【作者】 杨志东；

【导师】 李洪人；

【作者基本信息】 哈尔滨工业大学 ， 机械电子工程， 2009， 博士

【摘要】 振动环境模拟被广泛应用于航空、航天和国防军事等领域,目前已成为检测和提高装备可靠性的必备手段。通过使用特定的振动模拟设备和控制技术模拟产品在使用,运输以及存储等过程中所受到的振动激励,可以检测设计上的不足和缺陷,为及时地修正设计提供依据。本文以哈尔滨工业大学电液伺服仿真及试验系统研究所承接的“单水平向液压振动台数控系统改造”为项目背景,对模拟随机和正弦两类振动环境涉及的关键控制技术进行理论和实验研究,其研究成果对于充分发挥液压振动台在振动环境模拟方面的潜力具有重要的指导意义。液压振动台实时控制单元由伺服控制器和振动控制器两部分组成。伺服控制器实现运动平台的数字闭环控制,它是实现各类振动环境模拟的控制基础。本文针对液压振动台动力机构固有频率比较低、阻尼比很小的特点,引入三参量控制策略,提高系统稳定性和拓展系统频宽。同时,文中应用根轨迹设计法对控制器参数进行设计,充分考虑了伺服阀动态特性对激振器性能的影响,并通过实验验证了该方法的有效性。高斯随机振动环境是实际工程中经常遇到的一类振动环境,功率谱密度通常作为此类振动环境模拟的等价条件。而驱动平台运动的激励信号必须为时域信号,因此如何完成功率谱密度至时域信号的转换是此类振动环境模拟的首要工作。本文首先对该类振动环境的数值模拟技术进行了系统地论述,提出了基于自回归(Auto-Regressive,AR)参数模型和Ziggurat高斯随机数生成算法的数值模拟方法。实时构造的AR模型对平稳、各态历经和独立的高斯白噪声进行滤波,从而实现了对驱动功率谱密度的精确时域模拟。仿真结果表明这种方法不仅模拟精度高,而且计算量小,计算简单,同时能够保证时域驱动信号的各项统计特征要求。另外,该方法支持xPc Target快速控制原型技术,极大地简化了驱动信号的生成过程。在随机过程数值模拟的基础上,考虑到激振器和负载特性的影响,为保证足够的功率谱密度再现精度,需要对驱动功率谱进行迭代控制。控制点处的响应功率谱密度估计值是迭代控制过程中的反馈量(也称为被控制量),其估计精度与响应信号测量的统计自由度密切相关。为此,通过详细分析响应信号功率谱密度估计时的统计特性,提出了一种优化的功率谱平均方法,,该方法不仅提高了响应信号功率谱估计的置信度和精度,而且缩短了功率谱再现迭代控制中的回路时间,进而提高了控制过程的实时性。传统的频谱均衡控制算法(无论是线性域积分控制还是对数域积分控制)均采用控制误差直接补偿的闭环控制策略,其稳定性受到反馈增益和外界干扰的影响。为了保证系统的稳定性,通常需要选择较小的反馈增益,但是这样做却导致均衡时间变得很长。为了克服这种缺陷,本文提出了频域X滤波变步长LMS算法,基于此算法构造了随机振动功率谱再现的自适应控制律。在实时闭环控制的同时,阻抗函数(系统频响函数的逆)在频域内被不断修正,改善的阻抗函数可以允许更高的反馈增益,从而使均衡时间缩短而又不失稳定性。仿真结果表明,该方法相对传统的频谱均衡算法具有更好的稳定性、更快的收敛速度和更高的控制精度。正弦扫频振动试验既可以确定样品的共振频率和导致样品失效的危险频率,还可以进行扫频耐久性试验。本文针对正弦扫频信号实质上为瞬态信号的本质特性,提出了基于相位函数解析解的信号综合算法,并基于此信号综合算法设计了一种数字跟踪滤波器。该信号综合算法不仅可以提高输出波形的精度,而且可以与外差操作紧密结合,将响应信号在当前频率点的频谱平移至直流分量附近,该直流分量包含响应信号在当前频率点的基频幅度信息。直流中频检测器通过选择适当的低通滤波器释放出直流分量,即可实现跟踪滤波处理。仿真表明这种数字跟踪滤波器特别适合于正弦扫频信号的时变谱分析。另外,传统的幅度控制算法采用固定的压缩比,它无法解决对不同频段和不同系统特性的适应性问题。为此,本文通过引入优化概念,设计了一种分频段变压缩比积分控制器,提高了的正弦扫频振动控制的精度。仿真结果验证了这种控制技术的优越性。基于xPc Target快速控制原型技术开发了用于随机振动功率谱再现和正弦扫频振动的多级计算机控制系统,并针对各种控制策略进行了一系列的实验研究。实验结果进一步证明了仿真研究的结论,验证了本文所提出的各种控制技术的有效性和先进性。更多还原

【Abstract】 Vibration environment simulation has been applied in the fields of the aspects of aeronautics, astronautics, national defense and so on. Now the vibration environment simulation is an important means for increasing the reliability of the products. By simulating the particular vibration environment that products will exposed to in use, transform and storage using the specific test equipments and the control technologies, weakness of design can be early found and the information for correcting design is obtained in time.Upon the background of the project“Digital control system rebuilding of single horizontal direction hydraulic vibration table”which is developed by IEST (Institute of Electro-hydraulic Servo Simulation & Test System) of Harbin Institute of Technology, the key control technologies of the random and sine vibration environment simulation has been theoretically and experimentally studied,in order to enhance the capability of the hydraulic vibration table in the field of the vibration environment simulation.The real time control unit of the hydraulic vibration table consists of the servo controller and the vibration controller. The servo controller is the basic to provide vibration environment simulation by a digital closed loop. Due to the low resonance frequency and small damp ratio of the power element, the Three Variable Controller (TVC) is proposed in order to extent the frequency band and improve system stability. Considering the effect of the servo valve dynamic characteristics on the performace of the electro-hydraulic shaker, the root locus method is used to design the TVC parameters. And the validity of this method is confirmed by experiment.Gaussian random process is common in engineering. The power spectrum density (PSD) is equivalent condition. But the signal to excite vibration table must be in time-domain, therefore how to transform the PSD into the time-domain signal is the first work to simulate this kind of vibration environment. This paper firstly presents an overview of numerical simulation techniques to this kind of vibration environment. Then a numerical simulation method based on the Auto-Regressive (AR) model and the Ziggurat algorithm is proposed. The Stationary, ergodic and independent Gaussian White Noise is filtered by the AR model constructed in real-time, so that the accurate simulation of drive PSD is implemented. Simulation results show that this simulation algorithm has higher precision and low computation complexity, and it can also meet requirement about the statistics, besides, it can support xPc Target Rapid Control Prototyping (RCP) technology, so that the process generating time-domain drive signal is greatly simplified. On the basis of the numerical simulation, considering the effect of the dynamics of actuator and load on control accuracy, it is necessary to update drive PSD constantly. The PSD of the response signal at control points is not only the feedback quantity but also the control quantity. The estimated value of the quantity is closely related to the degree of freedom (DOF). So, the statistical characteristics about the response PSD estimation are analyzed in detail. An optimal method for averaging response PSD is proposed. The optimal method has increased confidence and accuray of the estimated response PSD, and has shortened the loop time,so that the real-time characteristics of the iterative process is increased.Classical spectrum equalization technique, either log-domain integral or linear- domain integral, corrects the control error by the closed loop feedback. In order to guarantee the stability, the feedback gain is usually enough small theorefore may result in very slow control system convergence. To solve the limitation, the frequency-domain filtered-X variable step-size least mean square (LMS) algorithm is introduced into control process, and thus adaptive control strategy of random vibration for PSD replication is proposed. The system impedence function,which is the inverse of the frequency response function, is refined in frequency-domain in real-time. The improved system impedance function has allowed a higher feedback gain parameter, which can speed up equalization process and guarantee the stability. The simulation results have validated that the control strategy has better stability, higher convergence rate and control precision.Swept sine testing can be used to determine resonance frequency and damage frequency, and to conduct durability testing. To non-stationary characteristics of the swept sinewave exciting signal, the swept-sine wave generation has been accomplished by solving the governing differential equation of the phase function associated with the linear sweep and the logarithmic sweep mode. The digital tracking filter is designed based on the signal synthesis method. This synthesis method has increased the accuracy of the output signal and has the suitability of heterodyne generation. The purpose of the heterodyne is to translate the response signal spectrum about 0-Hz. The term at DC represents the real and imaginary components of its AC amplitude.The 0-Hz intermediate frequency detector implements tracking filter by a low-pass filter that extracts the DC component. The simulation results have proved that the proposed digital tracking filter can be applied to time-variable spectral analysis of the swept-sine wave. On the other hand, the conventional amplitude control method can not be suitable for various frequency segment and various the system characteristics due to a fixed compression ratio. The optimal control concepts is applied to design an integral controller with a variable compression ratio (VCR) for frequency-devison section so as to provide better control accuracy. Simulation results have proved the effectiveness of the designed controller.The multi-level computer control system for the random vibration PSD replication and swept sine vibration is established using RCP technology based on xPc Target. A series of experiments are respectively carried out using various control strategies. Experimental results further proved the conclusions from the simulation study, and validated that the proposed control techniques are advanced and effective.更多还原