节点文献

滑动导向系统摩擦建模及动力学研究

Study on Friction Modeling and Dynamic Characteristics of Slide Guide System

【作者】 郭克尖

【导师】 孟光;

【作者基本信息】 上海交通大学 , 机械设计及理论, 2008, 博士

【摘要】 在机械系统的运行过程中往往有摩擦环节存在,常常对系统的性能产生很大的影响。尤其,由于摩擦在低速时表现出强烈的非线性,系统在启动、停止以及速度转向时受到的摩擦的影响最为严重,甚至由于摩擦的存在会使系统进入无序运动的混沌状态,极大的降低系统的性能。因此随着现代机械朝着高精度、高性能方向的发展,对摩擦环节的建模以及动力学特性的研究就变得越来越重要。摩擦现象在微观尺度上具有极其复杂的物理学本质,在一般的机械系统领域对摩擦的研究中,都是力求从宏观尺度上最好的解释和模拟摩擦现象和行为。因此本课题基于摩擦实验研究,对摩擦特性进行分析,提出新的宏观经验模型结构,并给出相应的摩擦信号处理和参数识别过程,此外,运用数值方法对滑动导向系统的动力学特性进行了探讨。论文首先概述了有关摩擦特性的实验和理论研究的国内外研究现状和研究进展,总结了各种摩擦现象的产生原因和表现形式,在此基础上,介绍了各种不同的摩擦力模型,并分析了它们的优缺点,简述了摩擦信号处理以及摩擦系统动力学特性研究等方面的研究方法。针对滑动导靴和导轨之间的横向和纵向摩擦的特点,设计了相应的实验装置,拟定了实验方案,并对实验结果进行了初步的分析和处理。通过分析和比较大量的实验结果,探讨了导靴与导轨之间的摩擦特性,包括摩擦力随相对运动的位移、速度、加速度,以及正压力、润滑状况和导靴类型等参数的变化情况,为进一步的研究工作提供实验基础。由于测量方法以及测量仪器的局限性,实验所测的摩擦信号不可避免的包含噪声信号和其他力信号成分,例如弹性力信号等。随着摩擦力模型的精细化和复杂化,参数辨识对实验数据中的噪声信号表现出高敏感性,对于被噪声和其它力信号污染严重的数据,部分辨识方法甚至失效。因此如何将真正的摩擦力信号从测量信号中分离出来在摩擦参数识别中至关重要。而传统的基于傅利叶变换的分析方法往往不适合处理此类非线性信号。本文提出了使用中值滤波和经验模态分解(Empirical Mode Decomposition)相结合的方法来分析处理实验中所测得的摩擦力信号。并将处理结果与经由基于傅利叶变换的方法(低通和带阻滤波,以及小波分解方法)处理的结果在时域上、Fourier谱空间以及Hilbert谱空间分别进行了比较,结果显示经验模态分解方法在摩擦力信号分析和处理上具有优越性。在对摩擦特性实验研究的基础上,通过引入规范化的Bouc-Wen模型,提出了一个新的动摩擦力模型结构。这个模型结构能够描述摩擦力在微滑动阶段、宏观滑动阶段以及过渡阶段的非线性特性。同时还提出了一个新的不可逆宏观滑动摩擦模型。给出了当滑动摩擦模型分别为速度依赖性模型和此不可逆模型时的参数识别过程,针对实验数据给出识别结果,并对实验中的摩擦过程进行仿真,仿真结果与实验结果在时间历程上以及摩擦力-相对运动位移和摩擦力-相对运动速度的坐标图上的比较,显示了此摩擦模型结构的有效性和描述摩擦过程的精确性。对于在一定牵引速度下受到横向位移激励的滑动导向系统建立了振动分析的二维模型,其中充分考虑了横向和纵向由于摩擦引起的耦合问题。数值仿真结果显示,在一定的参数状况下滑动导块有可能出现爬行运动状态。同时分析了牵引速度和横向激励的频率对横向位移响应的影响。并将二维模型的仿真结果与不考虑纵向振动的一维模型的结果作了比较。发现,由一维模型得到的导块的横向位移响应的幅值要大于二维模型的结果。但当横向位移激励的频率与导向系统横向振动的固有频率相差较远时,两种模型的数值仿真结果相差不大。建立了轿厢-导靴-导轨动力学耦合模型来研究电梯滑动导向系统的横向振动特性。并且由于导轨表面的不平顺引起横向位移激励的频率远小于导向系统横向振动的固有频率,因此不考虑由于摩擦耦合引起的导向系统纵向振动,认为在牵引方向始终以额定速度运动。在系统建模过程中,充分考虑了包括导靴与导轨的安装间隙和摩擦等非线性因素,和电梯运行时的负载不平衡等状况。并建立了该模型的Matlab/Simulink仿真程序。仿真结果与实测结果在频域上的比较表明了该模型的有效性。本文的研究工作对摩擦实验研究、摩擦特性研究、摩擦信号处理、摩擦建模和参数识别,以及摩擦系统动力学特性的研究等方面具有一定的参考价值。

【Abstract】 Friction exists in almost all mechanical systems, and often affects the performance of a system seriously, especially when the system in the states of starting、ending and velocity reversal. That is because friction shows strong nonlinear characteristics when relative velocity is slow. Sometimes the existence of friction can lead the motion of a system into chaos. Therefore the research on dynamics of mechanical system with friction is becoming a hot issue in the current times.Friction contact is extremely complicated from the microscopic view, and in the field of mechanical system, the study on friction is often from the view point of macroscopic to explain and predict friction phenomena and behaviors. On the basis of experimental studies and analyses for friction dynamic characteristics, a new dynamical friction model structure is proposed. The procedures of friction signal processing and parameter identification are also presented. In addition, the dynamic characteristics of systems with friction are studied using numerical techniques.First, the domestic and oversea research actualities of friction dynamic characteristics have been summaried, the manifestations and formings of these friction phenomena are discussed. Based on these, various friction models are introduced, and the advantages and disadvantages of them are expounded. The methods for friction signal processing are mentioned in brief, and the work on the dynamical characters of systems with friction is introduced.The setups and schemes for experimental study are designed according to the characters of friction contact between slide guide and rail along lateral and vertical directions. Experimental results are processed and analyzed. Through comparisons and analyzing of a large amount of cases, the dynamic friction characteristics between slide guide and rail, including the variation tendency of friction with the parameters just like relative displacement、relative velocity、relative acceleration、contact area、lubrication condition and slide guide type, are discussed. This part lays the foundation for furthering study.Due to the limitations of methods and instruments for friction measurement, the measured friction signal often contains noise and other force components, such as elastic forces. And because friction model becomes more and more complicated, parameter identification result often shows high sensitivity to noise contamination, and some methods even fails. How to extract the real friction signal from measured one is important to identification work. However, the traditional Fourier methods are not suitable to deal with nonlinear signal processing. The combination of median filter and empirical decomposition (EMD) method is used to analyze and process the measured friction signal. The processing results are compared with the results from Fourier-based methods (lowpass filter, band block filter and wavelet decomposition method) in the time history、Fourier spectral space and Hilbert spectral space. The comparisons show that the EMD method has superiority in processing measured friction signal.On the basis of experimental study, and through introducing the normalized Bouc-Wen model, a new dynamical friction model structure is proposed. This model has the capability to describe nearly all the nonlinear friction dynamic characteristics in presliding、sliding and their transition state. A new nonreversible friction model is presented for sliding regime to describe friction memory effect. Parameter identification procedures for both models are given. Identified friction models are used to predict friction courses, and results are compared with measured ones in the coordinates of friction vs. time、friction vs. relative displacement and friction vs. relative velocity. The effectiveness and accuracy of the proposed models can be shown from these comparisons.A two-dimensional model for the vibration of a slide guide system subjected to lateral displacement excitation is proposed. The coupling between lateral and vertical vibration due to the existence of friction is considered sufficiently in this model. Simulation results with this two-dimensional model show full sliding motion and stick-slip motion which may appear with some proper parameters. The variation tendencies of the vibration amplitude in lateral direction with the parameters in the model are simulated. Results derived from the two-dimensional model are compared with a one-dimensional model which does not included the vibration in sliding direction. The comparisons indicate the lateral response calculated from the abbreviative one-dimensional model is larger than the result from the two-dimensional model.An elevator cabin-guide-rail coupled lateral vibration model is developed to research the dynamical characters of slide guide system in an elevator. The mass of slide guide, the nonlinear factors such as the gap and friction between slide guide and rail, in addition to the load condition (balanced or unbalanced) are all included in this model. The comparisons of simulation results from Matlab/Simulink program with the experimental results illustrate the effectiveness of this model. The exploration provided an important theoretical foundation for the optimal design of slide guide and the determination of the gap between guide and rail in assembling. In the mean time, it can be helpful in the prediction and control of elevator vibration and noise.The results of this thesis may be worthily used to understand and develop the theories about friction experimental study, dynamic friction characteristics, friction signal processing, modeling and parameter identification, and dynamical characters of systems with friction.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络