Load Classification Method And Its Application

Load classification is basis of power system and load shifting management. It’s also the basis of TOU and load forecasting. Good load classification method can guide system planning and load shifting management. So this article will discuss some methodof load classification. And the method is applied to a grid of detailed load classification analysis. Load classification methods have already achieved some achievements, but thereare some problems. This paper presents a classification method based on the distance andshape of load curve. Work of this paper is as following:(1) Some basic issues involved in clustering theory are discussed, including data preprocessing method, similarity measure, how to determine number of clusters, evaluation functions and so on;(2) About substation load classification, multi-user load classification and time of single-user load classification, respectively describes several common load classification methods, and evaluate the advantages and disadvantages;(3) Load curve with time series, so this paper introduces a shape-based similarity measure, and defines what’s similar in shape, and the graphics of different degree of similarity is provide to further illustrate;(4) Principal component analysis is applied to determine the weights of weighted Euclidean distance. K-means clustering method and the AIC criterion are combined to get the best number of clusters and initial cluster center, the number of clusters as input clusters, the class centers as the initial cluster centers.(5) To overcome the disadvantages, similarity measure, the number of clusters and initialcluster centers to make improvements, so a method based on the distance of load curve andshape is proposed.(6) Analyze the load characteristics of a grid. The distance and shape classificationis used of industry’s classification. At last, compare with the K-means clustering method and fuzzy C-means method, the classification based on the distance and shape of load curve method. Result shows accuracy is improved by this method.