Research On Categorized Time-of-Use Power Price Based On Fuzzy C-Means Clustering

With the development of the tertiary industry and rising residential electricity consumption, the peak-valley difference and the maximum load of the grid is increasing, and a negative impact on stability of the grid and power supply quality is caused. In order to ease the situation of electric power shortage, the state energetically extends DSM (Demand Side Management), and TOU (Time-of-use) power price is an effective means of DSM incentives. Since the eighties of last century, our country began to try to implement TOU power price and some initial effect has been achieved in recent years, but there are still many problems in the implementation of TOU power price.At present, the state adopts the same TOU power price for different types of power users, yet different power users have different load characteristics. According to this feature, the traditional "cutting without differentiation" pricing method of TOU power price can be changed, and categorized TOU power price can be implemented, that is. dilferent types of power users develop different TOU power price to meet the way of their electricity consumption. Currently our country classifies power users through demand analysis and divides in accordance with the power users’ nature and marketing business needs, and it cannot rellect the load characteristics of power usersAimed to this problem, this paper make a clustering analysis for power users according to their load curve via FCM (Fuzzy c-Means) clustering algorithm, and power users are divided into six categories with obvious characteristics. FCM clustering algorithm is the most perfectly theoretical and widely used algorithm in algorithms based on the objective function.Because initialization and parameters setting of FCM algorithm have important influence on clustering results, this article optimizes them through the subtractive clustering algorithm, fuzzy decision algorithm and clustering validity function, and has achieved comparatively scientific and rational clustering resultsOn the basis of classification for power users, this paper divides power grid synthetic load into peak, flat and valley time periods via fuzzy degree of membership and FCM clustering algorithm, and analyzes the degree of demand response of different kinds of users through them synthetic load curves. It carries on the simulated analysis of the effect of them on the synthetic load of power grid through electric quantity transfer, and puts forward reasonable suggestions on the optimization of their TOU power price presently according to simulation results. These results provide the theoretical basis for realizing categorized TOU power price, and are of great significance to fully arousing the enthusiasm of users responding to TOU power price and better improving the load curve of power grid.