基于响应的机床切削自激励与动力学参数识别方法研究

发布时间：2018-10-24 15:36

【摘要】：机床的加工性能与其动态特性紧密相关,随着机床装备朝高速高精高可靠性方向发展,有关机床动态特性的研究也越来越重要。实验模态分析方法是研究静态下结构动态特性的主要手段。然而机床在加工状态与静止状态下的动态特性有很大区别,传统的实验模态分析识别的结果难以准确地表征机床工况时的动态特性。工作模态分析方法OMA (Operational or Output-only Modal Analysis)是一种仅需要测量响应来识别结构在工况下动态特性的新方法,因其简单实用,前景广阔而备受关注。然而,工作模态分析理论本身并不完善,例如由于不测输入无法估计频响函数,模态参数的识别依赖人工经验；同时,当OMA用于机床等机械结构时会带来一些新问题：首先,工作模态分析要求激励信号为白噪声信号,在频域上均匀分布,这对于机床等机械结构难以实现；其次,对于机床等旋转机械,其响应通常包含周期信号,导致模态参数识别受到谐频的严重干扰从而难以区分固有频率和谐频频率。本文旨在针对上述问题,在激励方法,信号处理,工作模态分析理论及模态参数自动识别算法方面做一些研究和探索,力图提出一套面向机床等机械结构的仅基于响应的模态分析解决方案。 首先,结合随机信号和脉冲信号的特点,采用随机脉冲激励信号,提出了多种切削激励方法,并实现机床结构动态特性与切削过程的解耦。建立了随机脉冲激励信号的数学模型,并通过该数学模型和仿真分析阐明了激励频带和能量与脉冲参数、切削参数的关系,总结了该信号的激励规律和特性,为实现激励的可控可调提供理论基础。分别从运动方式和切削工件两方面入手,提出主轴随机转动切削、随机多凸台、随机(曲线和折线)单凸台、渐变宽度凸台和凸点切削激励方法,从激励方法上对机床结构动态特性与切削过程解耦。 其次,提出了利用切削参数、刀具几何参数、工件材料特性及切削力系数等已知信息,预测切削激励力从而代替测量,并结合工作模态分析实测的振动响应信号,估计频响函数的新方法,弥补了OMA不测输入带来的缺憾。其中,为了考量转速对切削力的影响,通过对瞬时切削过程的功能分析,提出转速影响因子以改进预测切削力的瞬时刚性力模型。 第三,提出了识别工作模态分析中可能产生的伪模态(Spurious mode,即虚假模态)的方法。首先,根据产生的原因和表现出的特性,对可能的伪模态进行分类。然后,根据各类伪模态的特性和差异,提出了对应的识别方法。对于最难处理的谐频模态(Harmonic mode),本文提出了频率栅栏缝和谱突变度两种新工具,实现即使在不知谐频成分的复杂工况下也能识别谐频模态。 最后,对模态参数自动识别的两个关键问题,极点自动归组和从各组极点中自动识别各阶模态参数,进行了探讨。根据物理极点比较集中,而计算极点较为分散的特性,提出极点分布准则以用于清除计算模态。提出了基于最小方差的模态参数自动识别算法,并与伪模态识别算法、极点分布准则以及极点容差率准则整合,在Matlab上实现了适用于复杂工况的只基于响应的模态参数自动识别算法Auto-LSCE。 总体而言,本文提出了一种利用切削力激励机床而仅利用响应信号进行参数识别的模态分析方法,解决了OMA分析理论中频响函数无法估计、模态参数自动辨识及OMA用于机床等结构时的激励、伪模态识别等科学及技术难题。本文所取得的研究成果对推动OMA分析方法在机械工程领域的研究与推广具有积极意义。
[Abstract]:The machining performance of the machine tool is closely related to its dynamic characteristics. With the development of machine tool equipment toward high speed and high precision, the research on the dynamic characteristics of the machine tool is becoming more and more important. The experimental modal analysis method is the main means to study the dynamic characteristics of static structure. However, the dynamic characteristics of the machine tool in the machining state and the static state are very different, and the traditional experimental modal analysis and recognition results are difficult to accurately characterize the dynamic characteristics of the machine tool working condition. Working mode analysis method OMA (Operational or Function-only Modal Analysis) is a new method which only needs measurement response to identify dynamic characteristics of structure under working condition. However, the theory of modal analysis is not perfect, for example, because the input cannot estimate the frequency response function, the identification of modal parameters depends on the artificial experience; meanwhile, when OMA is used for mechanical structures such as machine tools, some new problems are brought forward: firstly, the working mode analysis requires that the excitation signal be a white noise signal, which is uniformly distributed in the frequency domain, which is difficult to achieve for mechanical structures such as machine tools; secondly, for rotating machines such as machine tools, the response typically comprises a periodic signal, It is difficult to distinguish the harmonic frequency of the natural frequency by causing the modal parameter to recognize the severe interference of harmonic frequency. In this paper, aiming at the above problems, some research and exploration are made on the excitation method, signal processing, working mode analysis theory and modal parameter automatic recognition algorithm. First of all, combining the characteristics of the random signal and the pulse signal, the random pulse excitation signal is adopted, a plurality of cutting excitation methods are proposed, and the dynamic characteristics and the cutting process of the machine tool structure are realized. The mathematical model of the random pulse excitation signal is established, and the relationship between excitation frequency band and energy and pulse parameters and cutting parameters is clarified by the mathematical model and simulation analysis. The excitation law and characteristic of the signal are summarized. On the basis of the two aspects of moving mode and cutting workpiece, this paper proposes random rotation cutting of main shaft, random multi-convex table, random (curve and broken line) single convex table, gradual width convex table and convex point cutting excitation method, and the dynamic characteristic and cutting process of machine tool structure are analyzed from the excitation method. Secondly, using the known information such as cutting parameters, tool geometry parameters, workpiece material properties and cutting force coefficients, the cutting excitation force is predicted to replace the measurement, and the measured vibration response signal is analyzed in combination with the working mode, and the frequency response function is estimated. The new method of number has made up for OMA's unmeasured input. In order to consider the effect of rotating speed on cutting force, the influence of rotating speed on cutting force is analyzed. By analyzing the function of instantaneous cutting process, the influence factor of rotating speed is put forward to improve the instantaneous cutting force. Thirdly, the pseudo-mode which may be generated in the modal analysis of the identification is presented. A method of making spurious modes. First, according to the causes and characteristics of the generation, it is possible to The pseudo-mode is classified. Then, based on the characteristics and differences of various pseudo-modes, it is proposed that In this paper, two new tools for frequency fence joint and spectral mutation are proposed for the most difficult harmonic mode (harmonic mode). It is also possible to identify the harmonic frequency mode. Finally, two key problems of automatic identification of modal parameters, the pole auto-return group and the automatic identification of each group from the poles of each group are identified. In this paper, the modal parameters of the order are discussed. According to the physical pole comparison set, the characteristic of the extreme dispersion of the poles is calculated and the pole distribution is put forward. In this paper, a modal parameter automatic recognition algorithm based on minimum variance is proposed, which is combined with pseudo-modal identification algorithm, pole distribution criterion and pole tolerance criterion. The algorithm Auto-LSCE. In general, this paper presents a modal analysis method which uses cutting force to excite the machine tool and uses the response signal only to identify the parameters, which solves the problem that the frequency response function in OMA analysis theory can't be estimated, the modal parameter can be recognized automatically. The excitation of OMA for machine tools and other structures The result of this paper is to push the OMA analysis method in mechanical engineering.
【学位授予单位】：华中科技大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TG506
，

本文编号：2291786