【作者】 冯宗杰；

【导师】 陈志华；

【作者基本信息】 国防科学技术大学 ， 机械工程， 2010， 硕士

【摘要】 活塞是发动机的核心部件,其外圆截面为非圆亦被称为异型截面。普通车削无法实现异型车削,专用伺服刀架可进行异型车削,但对于截面形状复杂的活塞,加工误差较大,无法满足设计精度要求。在现有异型外圆加工设备的基础上,本文提出一种改进策略,旨在提高其加工复杂截面的精度。迭代学习控制是智能控制领域的一个重要分支,对于重复或类重复问题具有学习功能,其良好特性对于具有类重复性质的活塞异型截面加工,可有效提高伺服刀架跟踪精度,进而有效提高复杂截面的加工精度。本文介绍了活塞异型外圆数控加工的基本原理,活塞截面数据特性和现有伺服刀架的基本控制方法。针对现有伺服刀架加工复杂活塞频响不足、精度较低的问题,提出了基于迭代学习控制的改进策略并给出了实现流程。介绍了迭代学习算法的基本原理,详细介绍了预期迭代学习控制及反馈前馈迭代学习控制算法,通过MATLAB数值仿真证明了迭代学习算法应用于活塞加工的可行性。对学习前活塞加工测量数据进行离线反馈前馈迭代学习控制,修正伺服刀架的控制输入数据并进行了再加工,对比了迭代学习控制前后的加工误差。在800r/min的转速下,利用迭代学习控制对复杂活塞外圆进行数控车削,最大误差由原来的0.0248mm大幅度减小到0.0086mm。实验结果表明,所应用的迭代学习控制算法能有效提高复杂截面活塞的加工精度,控制算法简单、实用,可行性强。本文的工作对于提高复杂异型截面活塞的加工精度具有较高的理论价值与应用价值。更多还原

【Abstract】 Piston is the one of the core part of the engine. The cross section of the piston is usually noncircular, which can’t be machined by the conventional lathe but can be realized by the special servo tool. However the tolerance of the machined part is often large if the cross section is very complicated.A modified control strategy to improve the machining precision of complicated cross section is put forward in this dissertation based on the existing noncircular cylindrical piston numerical-control equipment. Iterative learning control, an important branch of intelligent control area, which has the learning ability for iterative problems or the analogy ones, is used to improve tracking precision for an iterative signal, as the process of the piston noncircular turning has the property of analogy iteration. As a result, the machining precision of the piston can be improved. The main contents are as follows:Firstly, the basic noncircular section piston process principle was introduced, and the character of typical piston section data is analyzed. At the same time, the basic control method for the existing servo tool-post is also present. Secondly, aim at the limited frequency band of the servo tool-post and its low precision, an improved control strategy of iterative learning control with its realization flow is put forward. Thirdly, the basic iterative learning control principle is introduced, and the anticipatory iterative learning control and learning feed-back and feed-forward control are presented in detail. The MATLAB program numerical simulation shows the feasibility for the iterative learning control scheme to be used for the machining of piston. Fourthly, the process data is modified by the off-line learning feed-back and feed-forward control. Applying iterative learning control to complex section piston numerical control machining, maximum error could be decreased from 0.0248mm to 0.0086mm in the case of 800rmp. This result show that the machining precision was efficiently improved after the process data was modified by iterative learning control strategy.Briefly, the research in the paper will help the precision improvement of complicated noncircular section piston machining.更多还原