Research On Wind Power Forecasting And Optimal Scheduling Of Power System With Wind Power

With the continuous development of wind power technology,the proportion of wind power in the power system is gradually expanding.The randomness and volatility of wind power adversely affect the power grid.In order to reasonably arrange the schedule and ensure the safe and stable operation of the power system,it is necessary to accurately predict the wind power.Studying short-term wind power forecasting and power system economic optimal dispatching with wind farms have important engineering application value for improving wind energy utilization efficiency and optimizing the deployment of power grid dispatching operation.This paper studies the short-term wind power forecasting and the economically optimized dispatching of power systems with wind power.The main contents are as follows:1.Studied the characteristics of wind energy resources.The Weibull distribution function is used to establish the mathematical model of wind speed.Various factors affecting wind speed were studied.Three basic methods for short-term wind power forecasting were analyzed.In addition,the specific steps and evaluation criteria for short-term wind power forecasting were studied.2.ARMA time series method is used to predict wind speed and wind power respectively.A wind farm is simulated and analyzed by four forecasting methods such as ARMA method,BP neural network,RBF neural network and GRNN neural network.The results show that the prediction accuracy of the GRNN neural network is the highest and the prediction performance is the best.The equal-weight average combined forecasting method,the covariance optimal combination forecasting method and the entropy combination forecasting method were studied.It is concluded that the forecasting accuracy of the combined forecasting method is higher than that of the single forecasting method.The predictive performance of the covariance optimal combination forecasting method is better than the equal-weighted average combined forecasting method and the entropy combination forecasting method.3.Various influencing factors related to the output power of wind power were studied,and a wind power forecasting model based on multi-parameters was established.Through the simulation verification of different input combinations,the input combination of the best wind power prediction is wind speed,power,temperature and pressure.In this paper,a generalized regression neural network combination prediction model based on an improveduniversal gravitation search algorithm is proposed.Through simulation experiments on actual measured data of a wind farm,it is concluded that its prediction error is smaller than that of various combination forecasting methods,and it has better prediction performance and verifies the feasibility of the model.4.According to the output characteristics of the wind farm,the influence of wind power prediction error on grid uncertainty is considered.The cost of wind power is summed with the expected cost of wind power,the overestimation of wind power costs and the underestimation of wind power costs.The objective function uses the wind power cost and the thermal power cost,and establishes the power system optimization scheduling model that considers the wind power prediction error,sets up three different scenarios,and adopts the Wolf pack algorithm and the improved Wolf pack algorithm to optimize the different scenarios.The results show that the improved wolves algorithm solves the problem of the wolf colony algorithm falling into local optimization and has better optimization performance.Scenario 2 considers that grid-connected wind power can reduce the power generation cost of the grid,and Scenario 3 considers wind power prediction errors to further reduce the total operating cost of the grid.The results verify the validity of the model.