bearing temperature.The reason for the abnormal temperature of the main bearing.An effective knowledge model is used to reason about the cause of the abnormal temperature of the main bearing,and the protégé open source software is used to construct an intelligent reasoning model based on the main bearing temperature abnormality.The information rules of Web ontology language-OWL are added to realize information sharing,interaction and processing on the semantic level.Through knowledge reasoning,explore the reasons for the abnormal temperature of the main bearing.