基于GA-LSSVM的短期风功率预测研究

Short-term Wind Power Prediction Based on GA-LSSVM

随着风电渗透率的逐年增加,精确的风功率预测对于电力系统调度运行具有至关重要的意义。提出了基于GA-LSSVM的短期风功率预测研究。首先,利用基于密度的聚类算法对风功率历史异常数据进行识别与聚类分群,完成数据清洗;其次,通过GA(Genetic Algorithm,遗传算法)对LSSVM(Least Squares Support Vector Machine,最小二乘支持向量机)的惩罚系数γ以及核函数的参数σ进行动态寻优,构建GA-LSSVM的短期风功率预测模型;最后,通过风电场的历史数据进一步验证所提方法和所建模型的可行性。结果表明,所提出的通过GA优化LSSVM参数的方法可以提高风功率短期预测的精度。

As the penetration of wind power increases year by year,accurate wind power prediction is of importance in power system dispatching operation.Therefore,a short-term wind power prediction based on GA-LSSVM was proposed in this paper.Firstly,a density based clustering algorithm was used to identify and cluster historical abnormal wind power data to complete data cleaning;secondly,a short-term wind power prediction model of GA-LSSVM was constructed by dynamic optimization of the penalty coefficientγof least squares support vector machine(LSSVM)and the parameterσof kernel function by genetic algorithm(GA);finally,the feasibility of the proposed method and the established model were further verified through historical data of wind farms.The results show that the method proposed in this paper for optimizing LSSVM parameters by GA can be accurate for short-term prediction of wind power.