Abstract: Aiming at the optimization of hazardous materials transportation routes,a stochastic optimization model with the goal of the minimum expected value of transportation cost and risk coefficient under the condition of double uncertainty was constructed with the consideration of the uncertainty of the terminal demand on hazmat and population center.A hybrid method of multi-objective particle swarm optimization genetic hybrid algorithm based on priority and combining sample average approximation was designed for solving the stochastic model.Furthermore,an elitist solution set updating method based on dynamic crowding distance was adopted.The feasibility of the model and the effectiveness of the algorithm were verified by example analysis and algorithm comparison.Furthermore,the stability analysis was performed on the results under different sample sizes,and the sensitivity analysis was performed on demand and population.Experimental results showed that the Improved Muti-Objective Particle Swarm Optimization(IMOPSO)algorithm could effectively solve this problem with good convergence and variety.The uncertainties of different sample size,demand and population had impact on the route optimization.The results of transportation route optimization could provide references for decision makers.

Key words: hazardous material transportation, improved multi-objective particle swarm optimization algorithm, uncertainty, sample average approximation method, route optimization

摘要： 针对危险品运输路径优化问题,考虑危险品终端需求量和人口中心的不确定性影响,构建了双重不确定条件下以运输成本和风险系数的期望值最低为目标的随机优化模型。设计了基于优先权的多目标粒子群遗传混合算法结合样本平均近似法对随机模型进行求解,并进一步采用基于动态拥挤距离的精英解集更新。通过算例分析和算法对比验证了模型的可行性和算法的有效性,并进一步对不同样本规模下的结果进行了稳定性分析,对需求量和人口数量两个不确定条件进行灵敏度分析。实验结果表明改进的多目标粒子群算法可有效解决该问题,且具有很好的收敛性和多样性,不同样本规模和不确定性条件均会对路径规划产生影响,所得结果可以为决策者提供参考。

关键词: 危险品运输, 改进的多目标粒子群算法, 不确定性, 样本平均近似法, 路径优化