Short-term Wind Speed Prediction Based On GA-SVR Model

Wind speed forcasting is an important work to wind farm.An appropriate model shouldbe established in the process of forcasting.This paper establishes an support vector regressionbased on genetic algorithm to forcast wind speed. Genetic algorithm origins from computer'ssimulation studies of biological system.It is efficient,parallel and global search.It canautomatically get and accumulate the knowledge about the search space during the searchprocess.Moreover,it can also control the search progress in order to get optimal solution.Support vector regression machine is a new and effective regression method which is basedon the VC theory of statistical analysis and the structural minimization theory.This methodgets the better generalization ability through finding the best compromise between the limitedinformation model's complexity and learning ability.Firstly,this paper uses genetic algorithm to optimize the support vector regressionparameters.It means that using genetic algorithm to select the support vector regression'sparameters.It can avoid the disadvatage of selecting parameters caused by randomness Themethod give a new way to choose the parament of support vector regression.Secondly,this paper collects the actual metrical data from the wind farm .Then it appliesstatistical methods and principal componet analysis to process the data. Through analysing thestandard deviation,it selects the suitable input data.After that, this paper uses principalcomponent analysis to reduce the inpute data.Then using software of spss to deletes thesingular point.Finally,this paper applies the preterated data to establish support vector regression modelbased on genetic algorithm and solves the model.Then it gets the result of wind speedprediction.After that,this chapter vertifies the results and model.