Electricity distribution clustering and configuration study using KM-MST

Abstract

One of the most concerned things on the power system is how to build an optimum distribution configuration. Several techniques are developed in order to get the minimum number of distribution cost and also reduce total loss along the distribution system. Minimizing total cable spanning distance is the best way to achieve those objectives. Minimum spanning tree is appropriate to find the best configuration because of its ability to connect all nodes with minimum cost. The problem is, in this case study, the configuration must be deal with sub-system constrain. To deal with that, this study attempts to build the electricity configuration using KM-MST (K Means-Minimum Spanning Tree). By using KM-MST, the best number of cluster that can be built is 5 clusters with error 0.0958. Furthermore, as the final result, the optimal configuration for each sub-system in the whole system is achieved with total distance cost of 7.11222.