基于神经网络联合逆的4WID多电机协调容错控制

The four-wheel-independent-drive(4WID) electric vehicles(EVs) has indubitably become one of the most important trends for EVs. However, it suffers from the problems of coordinated control and reliability, which limit the practical applications. This project aims to improve the operation ability under the faulty condition by incorpotating the merits of the fault-tolerant motor, the multi-motor coordination and the active safety control of the vehicles. Thus, the coordinated fault-tolerant operation of the multi-motor system will be realized. This project will establish the "motor-tire-vehicle" united mathematic model which oriented to the torque coordinated control. Secondly, an integrated observation of the vehicle states will be investigated in which the neural network left inverse soft sensors theory will be adopted. Thirdly, by decoupling the vehicle longitudinal and the lateral dynamics, the multi-motor coordinated fault-tolerant control theory based on the neural network combined inverse will be proposed. This project will also build the faulty detection method and the optimal reduced-loss fault-tolerant control strategy. Finally, the multi-motor drive platform will be developed which can simulate the operation of the vehicles. Meanwhile, the basic problems of science can be clarified and the general rules of the multi-motor coordinated fault-tolerant control can be summarized. The project involves several disciplines such as control engineering, electrical engineering, and vehicle engineering. The completion of the project will lay a foundation for the development of domestic intellectual properties for EV drive system, and it can be applied into other multi-motor drive fields which requires continued operation.

四轮独立驱动（4WID）已成为电动汽车发展的重要方向之一，其驱动系统的协调控制及可靠性问题制约其走向实用。本项目旨在提高4WID电动汽车的带故障运行能力，融合容错电机、多电机协调与车辆主动安全控制，藉此实现多电机系统故障条件下的协调运行。建立面向转矩协调控制的"电机-轮胎-车辆"联合数学模型；研究基于神经网络左逆软测量的车辆状态集成观测方法；剖析车辆横向、纵向之间的动力学耦合特性，提出神经网络联合逆的多电机协调容错控制理论；建立系统故障诊断与最小损耗的容错控制策略；搭建多电机驱动整车模拟实验系统；提炼基础科学问题，探索故障条件下多电机协调容错控制的一般性规律。项目属控制工程与汽车工程交叉学科的应用基础研究，为我国研制开发具有自主知识产权的高可靠性电动汽车驱动系统奠定理论与实验基础，还可以推广到其它对系统连续运行有较高要求的多电机驱动领域。

四轮独立驱动电动汽车(4WID-EV) 中的冗余执行器系统，可用来优化整车动态特性，提高4WID系统带故障运行能力，从而提高4WID-EV的操控性、稳定性、可靠性。本项目从电机及其驱动系统，整车4WID容错协调控制入手，以提高4WID电动汽车的可靠运行，尤其是驱动电机故障下的整车运行的可靠性。. 本项目延续国家自然基金（51077066）的研究，对新型容错电机优化设计进行持续、深入研究，又提出了三种新型永磁容错电机，并总结了此类电机的优化设计准则，提高新能源车辆驱动电机的可靠性。针对新型五相永磁容错电机搭建了驱动控制系统，展开了多相电机的直接转矩控制、容错控制方法研究，提出了多种SVPWM控制方式，提高此类电机的控制性能，以满足电动汽车驱动所需。针对4WID电动汽车整车状态的集成观测、4WID驱动系统的协调运行的可靠性展开研究；采用神经网络联合逆理论与方法，结合其它状态观测理论与方法，实现对车辆运动状态中不直接可测量观测的同时，抑制整车横摆运动与侧向运动之间的耦合关系，改善整车控制性能。并针对4WID电动汽车冗余驱动的特点，从控制分配层中考虑驱动电机、轮胎与路面附着系数以及驱动电机故障等约束条件，建立优化准则，提高4WID的带故障运行能力，提高整车运行的可靠性。搭建轮毂电机驱动的分布式4WID电动汽车实验平台，以dSPACE实时仿真系统构建快速控制原型，对整车状态观测及容错控制进行实验验证。. 项目执行期内，完成的相关成果受到国内外专家的认可。获国家科学技术发明二等奖一项，发表SCI论文22篇，EI论文15篇，获授权发明专利8项。