基于One-class SVM的大工业负荷模型研究

Research on Great Industry Load Model Based on One-class SVM

随着新型能源的不断增长,电力系统的负荷特性不断变化,电网的供需关系特性改变,给电网运行调峰、有序用电、客户管理带来了新的挑战。提出基于单类支持向量机(One-class SVM)的大工业用户负荷模型,从典型日选取、相关系数、频谱分析、指标体系4个层面建立大工业负荷模型,按照典型客户、客户群、行业负荷3个层级研究多时间维度、多层次的大用户负荷特性,实现对大工业电力负荷数据的实时跟踪。实际应用表明,该方法能够有效分析各类大用户用电规律以及相关特性。

With the continuous growth of new energy generation,the development of power system load characteristics and supply-demand relation,the power grid peak regulation,orderly power utility and customer management are facing new challenges.Based on One-class support vector machines(One-class SVM) of large industrial users load model, the big industrial load model is set up from four aspects of the typical day selection, correlation coefficient, spectral analysis, index system. According to the three levels of typical customers, customer base and industry loads, the multi-time dimension and multi-level characteristics of large user load are studied to realize real-time tracking of large industrial power load data.The practical application shows that this method can effectively analyze the power consumption rules and related characteristics of various large users.