地铁具有重复运行模式、精确动力学模型难于建立、运行大数据可获取等突出特点。本项目将以数据驱动控制、学习控制与滤波理论为基础，探索先进、实用、数据和模式能被充分利用，且能满足地铁列车运行安全、高效、节能、舒适和准确等实际性能要求的，地铁列车无人智能驾驶运行控制基础理论与技术。具体讲，就是从单/多地铁的实际性能要求出发，以数据驱动迭代学习控制（ILC）、数据驱动无模型自适应控制(MFAC)、增强学习与深度学习、以及多源信息融合和卡尔曼一致性滤波等方法为基础，研究和设计出一套可直接应用于地铁列车运行的数据驱动无人智能驾驶控制新技术和新方法，为我国地铁列车运行的实际应用提供理论和技术支撑。同时在学术上给出适用于较一般未知非线性动态系统的数据驱动控制、数据驱动滤波估计、及前馈与反馈能优势互补的新方法和新策略。

Subway train operations has following outstanding characteristics, the train operates in repetive pattern according to train timetable everyday, the accurate train dynamics model is hard to be established and the big data of train operations is available. This proposal will target to develop the novel and applicable basic theories and technologies on the unmanned intelligent operation control methods for the subway train satisfying the practical critical performance requirements on the safety, efficiency, energy-efficient, comfort and accuracy, by making fully use of the operating big data and the repetitive pattern, based on the data driven control, iterative learning control and filtering estimation theories. Speaking specifically, for the single train and multiple trains with these practical performance requirements, this project will propose a series of novel data driven unmanned intelligent control methods for the subway trains based on data driven iterative learning control, data driven model free adaptive control, reinforcement learning and deep learning, multi-source information fusion and Kalman filtering theory, etc. which will provide theoretical innovation and technical support for the Chinese urban subway systems, and meanwhile, to propose a a series of new academic results on the data driven control strateges, the data driven filtering methods, and the complementarily modular approaches with feedforward add-on-to the feedback control.