基于卷积神经网络的柴油机DPF状态辨识研究  

Study on State Identification of Diesel Engine DPF Based on Convolutional Neural Network

作　　者：程德新 赵树恩[1] 张军 王欣伟 胡超超 CHENG Dexin;ZHAO Shu’en;ZHANG Jun;WANG Xinwei;HU Chaochao(School of Mechatronics and Vehicle Engineering,Chongqing Jiaotong University,Chongqing 400074,China;State Key Laboratory of Engine Reliability,Weifang 261000,China;Weichai Power Co.,Ltd.,Weifang 261000,China)

机构地区：[1]重庆交通大学机电与车辆工程学院,重庆400074 [2]内燃机可靠性国家重点实验室,山东潍坊261000 [3]潍柴动力股份有限公司,山东潍坊261000

出　　处：《客车技术与研究》2022年第2期36-40,共5页Bus & Coach Technology and Research

基　　金：内燃机可靠性国家重点实验室开放基金项目(SKLER-201912)。

摘　　要：针对柴油机颗粒捕集器功能失效的问题,应用GT-SUITE软件对柴油机颗粒捕集器进行仿真分析,研究颗粒捕集器温度、压降和碳烟浓度等与其状态的关联关系,并采用一维卷积神经网络对颗粒捕集器状态特征“自学习”,提高颗粒捕集器状态辨识的准确度。According to the failure problem of the diesel particulate filter function,the authors simulate and analyze the diesel engine DPF through GT-SUITE software and study the relationship between the DPF state with its temperature,pressure drop,and soot concentration.Then,they improve the accuracy of DPF state identification through applying the one-dimensional convolutional neural network self-learning the state features.

关 键 词：柴油机 颗粒捕集器 故障特征 状态辨识 

分 类 号：TK421.5[动力工程及工程热物理—动力机械及工程]