Research On Power Load Classification Based On Optimized K-means Algorithm

Scientific and accurate classification of load is very meaningful to planning and economic operation of power system. However, with the development of the power system, load data become more and more complex, types of power users is various, the current classification methods have been unable to meet the demand of power system, it is urgent research to find new load classification method to meet current load demand and provide a useful alternative to current power system analysis and decision-makers work.K-means algorithm is widely used in the fields of science and engineering, for its reliability theory, simple, fast convergence, and can effectively handle large data sets. But it still has the problem such as dependence on initial conditions, the method of traditional algorithm by choosing initial center point randomly and determining the cluster number in advance influence its effect. Based on in-depth study on the K-means algorithm, and power load have massive data, multiple dimensions and other characteristics, the paper selects GSA-based elbow judgment, distance cost function, histograms to determine the optimal number of clusters and density parameters to choose the initial cluster centers. Combination with these research results, comparative analysis of various optimization algorithms, select the most suitable combination to optimize algorithm for power load classification, by using GSA-based elbow judgment and density algorithm to optimize K-means algorithm.Demonstrate the effectiveness of the optimization algorithm by an example, the example choose the actual user loads data in certain areas, and the results showed that the optimized algorithm can reduce the number of iterations classification calculations and improve the accuracy of the load classification, achieve load classification automatically based on data characteristics, showed its superiority.