基于动态改进遗传粒子群--BP的重型车NOx排放预测模型研究

钱枫; 马骋; 祝能; 王明达; 王继广; 许小伟

车用发动机 ›› 2023, Vol. 0 ›› Issue (5) : 63-71.

栏目

基于动态改进遗传粒子群--BP的重型车NO_x排放预测模型研究

钱枫¹，马骋¹，祝能¹，王明达²，王继广³，许小伟¹

作者信息 +

NO_x Emission Prediction Model of Heavy-Duty Vehicle Based on Dynamic Improved Genetic Particle Swarm-BP Network

QIAN Feng¹,MA Cheng¹,ZHU Neng¹,WANG Mingda²,WANG Jiguang³,XU Xiaowei¹

Author information +

文章历史 +

摘要

为了降低重型车NO_x排放速率监测控制中OBD设备异常采集数据和数据耦合问题的影响，基于BP神经网络建立了排放预测模型。为了提高预测模型的准确性，引入了遗传粒子群组合算法，并对其进行动态改进，同时利用PCA分析提取数据特征。结果表明：对比传统遗传算法和粒子群算法，动态改进的遗传粒子群组合算法在适应度函数上提升了5.75%和3.37%；与其他9种预测模型相比，动态改进后的遗传粒子群-BP网络在评价指标MASE、RMSE和R²上表现最优，MASE、RMSE分别为0.024和0.033 6，R²为0.951，预测结果与原始数据基本吻合，所建预测模型具有较高的预测准确性。

Abstract

In order to reduce the influence of abnormal data acquisition and data coupling of OBD equipment in heavyduty vehicle NO_x emission rate monitoring and control, an emission prediction model was established based on BP neural network. In order to improve the accuracy of prediction model, the genetic particle swarm combination algorithm was introduced and dynamically improved. At the same time, PCA analysis was used to extract data features. The results show that the dynamic improved genetic particle swarm combination algorithm improves the fitness function by 5.75% and 3.37% compared with the traditional genetic algorithm and particle swarm optimization algorithm. Compared with the other nine prediction models, the dynamic improved genetic particle swarm-BP network performs best on the evaluation indexes MASE, RMSE and R², the first two are 0.024 and 0.033 6 respectively and the latter is 0.951. The prediction results are basically consistent with the original data, and the prediction model has higher prediction accuracy.

导出引用

钱枫，马骋，祝能，王明达，王继广，许小伟. 基于动态改进遗传粒子群--BP的重型车NO_x排放预测模型研究[J]. 车用发动机. 2023, 0(5): 63-71

QIAN Feng,MA Cheng,ZHU Neng,WANG Mingda,WANG Jiguang,XU Xiaowei. NO_x Emission Prediction Model of Heavy-Duty Vehicle Based on Dynamic Improved Genetic Particle Swarm-BP Network[J]. Vehicle Engine. 2023, 0(5): 63-71