Compound substation characteristics analysis based on multi-objective model and cluster-correct algorithm.

• Cluster model considering both load pattern and load composition is proposed. • A two-step algorithm is proposed to solve multi-objective cluster model effectively. • New algorithm reduces the chance of local optima when clustering a compound data case. • Seasonal load characteristics changes of substations are analyzed. • Substations share the similar load patterns may have different load composition. Extracting and analyzing abundant and accurate load characteristics of electricity distribution substations is the basis of works for load modeling. Substation load consists of a variety of subordinate loads with varied characteristics. Therefore, we propose a multi-objective clustering model taking both substation load pattern and load composition into consideration to cluster substations precisely. We propose a two-step algorithm called correct-cluster algorithm. We first cluster substations according to their load patterns. Then a new method is provided to correct these clustering results according to composition proportion of loads attached to these substations. The new method performs well in completing a multi-objective cluster in the test and practical cases. It is effective to prevent the result from falling into the local optimum, and it also achieves clustering results with more details. 39 substations with 19,407 users are taken as a sample in the practical case, where we use this method to analyze the load characteristics of substations in different seasons. Clusters of substations with similar load patterns and similar subordinate loads are formed, compared and analyzed. According to the cluster results, unusual substations share similar load patterns and diverse subordinate load composition proportions are also found and analyzed, which should be considered in future works about substation load modeling. [ABSTRACT FROM AUTHOR]