A New Index and Classification Approach for Load Pattern Analysis of Large Electricity Customers

Conducting load pattern analysis is an important task in obtaining typical load profiles (TLPs) of customers and grouping them into classes according to their load characteristics. When using clustering techniques to obtain the load patterns of electricity customers, choosing a suitable clustering algorithm and determining an appropriate cluster number are always important and difficult issues. Therefore, this paper proposes a stability index for choosing the most suitable clustering algorithm and a priority index (based on the stability index) for determining the priority rank of clusters. Based on three known clustering algorithms, an analysis approach is presented to demonstrate the use of these indices. In the approach, all load curves of customers are first clustered with the clustering algorithms under a serial given number of clusters. The two above-mentioned indices are then calculated. Following this, the most suitable clustering algorithm is chosen and the optimal number of clusters can be determined from the rank list for special application purposes. A case study with large electricity customers connected to a distribution network in Northern China illustrates the approach. The results prove the efficiency of the approach using the proposed indices in the classification and generation of the TLPs of large electricity customers.