基于潜在冲突贝叶斯推理的高速铁路列车运行图反馈优化理论研究

项目名称： 基于潜在冲突贝叶斯推理的高速铁路列车运行图反馈优化理论研究

项目编号： No.61503311

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 文超

作者单位： 西南交通大学

项目金额： 20万元

中文摘要： 目前我国高速铁路列车运行图编制较少考虑运行图的预期执行效果，影响了高速铁路的运输服务质量和运输能力，构建高速铁路列车运行图编制的反馈优化理论具有重要理论价值和现实意义。本项目通过对调度指挥仿真及实际列车运行数据的挖掘，揭示高速铁路列车运行图潜在冲突的链式机理，构建列车运行图潜在冲突结构化概率推理的链式效应图并将其转化为贝叶斯网。运用贝叶斯预测推理研究潜在冲突演化趋势预判及发展态势评估的方法, 设计潜在冲突态势的计算方法，基于贝叶斯诊断推理研究潜在冲突的断链机制，建立以潜在冲突态势为优化目标的高速铁路列车运行图反馈优化模型，构建基于潜在冲突贝叶斯推理的高速铁路列车运行图反馈优化理论。本项目从保障高速铁路运输服务质量及降低调度指挥复杂性的目标管理角度出发，进行高速铁路列车运行图编制模式和流程的新探索，发展和创新高速铁路列车运行图的编制理论，研究成果也能为调度指挥及列车运行调整提供一定理论支持。

中文关键词： 高速铁路；列车运行图编制；潜在冲突；贝叶斯推理；反馈优化

英文摘要： At present, the scheduling of train timetable in China scarcely considers the expected implementing performances, which brings adverse impacts on the transport quality and capacity of high-speed rail. Realizing the high speed train timetable feedback optimization is significant for improving the quality of train timetable design skills theoretically and practically. This project reveals the mechanism of chain law for latent conflicts in high-speed train timetable based on the mining of train operation data which comes from the dispatching simulation and actual train operation of high speed rail. Building a chain effect graph of latent conflicts of train operation in the form of a structured probabilistic reasoning and turning it into a Bayesian network, using Bayesian predictive inferences to achieve evolutionary trend forecasting of latent conflicts and post-conflict situation assessment. Based on Bayesian diagnostic reasoning, this project designs a method to calculate the latent conflict situation, proposes the latent conflicts scission mechanism under certain conflict situation, builds timetable adjustment model which takes the latent conflict situation as the optimized object, establishes the feedback optimization theory for high-speed train timetable using Bayesian inference of latent conflict. This project explores a new train timetabling methods and processes optimization for Chinese high-speed rail based on the management perspectives of reducing the complexity of train dispatching and guaranteeing the transportation service quality, which will be developments and innovations of theory for high-speed train timetabling. And it can also provide certain theoretical support for daily train dispatching and train operation adjustment.

英文关键词： High-speed rail;Train timetabling;Latent confilict;Bayesian inference;Feedback optimization