基于深度学习的柴油机故障诊断系统开发  

Development of a Diesel Engine Fault Diagnosis System Based on Deep Learning

作　　者：张益铭 白巴特尔 朱小龙 聂国乐 Zhang Yi-ming;Bai Ba-te-er;Zhu Xiao-long;Nie Guo-le(CATARC Automotive Test Center(Tianjin)Co.,Ltd.,Tianjin 300399,China;State Key Laboratory of Engines,Tianjin University,Tianjin 300072,China)

机构地区：[1]中汽研汽车检验中心(天津)有限公司,天津300399 [2]天津大学内燃机燃烧学国家重点实验室,天津300072

出　　处：《内燃机与配件》2025年第6期5-11,共7页Internal Combustion Engine & Parts

摘　　要：开展高效智能准确的柴油机故障诊断方法研究和系统研制对保证生产安全和能效提升具有重要意义。本文旨在设计并实现一种基于深度学习的柴油机故障诊断系统,以应对柴油机运行状态实时监控与高效维护的迫切需求。针对柴油机振动信号的非线性、非平稳特性,设计了一种基于改进卷积神经网络(CNN)的混合深度学习模型。该模型通过可分离卷积提取振动信号中的局部特征和空间层次信息,并减小模型参数降低过拟合风险的同时也方便部署,再使用全局平均池化层实现对故障特征的有效分类。之后,采用LabVIEW和python进行“端到端”的实时故障诊断系统开发,其功能包括数据采集、状态监测、故障诊断和数据存储等。经验证,该系统能够实现高精度的柴油机故障实时监测及诊断。Research and development of efficient,intelligent,and accurate diesel engine fault diagnosis methods and systems are of significant importance for ensuring production safety and improving energy efficiency.This paper aims to design and implement a deep learning-based diesel engine fault diagnosis system to address the urgent demand for real-time monitoring and efficient maintenance of diesel engine operating conditions.To tackle the nonlinear and non-stationary characteristics of diesel engine vibration signals,a hybrid deep learning model based on an improved Convolutional Neural Network(CNN)is proposed.The model employs separable convolutions to extract local features and spatial hierarchical information from vibration signals,reducing model parameters to mitigate overfitting risks while facilitating deployment.A global average pooling layer is then utilized for effective classification of fault features.Subsequently,a real-time"end-to-end"fault diagnosis system is developed using LabVIEW and Python,incorporating functionalities such as data acquisition,condition monitoring,fault diagnosis,and data storage.Experimental validation demonstrates that the system achieves high-precision real-time monitoring and diagnosis of diesel engine faults.

关 键 词：深度学习 故障诊断 可分离卷积 柴油机 

分 类 号：TK428[动力工程及工程热物理—动力机械及工程]