广西科技大学学报2025，Vol.36Issue(5)：1-9,9.DOI:10.16375/j.cnki.cn45-1395/t.2025.05.001
基于CPO-BP神经网络的双燃料发动机性能预测模型研究
Performance prediction model for dual-fuel engines based on CPO-BP neural network
摘要
Abstract
To address the challenges of high calibration complexity and cost in traditional methods for natural gas-diesel dual-fuel engines,a high-precision performance prediction model was constructed to provide technical support for engine control parameter optimization,energy conservation and emission reduction.Based on GT-Power,a one-dimensional simulation model of YC6K dual-fuel engine was constructed,which contained core modules such as intake and exhaust system,in-cylinder working process and Wiebe combustion model.The reliability of the model was verified by the external characteristics(maximum error 4.2%)and propulsion characteristics(cylinder pressure curve error≤4.0%).The Sobol space-filling test design method was used to extract 260 sample points in the 6-dimensional parameter space(rotational speed,pre-injection timing,main injection timing,rail pressure,air-fuel ratio and pre-injection fuel quantity).The CPO-BP neural network fusion algorithm was used to combine the crown porcupine optimisation(CPO)algorithm with the BP neural network.By dynamically adjusting weights and biases via the CPO algorithm,the proposed method mitigates the tendency of conventional BP networks to fall into local optima,thereby enabling accurate performance prediction for dual-fuel engines.关键词
GT-Power/双燃料发动机/Sobol空间填充试验设计/CPO-BP神经网络Key words
GT-Power/dual-fuel engine/Sobol space-filling experimental design/CPO-BP neural network分类
能源与动力引用本文复制引用
张清,王闯,莫清烈,李哲骏,姜峰..基于CPO-BP神经网络的双燃料发动机性能预测模型研究[J].广西科技大学学报,2025,36(5):1-9,9.基金项目
广西重点研发计划项目(桂科AB24010298、桂科AB24010293、桂科AB25069449) （桂科AB24010298、桂科AB24010293、桂科AB25069449）
广西科技重大专项项目(桂科AA24206064)资助 （桂科AA24206064）
