Research Of Power Load Characteristics And User Classification Based On Data Mining

With the highly developing of economy in recent years,electric power,as the basis of development of economy,is in a higher demand in different aspects which leads to a severe imbalance compared to the whole electric power supply.On the other hand,in the peak period the utilization of electric power declines and electric power supply could not meet the demand,in the valley period the demand is at a low level and rest capacity of electric power is huge.Thus the difference of power load between peak time and valley time is becoming larger.Considering the modem electric power feature,there should be a deep analysis to the power load charact eristics and a proposal to implement a different price between the electric power which is used in peak period and that used in valley period,which could improve the utilization of electric power and optimize the structure of electric power use in the whole electric power grids.For the huge scale of modern power grid users,there must be impossible to analyze the power user behavior from one to an other.So power user classification is the basis to analyze the whole power load characteristics.In this paper,a classification method of multi view clustering is presented.First of all,according to the electric power consumption,nine typical industries are selected in Shenzhen.The power use from nine typical industries is clustering an alyzed in the load fluctuation index and the feature of real-time load in real time.And two industries with same load fluctuation index or same real-time load feature would be concluded in one type.A ccording to the cluster result of load fluctuation index,the power grid user load fluctuation index feature could be gained.According to the cluster result of real-time load,every industry user power consumption model could be concluded.Thus a data base of power grid user load feature model from several typical industries in Shenzhen could be established and accomplishment to verify the available power grid user load feature could be gained.Furthermore a tracking check and update could be achieved to get a more suitable electric power load feature model.Applying to the real situation,the electric power grid user load feature presented in this paper resembles a lot with the real user power consumption feature and in high precision.The enterprises could take advantage of recent user power consumption feature model to classify them and study the user power consumption trend from different type to forecast and manage the related load.