基于机器学习的拖拉机发动机扭矩的预测模型  

Prediction Model of Tractor Engine Torque Based on Machine Learning

作　　者：罗森侨 王意东 王小龙 Luo Senqiao;Wang Yidong;Wang Xiaolong(Sichuan City Vocational College,Chengdu 610000,China;School of Automobile and Transportation,Xihua University,Chengdu 610000,China;School of Mechanical Engineering,Chengdu University,Chengdu 610000,China)

机构地区：[1]四川城市职业学院,成都610000 [2]西华大学汽车与交通学院,成都610000 [3]成都大学机械工程学院,成都610000

出　　处：《农机化研究》2023年第3期254-259,共6页Journal of Agricultural Mechanization Research

基　　金：四川省重大科技专项(2019YFS0499);四川省科技计划项目(2017KZ0075);四川城市职业学院2021年度科研重点项目(CS21KYZD01)。

摘　　要：拖拉机发动机扭矩实时测量需要进行大量的工作与昂贵的传感器,在农业生产中投入成本较高,不易进行广泛的推广使用。基于以上问题,采用了径向基函数(RBF)神经网络进行拖拉机发动机扭矩的预测,通过13种训练算法训练RBF神经网络模型,明确RBF神经网络模型基本结构与参数设置,并通过误差分析模型的预测性能。研究结果表明:RBF神经网络中R^(2)=0.99,RMSE=0.5,EF=0.99,预测值与试验值之间的拟合程度价高,可用于估算发动机扭矩。研究结果对于提高拖拉机发动机扭矩的预测和分析提供理论支撑与科学依据。The real-time measurement of tractor engine torque requires a lot of work and expensive sensors,and the input cost is high in agricultural production,so it is not easy to be widely used.Based on the above problems,the radial basis function(RBF) neural network is used in this study to predict the torque of the tractor engine.The RBF neural network model is trained through 13 training algorithms,and the basic structure and parameter settings of the RBF neural network model are clarified.Through the error analysis model the research results show that R^(2)=0.99,RMSE=0.5,EF=0.99 in the RBF neural network,the fitting degree between the predicted value and the experimental value is high,so it can be used to estimate the engine torque.The research results provide theoretical support and scientific basis for the prediction and analysis of improving tractor engine torque.

关 键 词：机器学习 RBF神经网络 拖拉机扭矩 预测模型 

分 类 号：S219.031[农业科学—农业机械化工程]