上海交通大学学报 ›› 2022, Vol. 56 ›› Issue (5): 604-610.doi: 10.16183/j.cnki.jsjtu.2021.231

• 机械与动力工程 • 上一篇    下一篇

基于外推高斯过程回归方法的发动机排放预测

王子垚, 郭凤祥, 陈俐()   

  1. 上海交通大学 动力装置与自动化研究所; 海洋工程国家重点实验室,上海 200240
  • 收稿日期:2021-06-29 出版日期:2022-05-28 发布日期:2022-06-07
  • 通讯作者: 陈俐 E-mail:li.h.chen@sjtu.edu.cn
  • 作者简介:王子垚(1996-),男,河南省郑州市人,硕士生,主要从事发动机排放预测、高斯过程回归等研究.

Engine Emission Prediction Based on Extrapolated Gaussian Process Regression Method

WANG Ziyao, GUO Fengxiang, CHEN Li()   

  1. Institute of Marine Power Plant and Automation; State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
  • Received:2021-06-29 Online:2022-05-28 Published:2022-06-07
  • Contact: CHEN Li E-mail:li.h.chen@sjtu.edu.cn

摘要:

为了提高训练集范围之外行驶工况的预测精度,提出外推高斯过程回归(GPR)方法.首先,采用训练集数据对GPR模型进行预训练,然后在正负3个标准差之间均匀采样构建宽域输入集,以该输入集的预测方差均值最小为目标优化GPR模型超参数.某直喷汽油机转毂试验的结果表明,外推GPR的平均绝对误差为0.53411,比传统GPR降低24.27%,比反向传播神经网络降低36.32%.所提方法可为降低试验成本,提高实际行驶过程排放预测精度提供参考.

关键词: 发动机, 排放, 高斯过程回归, 预测, 外推

Abstract:

Aimed at improving the prediction accuracy of engine emissions under driving conditions which are not covered by the training set, an extrapolated Gaussian process regression (GPR) method is proposed. First, the training set data is fed to the GPR model for pre-training, and then a wide-area input set is constructed by uniform sampling between plus/minus three standard deviations, and the hyperparameters are optimized for the goal of minimizing the prediction variance of the input set. The test results on a direct injection gasoline engine show that the mean absolute error of emission prediction using the extrapolated GPR is 0.53411, which is 24.27% lower than that of using the traditional GPR and 36.32% lower than that of using the back propagation (BP) neural network, which signifies the effectiveness of the proposed method in terms of reducing test costs and improving the accuracy of emission prediction during real driving.

Key words: engine, emission, Gaussian process regression (GPR), prediction, extrapolation

中图分类号: