The Optimization Decision Of Wind Power Prediction Models Based On Multi-index Fusion Evaluationand Case-based Reasoning

As the pollution of environment and depletion of traditional resources, theresearch of clean and renewable energy is got more and more attention. With itscharacteristics of no renewable and no pollution, wind power is widely favoredin the global country and the ratio of wind power in power grid is increasingyear by year. While, the wind has the characteristics of volatility and intermittentso that it may cause the certain impact to the grid, when the large-scale windpower integrated to grid system. It’s an effective method to forecast wind poweraccurately for arrangement of dispatching plan reasonably and the safe, stableand economic operation of power grid.There are varieties of wind power prediction models. In generally, themodels have different forecasting performance, each have advantages anddisadvantages. For selecting high accuracy prediction model to forecasting windpower, the optimization decision of wind power prediction models based onmulti-index fusion evaluation and case-based reasoning method is put forward.And a large amount of simulation research is conducted on MATLAB platform.The major research content is as listed below:(1) The background and significance of the wind power prediction are elaborated. The overseas and domestic research status and the commonclassification methods of wind energy prediction and evaluation of eachforecasting methods are summarized.(2) The wind turbine power model is analyzed and the main factors thatinfluencing the output of wind power are determined. The data characteristics ofthe main influence factors and the correlation between the main influencefactors sequence and wind power sequence are summarized.(3) The evaluation of models is the important basis to the modeloptimization. For the question that different evaluation indexes may lead todifferent evaluation results and it is one-sided to evaluate by single index, thispaper proposed the multi-index fusion evaluation method based on the idea ofmaximizing deviation and subjective correction coefficient for the optimizationof short-term wind power prediction models. Firstly, the more comprehensiveevaluation indexes system of wind power prediction models is established.Secondly, considering the traditional maximizing deviations method is tooobjective in determining the weight of evaluation indexes which ignoring theimportant degree of each index. So this paper joining the subjective correctioncoefficient to the traditional idea of maximizing deviations, and thecomprehensive weight of each evaluation index is determined with thesubjective experience and objective information. Finally, all the predictionmodels are ranked and the excellent model is selected according to the fusionevaluation value. (4) To improve the work efficiency of model optimization, the optimizationdecision method of wind power prediction models based on case-basedreasoning paper is put forward. Wind speed, wind direction and temperaturesequence are determined as the description characteristics of case. Theoptimization case library is established of each month based on historical data ofwind farm. In the case retrieve, the similarity matching algorithm based on thekey points and the piecewise linearization is used to retrieve the similar cases inthe case library, and the solution of matched cases as the optimization modelstypes of new case, so improving the work efficiency for optimizing models.