基于SSA-GRNN的汽油机过渡工况进气流量预测研究  

Research on Intake Flow Prediction of Gasoline Engine under Transient Conditions Based on SSA-GRNN

作　　者：陈侗 李岳林[1,2] 张五龙 谢清华 尹钰屹 Chen Tong;Li Yuelin;Zhang Wulong;Xie Qinghua;Yin Yuyi(Key Laboratory of Safety Design and Reliability Technology for Engineering Vehicle of Hunan Province,Changsha 410114;School of automotive and Mechanical Engineering,Changsha University of Science&Technology,Changsha 410114)

机构地区：[1]长沙理工大学,湖南省工程车辆安全性设计与可靠性技术重点实验室,长沙410114 [2]长沙理工大学,汽车与机械工程学院,长沙410114

出　　处：《汽车技术》2024年第12期54-62,共9页Automobile Technology

基　　金：国家自然科学基金项目“汽油机加速瞬态工况燃烧规律及控制方法研究”(51176014)。

摘　　要：针对过渡工况下汽油机进气流量预测难度较高的问题,构建了一种基于麻雀搜索算法(SSA)优化广义回归神经网络(GRNN)的进气流量预测模型。该模型利用SSA算法对GRNN的平滑因子进行寻优辨识,并采用斯皮尔曼法和对比分析法提取模型的特征参数,以达到较好的预测精度和泛化性能。运用过渡工况进气流量样本数据对模型进行训练和预测,结果表明:在加减速工况下,SSA-GRNN模型预测值的平均相对误差均小于1%;相较于BP、RBF和GA-SVR进气流量预测模型,SSA-GRNN模型具有更高的预测精度和泛化性能,更加适用于汽油机过渡工况进气流量的预测。To address the challenge of predicting gasoline engine intake flow under transition conditions,a prediction model based on a Sparrow Search Algorithm(SSA)optimized Generalized Regression Neural Network(GRNN)is developed.The model employs the SSA algorithm to optimize the smoothing factor of GRNN by identifying the best value and extracting feature parameters using the Spearman method and comparison analysis method to obtain improved prediction accuracy and generalization performance.The model is trained and tested using sample data of intake flow under transition conditions.The results show that the average relative errors of the predicted values of the SSA-GRNN model for acceleration and deceleration conditions are less than 1%.Compared with BP,RBF and GA-SVR intake flow prediction models,the SSA-GRNN model demonstrates higher prediction accuracy and generalization performance,making it more suitable for predicting gasoline engine intake flow under transition conditions.

关 键 词：汽油机 麻雀搜索算法 寻优辨识 广义回归神经网络 进气流量 过渡工况 

分 类 号：U464.4[机械工程—车辆工程]