|本期目录/Table of Contents|

[1]梅再武,周江嫚,陈绪兵*,等.融合多领域建模与强化学习的进给系统设计与分析[J].武汉工程大学学报,2025,47(04):434-441.[doi:10.19843/j.cnki.CN42-1779/TQ.202406001]
　MEI Zaiwu,ZHOU Jiangman,CHEN Xubing*,et al.Design and analysis of feed system integrating multi-domain modeling and reinforcement learning[J].Journal of Wuhan Institute of Technology,2025,47(04):434-441.[doi:10.19843/j.cnki.CN42-1779/TQ.202406001]
点击复制

融合多领域建模与强化学习的进给系统设计与分析
(/HTML)
分享到：

《武汉工程大学学报》[ISSN:1674-2869/CN:42-1779/TQ]

卷:
47
期数:
2025年04期
页码:
434-441
栏目:
智能制造
出版日期:
2025-08-29

文章信息/Info

Title:
Design and analysis of feed system integrating multi-domain modeling and reinforcement learning
文章编号:
1674 - 2869（2025）04- 0434 - 08
作者:
1. 武汉工程大学机电工程学院，湖北 武汉 430205；
2. 智能焊接装备与软件工程技术湖北省研究中心， 湖北 武汉 430205；
3. 武汉烽火富华电气责任有限公司， 湖北 武汉 430205
Author(s):
1. School of Mechanical & Electrical Engineering, Wuhan Institute of Technology, Wuhan 430205, China;
2. Hubei Research Center of Intelligent Welding Equipment and Software Engineering Technology, Wuhan 430205, China;
3. Wuhan Fenghuo Fuhua Electric Co., Ltd, Wuhan 430205, China
关键词:
Keywords:
分类号:
TH113；TP13
DOI:
10.19843/j.cnki.CN42-1779/TQ.202406001
文献标志码:
A
摘要:
针对滚珠丝杠进给系统的多领域耦合特性，以及系统综合设计与性能优化难点，提出了一种融合多领域建模与强化学习的进给系统设计与分析方法。通过深入的数学分析，基于Modelica语言实现了系统各部分的精准建模，并将子模型进行无缝集成，构建了一个全面反映该系统多领域耦合特性的综合模型。实验结果表明，该模型预测的跟踪误差与实验数据之间的最大偏差不超过6.8 μm，模型精准性良好。在此精准模型基础上，提出了一种基于强化学习的控制参数整定方法。该方法利用自学习机制，以系统控制性能为优化目标，通过大批量虚拟仿真，能够自主发掘出最优控制参数组合。本研究为滚珠丝杠进给系统的综合设计和分析提供了一种新的视角。
Abstract:
To address the multi-domain coupling characteristics of the ball screw feed system and the challenges in its design and performance optimization, a novel approach integrating multi-domain modeling with reinforcement learning for the feed system design and analysis was proposed. Through in-depth mathematical analyses, precise modeling of each component of the system mentioned was accomplished using the Modelica language, and the sub-models were seamlessly integrated to construct a comprehensive model that fully reflects its multi-domain coupling characteristics. Experimental results indicated that the maximum deviation between the tracking error predicted by the model and the experimental data does not exceed 6.8 mm, demonstrating good accuracy of the model. Based on this precise model, a control parameter-tuning method based on reinforcement learning was put forward, which utilizes a self-learning mechanism, takes the control performance of the system as the optimization objective and can autonomously find the optimal combination of control parameters through large-scale virtual simulations. The research herein offers a fresh perspective on the integrated design and analysis of ball screw feed systems.

参考文献/References:

[1] LIU C，ZHAO C Y，LIU Z D，et al. Dynamic analysis of ball screw feed system with the effects of excitation amplitude and design parameters[J]. Applied Sciences，2021，11（15）：7070.
[2] XU M T, LI C Y, ZHANG H Z, et al. A comprehensive nonlinear dynamic model for ball screw feed system with rolling joint characteristics[J]. Nonlinear Dynamics, 2021, 106(1): 169-210.
[3] LIU Z D, XU M T, ZHANG H Z, et al. Dynamic analysis and modeling of ball screw feed system with a localized defect on the support bearing[J]. Journal of the Brazilian Society of Mechanical Sciences and Engineering, 2023, 45(7): 367.
[4] TIAN F Q, ZHU J M, LI X R, et al. Modeling and dynamic characteristic analysis of a ball screw feed drive system based on receptance coupling[J]. Tehni[c]ki Vjesnik-Technical Gazette, 2021, 28(5): 1691-1700.
[5] LI C Y, XU M T, SONG W J, et al. A review of static and dynamic analysis of ball screw feed drives, recirculating linear guideway, and ball screw[J]. International Journal of Machine Tools and Manufacture, 2023, 188: 104021.
[6] GAO X S, ZHANG X R, YANG J S, et al. Dynamic modeling and analysis on lateral vibration of ball screw feed system[J]. The International Journal of Advanced Manufacturing Technology, 2023, 124(11/12): 4211-4229.
[7] ZHANG H J, ZHA J, DU C, et al. Variable-coefficient dynamic modeling method for a ball screw feed system in the no-extra-load running state[J]. Processes, 2023, 11(3): 793.
[8] 付振彪,王太勇,张雷,等．滚珠丝杠进给系统动力学建模与动态特性分析[J]. 振动与冲击, 2019, 38(16): 56-63.
[9] 张欣雨, 赵巍, 张雷, 等. 滚珠丝杠进给系统刚柔耦合动力学建模与时变动态特性分析[J]. 机床与液压, 2024, 52(13): 176-188.
[10] 伍盛华. 基于改进LuGre模型的滚珠丝杠进给系统摩擦力补偿方法[D].西安：西安理工大学, 2023.
[11] 陈晔, 程龙, 余嘉，等. 基于机床原生信息的进给系统位置-速度摩擦模型[J]. 机械设计与研究, 2019, 35(4): 173-177.
[12] LIU C C, TSAI M S, CHENG C C. Development of a novel transmission engaging model for characterizing the friction behavior of a feed drive system[J]. Mechanism and Machine Theory, 2019, 134: 425-439.
[13] TAO G, MA X L, LING Y. Optimal and nonlinear decoupling control of systems with sandwiched backlash[J]. Automatica, 2001, 37(2): 165-176.
[14] XI T D, KEHNE S, FUJITA T, et al. Condition monitoring of ball-screw drives based on frequency shift[J]. IEEE/ASME Transactions on Mechatronics, 2020, 25(3): 1211-1219.
[15] 官鹏飞，区均灌，李慧颖.基于机床进给伺服系统中频振动抑制研究[J].机床与液压,2023,51(19):92-98.
[16] 张忠奎，高峰，李艳，等．基于磁流变阻尼器的滚珠丝杠进给系统主动抑振研究[J]．机床与液压，2024，52(5)：169-172.
[17] YANG X J, LU D, LIU H, et al. Electromechanical integrated modeling and analysis for the direct-driven feed system in machine tools[J]. The International Journal of Advanced Manufacturing Technology, 2018, 98(5/6/7/8): 1591-1604.
[18] ZHANG X, ZHANG J, ZHANG W, et al. Integrated modeling and analysis of ball screw feed system and milling process with consideration of multi-excitation effect[J]. Mechanical Systems and Signal Processing, 2018, 98: 484-505.
[19] ANSOATEGUI I, CAMPA F J. Mechatronics of a ball screw drive using an N degrees of freedom dynamic model[J]. The International Journal of Advanced Manufacturing Technology, 2017, 93(1/2/3/4): 1307-1318.
[20] HERFS W, ?ZDEMIR D, LOHSE W, et al. Design of feed drives with object-oriented behavior models[J]. IFAC-PapersOnLine, 2015, 28(1): 268-273.
[21] ?ZDEMIR D, MOTSCHKE T, HERFS W, et al. Modelica library for feed drive systems[C]//Proceedings of the 11th International Modelica Conference. Link?ping: Link?ping University Electronic press, 2015: 117-125.
[22] 吴沁，周顺仟，王星联. 改进粒子群优化滚珠丝杠进给系统BP神经网络PID控制策略研究[J].西安交通大学学报,2024,58(6):24-33.
[23] KASHKI M, ABDEL-MAGID Y L, ABIDO M A. A reinforcement learning automata optimization approach for optimum tuning of PID controller in AVR system[C] / / Advanced Intelligent Computing Theories and Applications. Berlin: Springer, 2008: 684-692.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期：2024-06-02
基金项目：武汉东湖高新区“揭榜挂帅”项目(2024KJB325)
作者简介：梅再武，博士，讲师。Email：meizw@wit.edu.cn
*通信作者：陈绪兵，博士，教授。Email：chenxb@wit.edu.cn
引文格式:梅再武，周江嫚，陈绪兵，等. 融合多领域建模与强化学习的进给系统设计与分析［J］. 武汉工程大学学报，2025，47（4）：434-441.
更新日期/Last Update: 2025-08-29