基于多级聚类分析和支持向量机的空间负荷预测方法

为充分利用元胞负荷与元胞属性之间的相关联系来改善空间负荷预测效果,提出了基于多级聚类分析和支持向量机的空间负荷预测方法。首先生成元胞并建立元胞属性集合,根据各属性对元胞进行多级聚类分析,其中采用改进的k-均值算法确定聚类数目和初始聚类中心,来得到逐级细化的元胞分类;然后针对不同类型的元胞建立各自的支持向量机预测模型,同时利用遗传算法进行参数优化以提高预测模型的适应度;最后将待预测元胞的相关属性作为输入向量并代入所建立的预测模型中计算出目标年各元胞负荷最大值,从而实现空间负荷预测。工程实例分析表明了该方法的实用性和有效性。

A Spatial Load Forecasting Method Based on Multilevel Clustering Analysis and Support Vector Machine

In order to make full use of the relationship between the cellular load and cellular properties to improve the results of spatial load forecasting (SLF), a method of SLF based on multilevel clustering analysis and support vector machine (MCA-SVM) is proposed. Firstly, cells are generated and their properties set are established. The cells are classified by multilevel clustering analysis according to the properties, where k-means algorithm is used to determine the clustering number and the initial clustering centers. Secondly, the forecasting models based on support vector machine are developed for different cellular types. Meanwhile the genetic algorithm is used to optimize the parameters of the models for improving their fitness. Finally, the cellular properties are substituted into the models as input vectors to forecast the maximal load in the target year for all the cells, realizing spatial load forecasting. An actual SLF case shows the practicality and effectiveness of the proposed method. This work is supported by National Natural Science Foundation of China (No. 51177009) and Jilin Provincial Natural Science Foundation of China (No. 20140101079JC).