Intelligent evolutional algorithm for distribution network optimization

Abstract

Based on experimental comparison, this paper discusses GA applied solving methods of medium-scale (100 cities) time constraint Traveling Salesman Problem (TSP) that suit repetitive use in interactive simulation for optimizing a large-scale distribution network. To solve both energy problems and environmental problems simultaneously, it is important to optimize a large-scale distribution network shared by multiple enterprises.Recently, in addition to the distribution efficiency, transportation specified time-constraints are increasingly required to improve productivity through supply chain management. Moreover, the network optimality should be considered from various aspects by human experts. Thus, both practical optimality and interactive response time are required to this simulation. To satisfy these requirements, a "selfish-gene with limited allowance" type GA is proposed. Here, each gene of an individual satisfies only its constraints selfishly, disregarding the constraints of other genes in the same individual. Further, to some extent, even individuals that violate constraints can survive over generations. And, even inferior individuals have the chance of their reproductions over generations. Thus, these individuals get the chance of improvement. As a result of our experimental comparison, the proposed solving method could solve one hundred time-constraint TSPs within 10% errors, less than a few minntes.

Selfish-gene Tolerant Generic Algorithms to solve large-scale constraint TSPs

“…As an approximate solving method for TSPs, the authors realized a highly optimal and high speed solving method with GAs [2] utilizing the NI (Nearest Insertion) method embedded with knowledge in crossover and mutation operations. To realize the above concept through improving this method, an insertion method called selfish NI method" is applied to put nodes into a tour.…”

“…However, simple mutation operations using only random processes without a local search method on problem-oriented knowledge are not efficient for the GA's convergence to the optimal solution. Thus, we developed the block type mutation [2].…”

“…Here, through applying "selfish NI method" described in (1), the knowledge incorporated NI crossover operation already described [2] is extended to efficiently deal with time constraints. This extended crossover operation consists of the following steps: 1) Determine randomly a crossover point in one of the parents' chromosomes.…”