由于难以将行驶路线地形的影响从实际行驶排放（real driving emission，RDE）试验的其他试验边界的影响中独立出来，提出采用神经网络输入变量重要性算法以定量评估行驶路线地形试验边界对RDE试验的影响强度。以重庆地区RDE试验的37 256个数据窗口排放样本为基础，采用因子分析方法缩减数据并消除试验边界之间的信息重叠，建立神经网络模型预测污染物排放，并计算输入变量相对重要性占比。结果表明，行驶路线地形试验边界在二氧化碳（CO₂）排放中起主导作用，它的相对重要性远大于行程动力学试验边界。对于一氧化碳（CO）、颗粒数量（particle number，PN）、氮氧化物（NO_x）污染物排放，地形因素的影响力仍不可忽视，特别是在车辆高速行驶条件下，它对车辆行驶排放的影响与行程动力学因素大致相当。总体而言，在现有排放标准体系中，行驶路线地形试验边界对RDE试验的影响被严重低估。

It is difficult to separate the effect of route topography from that of other test boundaries in real driving emission (RDE) tests. We proposed an artificial neural network (ANN) weight method to quantitatively evaluate the impact of route topography on RDE tests. Based on 37 256 data window samples of RDE tests in Chongqing, a factor analysis method was used to reduce data and eliminate information overlap between test boundaries. Additionally, a neural network model was also established to predict pollutant emissions and calculate the relative importance of input variables. The results show that route topography significantly affects CO₂ emissions, with its relative importance far exceeding that of other test boundaries. Moreover, the influence of the route topography cannot be ignored for CO, PN (particle number), and NO_x emissions, having an impact on vehicle driving emissions comparable to that of trip dynamics, especially under high-speed driving conditions. However, the existing regulatory emission standards seriously underestimate the impact of the route topography on vehicle driving emissions.

U448.213

国家重点研发计划资助项目（2018YFB0106404）；重庆市技术创新与应用发展专项项目（cstc2019jscx-msxmX0016）；通用技术中国汽研检测事业部创新课题项目（JCCXKT-2021-002）。