Research On Day-Ahead Wind Power Prediction Of Regions Considering Multi-Source Wind Speed Correlation Characteristics

In recent years,the development of China's wind power industry has become increasingly vigorous.With the large-scale development and utilization of wind power in China,the problems of wind power consumption,grid connection and operation in the concentrated development of wind power have become a difficult problem to be solved.For the sake of effectively solving the problems caused by the safe and stable operation of wind power grids,improving the accuracy of wind power prediction has become a significant research point.Firstly,in the thesis,from the historical measured data of the wind farm wind tower and the corresponding numerical weather prediction(NWP)wind energy data,the relationship between the wind speed at different heights of the wind tower and the NWP wind speed at the hub of the wind turbine was qualitatively and quantitatively analyzed.The analysis results show that there is a gap between NWP wind speed and the actual wind speed of the wind tower.The difference between NWP wind mode and measured wind mode may be due to the fact that the calculation of mesoscale NWP does not take into account the relevant damage factors,such as complex terrain,wake and roughness,etc.Secondly,based on the analysis of the correlation between measured wind speed at different altitudes and corresponding NWP wind speed,a combined prediction model of wind power day-ahead based on corrected NWP and entropy weight method was proposed.The model uses long-short term memory(LSTM)to correct the NWP wind speed.Then,according to the NWP data corrected by the wind speed at different heights of wind tower,different day-ahead prediction models of wind power based on LSTM were established online.And then,the weight of each prediction result was determined dynamically by the entropy weight method,and the final prediction value is obtained by weighting.The results of the case study showed that the prediction indicators of the combined prediction model are better than the comparison models.Thirdly,in order to improve the day-ahead power prediction accuracy of wind power of regions,this thesis proposed an improved spatial up-scaling day-ahead prediction model of in regions of wind power based on fractal scaling factor transformation.The spectral clustering algorithm was used to cluster the grey correlation degree series with spatial-temporal characteristics among the NWP wind speed of each wind farm of regions,and the optimal segmentation results of sub-regions were obtained.Then the fractal scaling factor was used to optimize the traditional spatial up-scaling prediction process.Finally,the prediction results of the sub-region with the highest gray correlation degree of the overall output of regions were scaled up to the day-ahead prediction results of regions.The case study show that the prediction model can effectively improve the prediction accuracy.Finally,according to the day-ahead wind power prediction results of the case study,the spatial-temporal propagation characteristics between the prediction errors of each region were studied.This thesis verifies and analyses whether there was a significant correlation between prediction error within and between regions.On the basis of the second,it explored the relationship between wind speed and wind direction and prediction error of regions.The results show that there is a significant auto-correlation between the prediction errors of regional wind farms with different lag time.The cross-correlation model of prediction error between regional wind farms is obviously constrained by the main weather conditions,mainly factors are wind speed and wind direction.The influence of wind direction is crucial,and the impact of wind speed is more complicated.