Traveling salesman problem solving method fit for interactive repetitive simulation of large-scale distribution networks

Abstract

Based on experimental comparison, this paper discusses approximate solution methods of medium-scale traveling salesman problems (TSPs) that suit repetitive use in interactive simulation for globally optimizing a large-scale distribution logistic network. For constructing a globally optimized large-scale logistic network, the problem is decomposed into hundreds of sub-problems. And each sub-problem including above-mentioned TSPs should be repetitively solved. Thus, it is essential to find approximate solution methods of medium-scale TSPs that suit the heavily repetitive use in interactive simulation for globally optimizing a large-scale distribution logistic network. Accordingly, we made an experiment for comparison among approximate methods using random restart strategy that iterates the combination of random initialization and local search. As a result of this experimental comparison, we discovered one of above approximate methods could obtain solutions ensuring errors below 2-3% within 0.1 second.Thus, this method is considered promising to realize the system that enables to make above-mentioned interactive simulations repetitively for constructing a globally optimized large-scale logistic network.

Solving large batches of traveling salesman problems with parallel and distributed computing

“…Each TSP instance can be solved by calling a TSP Solver in parallel. Applications of large batches of TSPs include design of order picking warehouses [2], large scale distribution network simulation [3,4], case-based reasoning for repetitive TSPs [5], and delivery route optimization [6]. In these applications the TSP solving consumes most of the computational effort.…”

A Multi-World Genetic Algorithm to Optimize Delivery Problem with Interactive-Time

“…So-called random restart methods (Yanagiura & Ibaraki, 2000), which apply local search such as 2-opt for improving random initial solutions, can obtain near-optimal solutions. These include GRASP (Feo et al, 1994) or the elaborated random restart method (Kubota et al, 1999) that can guarantee responsiveness by limiting the number of repetitions. However, according to our experiments, the above-mentioned elaborated random restart method needed about 100 milliseconds to solve 40 cities TSP and to guarantee less than 3% errors (Kubota et al, 1999).…”

“…These include GRASP (Feo et al, 1994) or the elaborated random restart method (Kubota et al, 1999) that can guarantee responsiveness by limiting the number of repetitions. However, according to our experiments, the above-mentioned elaborated random restart method needed about 100 milliseconds to solve 40 cities TSP and to guarantee less than 3% errors (Kubota et al, 1999). As for the Genetic Algorithms (GA) to efficiently solve TSP, various techniques are proposed.…”

“…We developed an efficient method for solving the TSP by elaborating a random restart method. The developed method enables to guarantee the responsiveness by limiting the number of repetitions and by devising component methods and heuristics (Kubota et al, 1999). However, to meet the required guarantee of expert-level accuracy (below 3% of errors), it took more than 100 milliseconds to solve one TSP, which caused the time to solve one TSP should be significantly decreased.…”

Intelligent evolutional algorithm for distribution network optimization

“…Since a large-scale distrihution network has hundreds of distribution routes, one route has to he created withm a few seconds. And as the creation of each route is equivalent to a TSP (Traveling Salesman Problem) of tens of cities, an approximate solving method that enables both interactive responsiveness and practical optimality is required [2]. As a method for solving distribution problems with delivery time constraints, a method based on IP (Integer Programming) [3] has been proposed.…”