一种智慧地铁轨道状态预测和维修决策优化系统

为提高地铁轨道智能化管理水平，设计了一种新的智慧地铁轨道管理系统；分别设计了朴素贝叶斯分类器、Logistic回归分类器和支持向量机分类器，并构建了基于Stacking集成算法的轨道状态预测模型；利用XX地铁1号线、2号线和6号线在2015-2021年的设备数据、检测数据和维修数据，验证了模型有效性；引入自适应学习的马尔可夫决策过程(AL-MDP),构建了基于Stacking-SVM的轨道维修决策优化模型；利用模型对地铁轨道进行运行状态感知和异常状态预测，通过自适应学习的过程不断学习地铁轨道的劣化机理，并且为轨道的状态预警和维修策略提供个性化、精细化的决策建议；最后，设计并实现了智慧地铁轨道管理系统；引入AL-MDP后进一步降低地铁轨道的维修成本，能够实时掌握地铁轨道的运作状态；该研究给管理者和工作者提供精细化、个性化、更科学的维修优化决策，对维修成本和轨道安全实现双重精准控制。

A Intelligent Subway Track State Prediction and Maintenance Decision-Making Optimization System

Improve the intelligent management level of subway track, a new intelligent subway track management system was designed. Naive Bayes Classifier, Logistic Regression Classifier and Support Vector Machine Classifier were designed respectively, and a track state prediction model based on Stacking ensemble algorithm was constructed. Using the equipment data, inspection data and maintenance data of XX Metro Line 1, 2 and 6 from 2015 to 2021, the validity of the model was verified. Furthermore, an Adaptive Learning Markov Decision Process (AL-MDP) was introduced, and an optimization model of track maintenance decision based on Stacking-SVM was constructed. The model was used to sense the running state of the subway track and predict the abnormal state, and continuously learned the deterioration mechanism of the subway track through the process of adaptive learning, and provided personalized and refined decision-making suggestions for the state warning and maintenance strategy of the track. Finally, the intelligent subway track management system was designed and implemented. After the introduction of AL-MDP, the maintenance cost of the subway track was further reduced, and the operation status of the subway track can be grasped in real time. This research provides managers and workers with refined, personalized and more scientific maintenance optimization decisions, and achieves dual precise control of maintenance costs and track safety.