天然气发动机燃烧放热的预测研究  

Prediction of the combustion heat release in a natural gas engine

作　　者：崔瀚林 刘庆伟 丁顺良 高建设[1] 宋恩哲 CUI Hanlin;LIU Qingwei;DING Shunliang;GAO Jianshe;SONE Enzhe(School of Mechanical and Power Engineering,Zhengzhou University,Zhengzhou 450001,China;Luoyang Tractor Research Institute Co.,Ltd.,Luoyang 471039,China;Yantai Research Institute,Harbin Engineering University,Yantai 264000,China)

机构地区：[1]郑州大学机械与动力工程学院,郑州450001 [2]洛阳拖拉机研究所有限公司,河南洛阳471039 [3]哈尔滨工程大学烟台研究院,山东烟台264000

出　　处：《重庆理工大学学报(自然科学)》2025年第3期225-234,共10页Journal of Chongqing University of Technology:Natural Science

基　　金：国家自然科学基金项目(51906225);船舶动力工程技术交通运输行业重点实验室(武汉理工大学)开放基金项目(KLMPET2021-05);教育部“春晖计划”合作科研项目(HZKY20220281);中国国家留学基金项目(202308410392)。

摘　　要：为探究低负荷工况下过量空气系数(λ)对天然气发动机燃烧放热规律的影响,在1台电控多点顺序喷射天然气发动机上进行试验并采集数据,结合遗传算法(GA)与反向传播(BP)神经网络,构建了3种燃烧放热的预测模型:传统BP神经网络模型、传统GA优化BP神经网络模型和双阶段遗传算法(double-stage genetic algorithm,DSGA)优化BP神经网络模型,并将各模型的预测值与试验值进行对比。结果表明:3种预测模型均展现出较好的预测性能,相对于传统BP神经网络模型和传统GA优化BP神经网络模型,DSGA优化BP神经网络预测模型对放热率(HRR)、累计放热量(HR)以及燃烧反应百分率(CRR)的预测精度最高且效果最优,在稀燃极限(λ=1.9)工况下,虽然对燃烧循环后期的放热规律预测精度有所降低,但仍可对燃烧循环前期和中期的放热规律准确预测。To investigate the impact of excess air coefficient(λ)on the combustion heat release characteristics of natural gas engine under low load conditions,we conduct experiments on an electronically controlled multi-point sequential injection natural gas engine and collect data.Three combustion heat release prediction models are built by integrating the genetic algorithm(GA)with the backpropagation(BP)neural network:the standard BP neural network model,the GA-optimized BP neural network model,and the double-stage GA-optimized BP neural network model.The predicted outcomes from each model are evaluated and compared with our experimental results.Our findings indicate the three prediction models deliver impressive predictive performances.Compared to the standard BP neural network model and the traditionally GA-optimized BP neural network model,the DSGA-optimized BP neural network prediction model achieves the most accurate prediction for heat release rate(HRR),cumulative heat release(HR),and combustion reaction rate(CRR).Under the lean burn limit(λ=1.9)condition,the DSGA-optimized model still accurately predictes the heat release pattern during the early and middle stages of the combustion cycle,although its prediction accuracy decreases in the later stages.

关 键 词：天然气发动机 燃烧放热 双阶段优化 遗传算法 神经网络 

分 类 号：TK464[动力工程及工程热物理—动力机械及工程]