基于数据驱动预测控制的混合动力汽车转矩协调控制策略研究

Hybrid electric vehicle(HEV) is one of the main orientations of new energy vehicles at present. The powertrain system is the heart of HEV, and its torque coordination control is the most effective way to improve the ride comfort, the dynamic performance and the durability. HEV torque coordination control can be concluded as a class of nonlinear optimization problem with multiple objectives and multiple constraints, which is difficult to be solved by traditional control approaches. Therefore, it is very urgent to seek new theories and methods to resolve it. Starting with the complex nonlinear characteristics of HEV powertrain, torque coordination control will be researched based on data-driven predictive control theory in this subject. Firstly, the data model of HEV powertrain will be built using the subspace identification method of data-driven theory. Secondly, taking the integrated optimum of the ride comfort, the dynamic performance and the durability as the design objective, the torque coordination control strategy will be designed which is based on the predictive control theory considering multiple constraint characteristics. Then, based on the historical objective function benchmark, the torque coordination control strategy will be monitored real-timely and revised in time to ensure the multiple performance indexes of HEV. Finally, the effectiveness of the proposed new theories and approaches will be illustrated on the HEV powertrain experimental platform. This research project will not only play a significant role on the key technology development and the industrialization course of HEV but also promote the related theory and application research of certain related subjects obviously.

混合动力汽车（HEV）是目前新能源汽车发展的主流方向之一。动力总成系统是HEV的心脏，其转矩协调控制是提高HEV舒适性、动力性和耐久性的最有效途径，且可归结为一类多约束的非线性多目标优化问题，常规控制方法难以解决，亟待寻求新理论和新方法予以突破。本项目拟从探究HEV动力总成系统的复杂非线性特性入手，基于数据驱动预测控制理论研究其转矩协调控制问题。首先，采用数据驱动子空间辨识方法建立HEV动力总成系统的数据模型。其次，以舒适性、动力性和耐久性综合最优为目标，并考虑多种约束条件，利用预测控制理论设计转矩协调控制策略。然后，基于历史目标函数基准方法实时监控并及时修正所设计的转矩协调控制策略，以保证HEV的各项性能指标。最后，在所搭建的试验平台上验证新理论方法的有效性。本项目不仅对发展我国HEV关键技术进而推进其产业化具有重要意义，而且对相关学科的理论和应用研究有明显促进作用。

混合动力汽车（HEV）是目前新能源汽车发展的主流方向之一。动力总成系统是HEV的心脏，其转矩协调控制是提高HEV舒适性、动力性和耐久性的最有效途径，且可归结为一类多约束的非线性多目标优化问题，常规控制方法难以解决，亟待寻求新理论和新方法予以突破。按照研究计划，本项目从探究HEV动力总成系统的复杂非线性特性入手，基于数据驱动预测控制理论研究其转矩协调控制问题。首先，采用数据驱动子空间辨识方法建立了HEV动力总成系统的数据模型。其次，以舒适性、动力性和耐久性综合最优为目标，并考虑多种约束条件，利用预测控制理论设计了转矩协调控制策略，在所搭建的试验平台上验证了所提新理论方法的有效性，提高了HEV的模式切换质量。此外，课题组还在原有研究计划基础上，取得了一定的超越：提出了基于模糊滑模控制的HEV模式切换控制策略；提出了基于神经网络的混合动力装载机的自动换挡策略；针对HEV用永磁同步电机的高性能控制进行了深入研究，提出了基于预测控制、无差拍预测控制、滑模控制的永磁同步电机高性能控制策略。本项目不仅对发展我国HEV关键技术进而推进其产业化具有重要意义，而且对相关学科的理论和应用研究有明显促进作用。