基于耕作过程下拖拉机发动机扭矩预测模型的研究  

Prediction Model of Tractor Engine Torque Based on Tillage Process

作　　者：王亮 Wang Liang(Zhoukou Vocational and Technical College,Zhoukou 466000,China)

机构地区：[1]周口职业技术学院,河南周口466000

出　　处：《农机化研究》2025年第5期220-225,共6页Journal of Agricultural Mechanization Research

基　　金：河南省科学技术厅科学研究重点项目(20B413011)。

摘　　要：拖拉机发动机扭矩是评价变速器疲劳寿命分析、优化设计和加速寿命测试的重要指标,而现有的拖拉机发动机扭矩测量方法需要一个昂贵的扭矩传感器。为此,提出一种基于粒子群优化(PSO)和极限学习机(ELM)的拖拉机发动机扭矩预测模型。首先,采集了拖拉机发动机的试验数据,并将其分为训练集和测试集;然后,使用PSO算法优化ELM的参数,以获得更好的预测性能;最后,将优化后的ELM模型与其他机器学习模型进行比较,并使用误差指标来评估预测性能。试验结果表明:所提出的PSO-ELM模型具有更好的预测精度和泛化能力,R2达到0.9342,MAPE仅为0.95%,RMSE为2.10,预测值与试验值之间拟合程度较高,可以有效地用于拖拉机发动机扭矩的预测。Tractor engine torque is an important indicator for evaluating transmission fatigue life analysis,optimal design and accelerated life testing.Existing tractor engine torque measurement methods require an expensive torque sensor;therefore,proposed a tractor engine torque prediction model based on Particle Swarm Optimization(PSO)and Extreme Learning Machine(ELM).Firstly,experimental data of tractor engines were collected and divided into a training set and a test set.Then,the parameters of the ELM were optimized using the PSO algorithm to obtain better prediction performance.Finally,the optimized ELM model was compared with other machine learning models and the error metrics were used to evaluate the prediction performance.The experimental results showed that the proposed PSO-ELM model had better prediction accuracy and generalization ability,with R 2 reaching 0.9342,MAPE only 0.95%,RMSE 2.10,and a good fit between predicted and tested values,which can be effectively used for tractor engine torque prediction.

关 键 词：拖拉机 发动机扭矩 耕作过程 预测模型 机器学习 

分 类 号：S219.031[农业科学—农业机械化工程]