基于PSO-SVR的重型柴油车NOx排放预测-《内燃机学报》

文章信息/Info

Title:: Prediction of NOx Emissions of a Heavy-Duty Diesel Vehicle Based on PSO-SVR

作者:: 王志红; 董梦龙; 张远军; 胡杰; 1. 武汉理工大学汽车工程学院，湖北武汉 430070；2. 襄阳达安汽车检测中心有限公司，湖北襄阳 441004

Author(s):: Wang Zhihong; Dong Menglong; Zhang Yuanjun; Hu Jie; 1. School of Automotive Engineering，Wuhan University of Technology，Wuhan 430070，China； 2. Xiang Yang DA’AN Automobile Testing Center Limited，Xiangyang 441004，China

Keywords:: heavy-duty diesel vehicle; portable emission measurement system; principal component analysis; particle swarm optimization(PSO); support vector regression(SVR)

摘要:: 结合重型汽车国Ⅵ污染物排放法规，采用车载便携式排放测试设备(PEMS)进行了某重型柴油车实际道路排放测试．对测试数据进行数据对齐，剔除无效数据后，采用灰色关联分析提取了对NOx排放影响较大的参数，引入主成分分析(PCA)对输入数据进行降维，引入非线性递减惯性权重粒子群算法(PSO)对支持向量回归(SVR)模型进行优化，最终得到重型柴油车实际道路NOx排放预测模型，测试集均方根误差(RMSE)为1.3816mg/s，平均绝对百分比误差(MAPE)为19.88%，决定系数R2为0.9081．该研究为车载NOx传感器故障诊断以及重型车NOx排放在线监管提供一种可能性方法．

Abstract:: Followed with China Ⅵ emission regulations for heavy-duty vehicles，a portable emission measurement system(PEMS) was used to measure the real-world emissions of a heavy-duty diesel vehicle. After aligning the test data and removing the invalid data，the parameters with great impact on NOx emissions were extracted by gray correlation analysis. Then，principal component analysis(PCA) was introduced to reduce the dimension of input data，and particle swarm optimization(PSO) was introduced to optimize the support vector regression(SVR) model. Finally，a real-world NOx emission prediction model of the heavy-duty diesel vehicle was obtained. The results show that the root mean square error(RMSE) of the test dataset is 1.3816mg/s，the mean absolute percentage error(MAPE) is 19.88% and the R2 is 0.9081. This study provides a possible method for on-board NOx sensor fault diagnosis and on-line NOx emission monitoring for heavy-duty diesel vehicles.