Application of an ExChange-Tree Operator in Two Stages of the Algorithm for Solving Large-Scale Network Problem

摘要

This paper focuses on the problem of the efficiency and accuracy of the algorithm for solving the large-scale traffic network layout optimization. According to the characteristics of these two problems, two new evolutionary strategies are presented through comprehensive consideration of the peculiarities of the crossover and mutation operation for improving the efficiency and accuracy of the algorithm in the traffic network layout environment. For solving the efficiency problem during generating feasible solutions, an ExChange-Tree Operator (ECTO) that a genetic algorithm crossover and mutation probability self-adaption is proposed. For lifting the accuracy of feasible solutions, the paper presents a bidirectional stochastic optimization strategy. Considering the best state of improvement, the experiment analyses from multi-angle the variation of different parameters from different scales and compares the gap ratio and precision ratio between two algorithms under the optimal parameter combination, showing that optimization results for two strategies based on a maximize performance analysis in solving efficiency and accuracy agrees fairly well with the experimentally determined values, which proves the effectiveness of the algorithm.