Clustering Method of Daily Load Curve based on Dimension Reduction of Distribution Dynamic Characteristic Index

In order to adapt to the characteristics of the load in the power big data that tends to be massive and multidimensional, accurately extract the characteristic information of load, and support the application of the power big data middle platform, a daily load curve clustering method based on the dimension reduction of the distribution dynamic characteristic index was proposed by comprehensively considering the distribution characteristics and dynamic characteristics of the curves. Firstly, according to the distribution characteristic indexes such as daily load rate and daily peak valley difference, the dimension reduction dataset of distribution characteristics was extracted, Then, according to the peak, flat peak and low peak periods of the load, sampling time periods was divided, and the dynamic feature dimension reduction dataset was extracted according to the difference of sampling values in adjacent sampling time periods, Finally, the two types of reduced dimension datasets were combined to form a distributed dynamic feature reduced dimension dataset, and the pattern index clustering (PIC) method was adopted to cluster the reduced dimension datasets. The example showed that the clustering quality and robustness of the proposed method have certain advantages, the operation efficiency was in an acceptable range, and the comprehensive performance was great, which is suitable for accurate clustering of power load.