State Prediction Of Wind Turbines Based On Neural Network And Fuzzy Comprehensive Evaluation

With the adjustment of the energy structure,new environmentally friendly energy sources have developed rapidly.Wind power generation has received extensive attention and application due to its rich resources,clean and environmentally friendly,and mature technology.Wind energy is currently one of the most promising new energy sources,with huge environmental and economic benefits.However,the environment of the wind turbine is harsh and the operating conditions are complex and changeable.In order to ensure the safety and efficiency of power generation,it is of great significance to predict the important components of the wind turbine and the overall unit.First of all,this article introduces the working principle and basic structure of the wind turbine,and counts the fault information of each component of the wind turbine.This paper introduces and classifies the monitoring parameters of the SCADA system of wind turbines;deals with the useless data and abnormal data of SCADA,and establishes a feature selection model based on MIV-BP neural network.Secondly,three state parameters that have a greater impact on wind turbines are selected for predictive modeling,which are gearbox oil temperature,rotor speed,and active power.In this paper,the correlation of input parameters is sorted through the MIV-BP neural network and finally the input parameters of the prediction model are determined.Established wind turbine state parameter prediction models based on DBN and LSTM,and tested the parameters of the neural network to determine the prediction model structure.According to the three evaluation indicators of RMSE,MAPE and MAE,the effectiveness of the neural network prediction model is verified.Finally,in view of the problem that the wind turbine state parameter prediction model can only evaluate a single index parameter,a fuzzy comprehensive evaluationbased wind turbine overall operation state evaluation model.The model establishes a comprehensive evaluation index system for the operation status of wind turbines,and divides the operation status of wind turbines into four levels: "healthy,good,attention,and fault".The model revises the temperature index,obtains the index degradation degree,uses the combined weighting method to determine the index weight,and replaces the traditional membership function with the normal cloud model to determine the index membership degree.Finally,the health evaluation matrix is obtained and the health status of the unit is judged according to the principle of maximum membership degree.The experimental results verify the effectiveness of the model.