基于粒子群卷积算法的转杆机械损伤预测研究  

Study on Prediction of Rod Mechanical Damage Based on Particle Population Convolution Algorithm

作　　者：徐立[1] 元春波 徐敏[1] 邵坚铭[1] 冯海[1] 王安琪 XU Li;YUAN Chunbo;XU Min;SHAO Jianming;FENG Hai;WANG Anqi(Ningbo Cigarette Factory,China Tobacco Zhejiang Industrial Co.,Ltd.,Ningbo 315040,China)

机构地区：[1]浙江中烟工业有限责任公司宁波卷烟厂,浙江宁波315040

出　　处：《机械与电子》2024年第10期76-80,共5页Machinery & Electronics

摘　　要：针对通用旋转机械中转杆在实际使用过程中的机械损伤现象,传统的预测方法依赖于经验公式和简单的统计模型,难以准确捕捉转杆在实际工作环境下复杂的应力状态和损伤演化过程,进而这些方法在预测精度上存在局限性,难以实现早期损伤预警和寿命预测。因此,结合卷积神经网络和粒子群算法,提出一种基于粒子群卷积算法下转杆的机械损伤预测。通过与实际的机械损伤对比,可实现机械损伤预测避免网络陷入局部最优,提高了计算效率和预测准确性,满足实际工程中杆机械损伤预测的需求。There are mechanical damage phenomena in the actual use of universal rotating machinery rotating rod.The traditional prediction methods rely on empirical formulas and simple statistical models,and it is difficult to accurately capture the complex stress state and damage evolution process of the rotating rod in the actual working environment.Therefore,these methods have limitations in the prediction accuracy,and it is difficult to achieve early damage warning and life prediction.Combined with convolutional neural network and particle swarm optimization algorithm,a kind of mechanical damage prediction of rotating rod based on particle swarm optimization algorithm was proposed.By comparing with actual mechanical damage,mechanical damage prediction can be realized to avoid the network falling into local optimal,and the calculation efficiency and prediction accuracy are improved,meeting the needs of mechanical damage prediction of rod in practical engineering.

关 键 词：粒子群卷积算法 转杆 机械损伤 预测 

分 类 号：TH117[机械工程—机械设计及理论] TP301[自动化与计算机技术—计算机系统结构]