农业拖拉机焊接结构振动疲劳寿命预测模型研究  

Research on Vibration Fatigue Life Prediction Model of Agricultural Tractor Welded Structure

作　　者：武守辉 陈思杰[2] Wu Shouhui;Chen Sijie(Huanghe Jiaotong University,Jiaozuo 454000,China;Henan Polytechnic University,Jiaozuo 454150,China)

机构地区：[1]黄河交通学院,河南焦作454000 [2]河南理工大学,河南焦作454150

出　　处：《农机化研究》2025年第5期215-219,共5页Journal of Agricultural Mechanization Research

基　　金：河南省科技攻关项目(142102210434);河南理工大学金属学科基金项目(660642/002,036)。

摘　　要：焊接结构广泛应用于实际工程中,常常承受着复杂的随机动载荷,当动载荷频率与结构固有模态频率接近时,接头会发生疲劳破坏,而传统的单轴静载疲劳分析方法无法准确预测这种情况下的疲劳寿命。因此,需要采用更加精细和准确的方法来分析焊接结构的疲劳寿命,以保证结构的安全性和可靠性。为此,基于统计学和机器学习方法构建了预测模型,包括基于逐步回归的多元线性回归模型、基于支持向量机的非线性回归模型和基于神经网络的深度学习模型。同时,以农业拖拉机为研究目标,对比分析了3种模型的预测性能,结果表明:基于神经网络的深度学习模型具有最佳的预测精度和稳定性。Welded structures are widely used in practical engineering,but are often subjected to complex random dynamic loads during operation.When the dynamic load frequency is close to the intrinsic modal frequency of the structure,fatigue damage occurs in the joint.The traditional uniaxial static load fatigue analysis method cannot accurately predict the fatigue life in this case.Therefore,a more refined and accurate method is needed to analyze the fatigue life of welded structures to ensure the safety and reliability of the structure.For this,prediction models were constructed based on statistical and machine learning methods,including a multiple linear regression model based on stepwise regression,a nonlinear regression model based on support vector machines and a deep learning model based on neural networks.Taking agricultural tractors as the research target,the prediction performance of the three models was compared and analyzed,and the results showed that the deep learning model based on neural networks has the best prediction accuracy and stability.

关 键 词：拖拉机 焊接结构 振动疲劳 寿命预测 机器学习 深度学习 

分 类 号：S219.03[农业科学—农业机械化工程]