针对柴油机获取故障数据较难的问题，利用AVL-BOOST模块化的方法建立柴油机气缸模型、中冷器模型等各系统模型，搭建柴油机的整机模型，采用额定工况数据验证了模型精度较高。模型模拟喷油提前、排气滞后等7种单一故障和3种复合故障，分析不同故障引起的参数变化，选取了油耗、有效功率、最大燃烧压力等12种参数作为特征参数，构造故障样本集，为故障状态检测提供了数据支撑。为实现柴油机运行状态精准检测，采用变分自编码器模型（VAE）对柴油机进行故障检测，利用正常数据确定统计量阈值判断是否发生故障，实验结果表明，对单一故障和复合故障2种类型检测都取得了较好的效果。
In addressing the challenge of acquiring fault data for diesel engines, AVL-BOOST employs a modularized approach to establish models of the diesel engine cylinder, intercooler and other systems. This method is then used to construct a comprehensive diesel engine model, with the model's accuracy being verified using data from rated working conditions. The model simulates seven kinds of single faults and three kinds of compound faults, such as injection advance and exhaust lag. It then analyses the parameter changes caused by different faults and selects 12 kinds of parameters, such as fuel consumption, effective power, and maximum combustion pressure, as the characteristic parameters. Finally, it constructs a sample set of faults, which provides data support for the detection of the fault state. In order to achieve accurate detection of the operational state of a diesel engine, the Variational Autoencoder (VAE) model is employed for fault detection. The normal data is utilized to establish the statistical threshold, which is then employed to determine whether a fault has occurred. The experimental results demonstrate that enhanced outcomes have been attained for the detection of both single and composite faults.
2026,48(4): 63-69 收稿日期：2025-6-4
DOI：10.3404/j.issn.1672-7649.2026.04.010
分类号：U664
基金项目：国家重点研发计划项目（2019YFE0104600）；国家自然科学基金资助项目（51909200）
作者简介：邹思涵(2000-),男,硕士研究生,研究方向为动力机械建模仿真与监测诊断
参考文献：
[1] 张天野. 船舶柴油机[M]. 哈尔滨: 哈尔滨工业大学出版社, 2017.
[2] LIM J, KIRCHEN P, NAGAMUNE R. LPV controller design for diesel engine SCR aftertreatment systems based on Quasi-LPV models[J]. IEEE Control Systems Letters, 2021 (5): 1807-1812.
[3] PRATAP SINGH A, KUMAR V, KUMAR AGARWAL A. Evaluation of reactivity controlled compression ignition mode combustion engine using mineral diesel/gasoline fuel pair[J]. Fuel, 2021, 301: 120986.
[4] JIANG F, CAO W, TAN X, et al. Optimization analysis of locomotive diesel engine intake system based on matlab-simulink and GT-power[J]. Processes, 2022, 10(1): 157.
[5] BELLÉR G, ARPAD I, KISS J T, et al. AVL Boost: a powerful tool for research and education[J]. Journal of Physics: Conference Series, IOP Publishing, 2021 (1): 12–15.
[6] 赵志强, 张嘉锐, 罗炽恒, 等. 基于AVL BOOST的船用柴油机典型故障仿真及其数据分析[J]. 中国舰船研究, 2021, 16(2): 176-181.
ZHAO Z Q, ZHANG J R, LUO C H, et al. AVL BOOST-based typical fault simulation and data analysis of marine dies el engine[J]. Chinese Journal of Ship Research, 2021, 16(2): 176-181.
[7] 闫浩栋. 基于改进多模型的船用柴油机故障诊断[D]. 哈尔滨: 哈尔滨工业大学, 2022.
[8] NADERI E, MESKIN N, KHORASANI K. Nonlinear fault diagnosis of jet engines by using a multiple model–based approach[J]. 2012, 134(1): 254–262.
[9] 戴琳. 基于热力学原理的智能低速机故障模拟研究[D]. 哈尔滨: 哈尔滨工程大学, 2020.
[10] 王英敏. 基于融合模型的柴油机空气管理系统故障诊断[J]. 科学技术与工程, 2020, 20(25): 10280-10286.
WANG Y M. Fault diagnosis of diesel engine air management system based on fusion model[J]. Science Technology and Engineering, 2020, 20(25): 10280-10286.
[11] 白雲杰, 贾希胜, 梁庆海, 等. 基于深度学习的柴油机气门健康状态评估[J]. 科学技术与工程, 2022, 22(10): 3941-3950.
BAI Y J, JIA X S, LIANG Q H, et al. Evaluation of diesel engine valve health status based on deep learning[J]. Science Technology and Engineering, 2022, 22(10): 3941-3950.
[12] 刘伊凡, 黄连忠, 孙培廷, 等. 实船数据环境下船舶主柴油机性能评估方法[J]. 内燃机学报, 2018, 36(2): 182-191.
LIU Y F, HUANG L Z, SUN P Y, et al. A performance estimation method for main diesel engine in shipping data environment[J]. Transactions of CSICE, 2018, 36(2): 182-191.
[13] KINGMA D P, WELLING M. Auto–encoding variational bayes[J]. arXiv preprint arXiv: 1312.6114, 2013.
[14] CAI Z, CAI Z. Pevae: A hierarchical vae for personalized explainable recommendation[C]//Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval. Madrid: ACM, 2022.
[15] YAO Y, YANG Y, SHANG Q et al. Research on fault diagnosis of marine diesel engine based on DA-DBN[C]//2024 International Conference on Energy and Electrical Engineering (EEE), China, 2024.
