近年来，道路危险货物运输事故时有发生。本项目以道路危险货物运输网络为研究对象，在分析危险货物运输系统特点的基础上，构建新的危险货物运输风险模型，进而研究危险货物“显性”禁行路段结构化筛选算法、基于Levenberg-Marquardt神经网络和遗传算法的“隐性”禁行路段智能筛选算法；然后分常态和防恐背景2种情形，分别构建常态下考虑公平性与畅通可靠性的多起讫点间危险货物运输网络多目标鲁棒优化模型、防恐背景下考虑恐怖袭击风险和运输网络点边重复度的多起讫点间危险货物运输网络多目标鲁棒优化模型；进而综合运用鲁棒优化理论和改进的多目标智能优化算法对鲁棒优化模型进行求解；最后以甘肃省陇东地区危险货物运输网络鲁棒优化为例展开实证研究。研究成果可为保障危险货物运输安全提供基础理论支持。

Transportation accidents of road hazardous materials happen sometimes in recent years. This project deals with the network optimization problem associated with hazardous materials transportation (HMT) and attempts to provide a relevant and rigorous research. Firstly, a new model is built to measure HMT risks, followed by a structured algorithm for the case of explicitly-restricted roads and an intelligent screening algorithm based on the Levenberg-Marquardt algorithm and genetic algorithm for the case of implicitly-restricted roads. Then, two kinds of multi-objective robust optimization models are proposed depending on whether or not anti-terrorism is taken into consideration. Under normal circumstances, fairness and reliability are considered while under the anti-terrorism circumstances, the risk of terrorist attacks and repetition degree of nodes and edges are considered for hazmat from multiple origins to multiple destinations. The robust optimization theory and improved multi-objective intelligent algorithms are combined to solve the models. Finally, the Longdong area of Gansu province is taken as an example for empirical study. The study results can provide the theoretic support for protecting the transportation safety of hazardous materials.