Variable length encoded genetic algorithm for optimal electrical distribution network routing

We describe a genetic algorithm approach to the practical problem of electrical distribution network routing through areas of relatively sparse population. The objective is evaluated through minimizing the total network investment cost, comprised of capital and operational costs, over the selected time period. The genetic algorithm benefits from a variable length genome, a number of interesting mutations based on Steiner points and careful management of population diversity. Evaluation on a real network benchmark shows improved performance to previous approaches under the assumption of free path traversal, and warrants further investigation.