CASIA OpenIR  > 毕业生  > 博士学位论文
基于鲁棒自适应迭代学习的注塑控制策略研究
其他题名Strategies of Injection Molding Control Based on Robust and Adaptive Iterative Learning
杨雁
学位类型工学博士
导师王云宽 ; 徐波
2008-01-08
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业控制理论与控制工程
关键词迭代学习 鲁棒 自适应 注塑成型 注塑成型 Iterative Learning Control Robust Adaptive Injection Molding Injection Molding Machine Control System
摘要目前,我国注塑机控制系统与国外同类产品在控制水平上存在较大差距,因此提高注塑控制技术水平,研制基于先进控制技术的高端注塑机控制系统具有重大的现实意义和市场前景。本文介绍了自主研制的注塑机通用控制系统,并以此为平台,对鲁棒自适应迭代学习控制理论及其在注塑工艺控制中的应用展开研究。首先,本文介绍了我国注塑机市场的现状及发展趋势,并对注塑控制技术的现状和发展进行了总结和讨论。针对迭代学习控制在注塑成型过程控制中的应用,重点对迭代学习控制理论的研究发展进行了综述。其次,系统分析了常规迭代学习控制算法的暂态学习特性,介绍了在工程中常用几种增强迭代学习算法鲁棒性的方法。并从时域和频域的角度分别对离散滤波器型迭代学习控制算法进行了系统的鲁棒稳定性分析。针对滤波器型迭代学习控制参数设计问题,提出了一种可变学习增益的自适应迭代学习控制方法,以及一种基于时频分析的带宽自适应滤波器型迭代学习控制改进方法。第三,针对具有重复运动特性的摩擦力模型参数时变伺服系统,提出了一种基于迭代学习的自适应摩擦补偿方法,即类似自适应摩擦补偿算法,通过重复动作间的迭代学习来提高摩擦力模型参数的估计精度,从而达到提高伺服系统性能目的。第四,介绍了笔者参与研制的系列注塑机通用控制系统产品,以及S80注塑机通用控制系统测试及算法验证实验平台的建立。第五,针对注塑特殊工艺的不断开发和推广,对开合模机构定位精度越来越高的要求,提出一种用于提高注塑机开合模机构定位精度的自适应学习增益的迭代学习算法。另外,从基于比例阀的注塑成型液压控制系统动态特性角度出发建立了非对称注射缸的液压系统模型。接着针对目前国内注塑机技术现状,讨论了注塑成型过程控制策略,采用基于反馈+预期迭代学习控制对注塑阶段中的注射速度进行控制,并探讨了时频自适应滤波器迭代学习方法在注射速度控制中的应用。 最后,对本文所取得的研究成果做了总结,指出尚待解决的问题并对后续工作进行了展望。
其他摘要The advanced Injection Molding Machine (IMM) control system is urgently needed because there is an increasing gap of control technology between domestic and overseas IMM controllers. In the paper, robust and adaptive designs of Iterative Learning Control (ILC) are investigated, and the related algorithms are implemented into injection velocity control and clamping mechanism’s position control. Firstly, the dissertation describes the current status and the developments of domestic IMM market. Then the IMM control technologies are surveyed briefly. Especially, the ILC algorithm is introduced and surveyed. Secondly, the systems analysis of robust and monotonically-convergent ILC is studied. Then several approaches that have been proposed in the literature to add convergence robustness to ILC are presented. Especially, the discrete Filtering ILC (FILC) is fully studied in the time and frequency domains. A novel adaptive update rule ILC algorithm and an improved time-frequency adaptive filter ILC algorithm are proposed. Thirdly, an adaptive friction compensation approach based on iterative learning is developed to address a class of nonlinear servo systems with repeatable actions and time variant friction model uncertainties, which are iteration independent. The simulation illustrations are given to show the approach can achieve high-accuracy trajectory control. Fourthly, the series products of real-time IMM control system are developed. Then, the hardware and software setup to test IMM are introduced. Fifthly, the adaptive update rule ILC is applied to improve the clamping mechanism’s position precision of IMM. Furthermore, a changeable update rule is introduced to accelerate the convergence speed of ILC. Next, a physics-based injection molding model is developed by incorporating the dynamics of the proportion flow valve and the force equation of asymmetric cylinder. The model provides an excellent simulation environment for rapid controller design and tuning. Then the injection molding control strategies are discussed, and a discrete anticipatory learning control plus PI control is applied to ram velocity profile tracking in injection molding process. Last, the time-frequency adaptive filter ILC algorithm application in ram velocity profile tracking is also studied. The ILC approaches are shown to provide a significant increase in system performance. Finally, all the results are summarized and the future works are put forth.
馆藏号XWLW1160
其他标识符200318014603038
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/6040
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
杨雁. 基于鲁棒自适应迭代学习的注塑控制策略研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2008.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
CASIA_20031801460303(2331KB) 暂不开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[杨雁]的文章
百度学术
百度学术中相似的文章
[杨雁]的文章
必应学术
必应学术中相似的文章
[杨雁]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。