Optimization of urban MV multi-loop electric power distribution networks structure using Artificial Intelligence methods

Urban medium voltage (MV) electric power distribution networks are supplied with primary (HV/MV) substations. These networks supply secondary (MV/LV) transformer substations and are often built as closed structures - loop arrangements. The design problem of optimal urban MV distribution network structure consists of determining the number of primary substations, establishing the number of MV loops supplied with the primary substations, and assigning the secondary MV/LV transformer substations to the MV loops. The optimization task becomes especially complex when the number of the primary substations is greater than one. The minimum of total annual costs is sought. The total annual costs include: fixed (investment) costs, variable (operating) costs and supply-interruption costs. Typical constraints are also accounted for. The so defined optimization problem is a complicated mathematical problem in respect of computational effort. In order to resolve the mathematical model of the optimization problem, evolutionary algorithms and artificial neural networks have been used. Exemplary computational experiments have been executed on the model of urban MV multi-loop electric power distribution networks. The results from the evolutionary algorithm and the artificial neural network calculations have been compared.