基于IPSO-GRNN的汽车零部件生产过程碳排放预测  

Carbon-emission prediction of auto parts in production process based on IPSO-GRNN

作　　者：郑辉 郭晓[1,2] 郭子健 郭瀚文[1,2] 刘苑农 ZHENG Hui;CUO Xiao;CUO Zijian;CUO Hanwen;LIU Yuannong(School of Economics and Management,Tianjin University of Science and Technology,Tianjin 300222;Lean Management Research Center,Tianjin University of Science and Technology,Tianjin 300222)

机构地区：[1]天津科技大学经济与管理学院,天津300222 [2]天津科技大学精益管理研究中心,天津300222

出　　处：《机械设计》2023年第11期69-73,共5页Journal of Machine Design

基　　金：天津市教委社会科学重大项目(2022JWZD11)。

摘　　要：为提高汽车零部件生产过程的碳排放预测精度,文中提出了一种改进粒子群优化算法(IPSO)与广义回归神经网络(GRNN)相结合的碳排放预测方法。通过改变惯性权重与自适应变异因子来更新粒子位置和速度,以确定GRNN最优光滑因子δ,并得到模型输入层适应度函数。运用金属带式无极变速器的实际数据对该模型进行验证,同时,将此模型与PSO-GRNN和GRNN模型对比选优。结果表明:文中构建的IPSO-GRNN模型的预测值与实际值的均方根误差RMSE仅为1.5198,平均绝对误差MAE为1.2749,决定系数R2为0.9001,收敛速度和寻优精度明显优于其他模型,可为汽车产业的碳排放预测提供有力支撑。In this article,in order to enhance the accuracy in carbon-emission prediction of auto parts in the production process,efforts are made to introduce a method combining Improved Particle Swarm Optimization(IPSO)with Generalized Regression Neural Network(CRNN).By adjusting the inertia weight and the adaptive mutation factors,the particles'position and velocity are updated to determine GRNN's optimal smoothness factor and work out the fitness function of the model's input layer.The real-world data obtained from a metal-belt CVT is used to validate this model,and it is benchmarked against the PSOGRNN and GRNN models.The results show the IPSO-GRNN model has RMSE of 1.5198,MAE of 1.2749,and R?of 0.9001,thus outperforming other models in terms of convergence speed and optimization accuracy.It provides substantial support for carbon-emission prediction in the auto industry.

关 键 词：零部件生产 碳排放 预测 PSO GRNN 

分 类 号：TG659[金属学及工艺—金属切削加工及机床]