Abstract: Due to the characteristics of hybrid electric vehicle, its design, development and testing are relatively complicated. The application of some virtual simulation methods can effectively improve the control strategy and verify the control unit and save the cost of real vehicle test. In this paper, a set of high-precision engine and vehicle physical model suitable for the study of emission characteristics of hybrid electric vehicles is built with Cruise M. The emission model is established by using the bench and vehicle emission test data through the neural network algorithm, and the hardware in the loop simulation platform is used for real-time calculation, and the calculated results are compared with the real vehicle test data, and the results are verified by the real vehicle.

Key words: Hybrid electric vehicle; Neural network; Emission model; Hardware in the loop

摘要： 由于混合动力汽车自身的特点，其设计、开发和测试相对比较复杂，应用一些虚拟仿 真的方法可以有效改善控制策略和验证控制单元并节约实车试验的成本。文章首先利用 Cruise M 搭建了一套适用于混合动力汽车排放特性研究的高精度的发动机以及车辆物理模 型，利用台架以及整车排放测试数据，通过神经网络算法建立排放数学模型，在硬件在环仿 真平台集成后进行实时计算，并将计算结果与实车测试数据对比，结果经实车验证可靠。

关键词: 混合动力汽车；神经网络；排放模型；硬件在环