Research Of The Power Customers’ Peaking Averting Potential Based On Data Mining

With the rapid development of economy and the improvement of people’s living standard,great changes have taken place in the power consumption structure of various industries and the load characteristics of power customers, the power sector is facing a severe short supply situation. Especially in summer and winter, the demand for power load is much higher than the grid power supply. The phenomenon of imbalance between power supply and demand occurs in most regions of China. Power imbalance between supply and demand not only brings trouble to the grid, but also can not provide customer with high-quality and stable electrical energy. In order to ensure safe and economic operation of power grid, the power sector should know the electricity consumption behavior of customer in time and adopt a series of measures when the supply of power grid is less than the demand of power customers,so as achieve better electricity order. At present, orderly electricity consumption formulated by power grid mainly contains peak averting, peak shifting, days off and maintenance. Peak averting is simple, flexible, easy to implement the "being up quickly and down quickly" and able to deal with emergency in time, so peak averting of orderly power consumption should be studied further. It is particularly important how to analyze the peak averting potential of massive power customers and to provide the customers with decision support about peak averting.Because the data of electricity load of customers is huge, we need to analyze the power mode of customers by using the related algorithm of data mining, which can simplify the huge data of electricity load and is convenient to analyze the typical electricity consumption behavior of customers. In this paper, a lot of historical load data is clustered by using the K-means algorithm for electricity customers and the typical daily load curves of power customers are obtained. Then the typical daily load curve of each power customer is used as its most representative load curve, which provides data for the peak averting analysis of each electric power customer in this paper.In this paper, based on the traditional load characteristic indicators and according to the actual demand, several new indicators applied to peak averting analysis are defined. Then the model of peak averting value is established with combining the new load characteristicindicators. And according to the model, the values of various characteristic indicators and the peak averting of power customers are calculated quantitatively. On the basis of the above work, the vast power customers are clustered into five categories and according to the clustering results, a guiding sorting table of peak averting is generated. Sort the values of each cluster center from large to small and the results of rank ordering represent that the ability of peak averting are very strong, strong, relatively strong, generally and poor. The group with strong ability should give priority to avert peak, and the customers in the same group should avert peak at the same time.Based on the algorithm design in this paper, the actual load data of several electric power customers and power grid of some districts are chosen to verify and analyze the correction of the peak value of the model. The result shows that the sort way of peak averting is correct and effective, which provides value guide for the power sector to find the valuable electricity customers for peak averting.