【摘要】 提出一种基于改进K-means聚类算法的ARMA预测模型来提高短期负荷预测的精度。采用改进K-means算法解决了传统K-means聚类算法聚类数不确定、初始聚类中心盲目选取的缺点,同时还可提高聚类的效率。在考虑气象因素的情况下,建立了气温与湿度的关系模型,通过负荷与气温的聚类分析,在不同簇内建立相应的ARMA预测模型进行负荷预测,对预测结果进行修正与综合。实例分析表明:运用该方法可将预测结果的平均相对误差控制在2%以内,夏季和冬季的预测平均误差更是低至1.78%和1.48%;相比于单纯的ARMA预测法和考虑气象因素的传统K-means聚类的ARMA预测法,该方法在提高预测精度上有更明显的优势。更多还原

【Abstract】 To improve the short-term load forecasting accuracy and efficiency, this paper proposes an ARMA forecasting model based on improved K-means clustering algorithm. Adoption of the improved K-means clustering algorithm helps to overcome disadvantages such as uncertainty of the cluster number by the traditional K-means clustering algorithm and blind choice of the initial clustering center and also improve the efficiency of the clustering. The relationship model of the temperature and humidity is established considering the meteorological factors,and through the cluster analysis of the load and temperature,the corresponding ARMA prediction model is established in different clusters to predict the load,and the prediction results are then corrected and integrated. Case analysis shows that the method proposed in this paper is able to control the average prediction error of the prediction results within 2%,while the figure can be as low as 1.78% and 1.48% in summer and winter respectively.Compared with the pure ARMA prediction method and the ARMA forecasting model based on the traditional K-means clustering algorithm considering the meteorological factor,the proposed method has more obvious advantages in improving the prediction accuracy.更多还原