Performance Analysis And Prediction Of Wind Turbine Based On SCADA Data

Due to historical reasons,the traditional way of wind farm location is usually to measure the wind resources in the local area for at least one year,which takes a long time and has a certain degree of human orientation;most of the wind farms are located in high altitude areas with abundant wind resources,where the environmental conditions are harsh and complex,and the operation status of wind turbines is difficult to grasp due to the impact of service environmental factors Control and other issues.Therefore,this paper proposes a Monte Carlo simulation based wind turbine power generation estimation method to achieve fast and accurate location of wind farms;through analyzing the environmental factors affecting the performance of wind turbines,this paper proposes a performance prediction method based on SCADA data,which is of great significance to the state pre-warning of wind turbines.The main work is as follows:(1)This paper briefly introduces the system composition and working principle of direct drive wind turbine,and analyzes the distribution characteristics of wind speed,wind direction,environmental temperature,power and hub speed in SCADA data of wind turbine.(2)Based on Monte Carlo simulation of wind turbine power generation estimation method.Firstly,the distribution characteristics of wind speed in the wind turbine service environment are analyzed,and the probability density function of wind speed is obtained;secondly,the Monte Carlo simulation method is used to simulate the wind speed according to the wind speed distribution characteristics,and the Monte Carlo simulation accuracy is calculated;finally,the wind turbine power generation capacity is estimated by comprehensively considering the wake effect and other energy consumption.(3)Power curve evaluation method of wind turbine based on SCADA data.Firstly,the original data of wind turbine SCADA is extracted and preprocessed,and the performance reliability model of wind turbine is built based on the performance reliability theory;secondly,the power discreteness of wind turbine is considered,and the power variation coefficient is calculated;then,the performance scoring rules are established by integrating the performance reliability and power variation coefficient of wind turbine.(4)Wind turbine power prediction method based on twin support vector regression.Firstly,in twin support vector regression method,linear kernel function and polynomial kernel function are used respectively,and genetic algorithm is used to establish prediction model in parameter optimization;secondly,environmental factors affecting the performance of wind turbine are analyzed to determine the wind speed,wind direction and ambient temperature as input,and the power of wind turbine as output,and data preprocessing is carried out;finally,2 MW wind turbine as the research object,through comparison,the prediction accuracy of twin support vector regression method with polynomial kernel function is higher.(5)The prediction method of wind turbine service quality grade based on meteorological information.Combined with the wind turbine service quality evaluation method and the twin support vector regression based wind turbine power prediction method to predict the operation state,through the combination of the two methods,a wind turbine service quality prediction method based on meteorological information is obtained;finally,the feasibility of the method is verified by a case.