基于DNN整机建模的滚珠丝杠进给系统关键结合部动态特性参数辨识  

Dynamic characteristics parametric identification of key joint of ball screw feed system based on DNN modelling

作　　者：朱迪 张玮[1] 黄之文[1] 朱坚民[1] ZHU Di;ZHANG Wei;HUANG Zhiwen;ZHU Jianmin(School of Mechanical Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学机械工程学院,上海200093

出　　处：《振动与冲击》2023年第3期243-254,279,共13页Journal of Vibration and Shock

基　　金：国家自然科学基金(51775323);上海市科委科研计划项目(1706052600)。

摘　　要：针对滚珠丝杠进给系统关键结合部动态特性参数的辨识精度不高等问题。提出利用可表征结合部动态特性参数与整机固有频率之间映射关系的深度神经网络(deep neural network, DNN)建立进给系统整机的等效动力学模型;结合进给系统固有频率的DNN预测值与实验模态分析值,采用粒子群优化(particle swarm optimization, PSO)算法对进给系统关键结合部的不同方向的刚度、阻尼参数同时辨识。以自行设计制造的进给系统实验台为实例进行整机建模、实验、参数辨识等分析;最终的辨识结果达到很高精度,说明该方法是可行、有效的。Here, aiming at the problem of lower identification accuracy of dynamic characteristics parameters of key joint of ball screw feed system, a deep neural network(DNN) being able to characterize mapping relation between dynamic characteristics parameters of joint and natural frequencies of the whole feed system was proposed to establish the equivalent dynamic model of the whole feed system. Combining DNN prediction values and experimental modal analysis values for natural frequencies of the feed system, the particle swarm optimization(PSO) algorithm was used to simultaneously identify stiffness and damping parameters in different directions of key joint of the feed system. Taking the self-designed and manufactured feed system test platform as an actual example, whole feed system modeling, experiments, parametric identification were conducted. It was shown that the final identification results can reach very high accuracy;the proposed method is feasible and effective.

关 键 词：滚珠丝杠进给系统 关键结合部 动态特性参数 深度神经网络(DNN) 整机建模 

分 类 号：TH132[机械工程—机械制造及自动化] TG502.1[金属学及工艺—金属切削加工及机床]