【作者】 于慧君；

【导师】 王庆丰； 陈章位；

【作者基本信息】 浙江大学 ， 机械电子工程， 2009， 博士

【摘要】 产品结构的可靠性,反映了其在使用及运输过程中对实际振动环境的承受能力。长时间历程复现振动试验作为一种产品结构可靠性测试的有效手段,越来越受到重视。它将记录下来的振动信号,经预处理得到复现试验的目标信号,并在试验室的振动台上复现该目标信号,以对产品进行可靠性测试。长时间历程复现振动试验可有效应用于模拟汽车、船舶及飞行器等运行过程中所承受的振动环境。为得到长时间历程复现试验控制系统的目标信号,本文对长时间历程信号预处理技术进行了分析,并着重研究了信号重采样及多段信号平滑技术。文中提出了基于最小二乘法的信号重采样算法,即应用最小二乘法将重采样信号处理中低通滤波器系数的求取问题转化为相应矩阵方程的求解,并应用QR分解方法提高求解精度。数值及实验结果表明该算法可有效地应用于长时间历程信号的重采样处理,并能较为直观地得到精确的结果。对于多段信号的平滑技术而言,文中提出了基于加窗重叠的多段信号平滑连接算法。通过对两端信号分别加窗,减小了由边界不连续性导致的频谱泄漏,将加窗后的两段信号重叠相加从而减轻时域信号波动以避免在振动过程中产生过冲的现象。在得到了长时间历程复现目标信号后,后续的工作是如何对其进行复现控制。文中对实际环境振动信号的统计特性进行了研究,将长时间历程复现控制分为高斯及非高斯复现控制,并提出了相应的控制算法,即基于相关分析和高阶累积量的自适应逆时域复现控制算法。前者采用相关分析的方法实现系统逆传递特性的估计,以达到实时精确的控制效果。后者,考虑到在非高斯情况下,二阶统计量不能完全描述系统特性,因此引入高阶累积量来估计系统的逆传递特性,以提高相位估计精度及减小噪声干扰,提高控制精度。一般而言,长时间历程复现试验大多需要能提供大承载、大推力的电液振动台。考虑到电液振动台存在波形失真大及频响带宽较窄等缺点,本文对电液振动台建模方法进行了研究,并在此基础上,提出了基于广义预测控制的自适应三参量控制算法。三参量控制技术被广泛应用于电液振动台控制,其性能完全取决于控制参数。注意到,三参量控制技术并不完全适用于未知时滞及参数时变的系统。因此本文应用广义预测控制方法来对三参量控制参数进行调节,并且应用递归最小二乘法辨识系统模型,从而克服了由参数时变性对控制性能产生的影响。仿真及实验结果表明该方法能有效改善系统的波形失真度。更多还原

【Abstract】 The structural reliability of mechanical product shows its ability to endure the real environmental vibration in operation and transportation.As a versatile tool for examining structural reliability,long time history replication(LTHR) vibration testing have been paid more and more attention in automotive,ship and aeronautical engineering in recent years.LTHR vibration testing can be applied effectively to vibratory simulation of real environmental vibration with the electro-hydraulic shaker which implementing preprocessed target signal acquired from real one.In order to obtaining the target signal of LTHR testing control system,the long time history (LTH) preliminary processing techniques,including signal resampling and multiple segments signal smoothing algorithms will be presented in this dissertation.The former is used to acquire the low-pass filter coefficients in resampling process through proper transformation.With the transportation,the filter coefficients can be obtained from the corresponding matrix equation using by QR decomposition in high precision.The latter is used to minimize the effects of energy leakage in signal spectrum.The specially shaped windows are applied to the two signal segments after they are resampled and the two windowed time waveforms will be overlapped in some extent that the strong fluctuation and overshoot of shaker can be prevented effectively.After the acquisition of the target signal of LTHR vibration testing,the successive work is how to control the electro-hydraulic shaker in accordance with the given target signal for exact reproduction of real environmental vibration.In this dissertation,we find that the LTH control can be classified into two main categories,namely Gaussian and Non-Gaussian LTHR control, after systematical study on the statistical properties of real environmental vibration.Aimed at two different categories above,the adaptive inverse time domain replication control algorithms are proposed correspondingly with correlation analysis and high-order cumulants.The Gaussian one is used to achieve the real-time and high degree of accuracy control through the precise estimation of system inverse transfer behavior by correlation analysis.The Non-Gaussian one is used to improve the control accuracy and phase estimation under noise disturbance.And considering second-order statistical quantities are not appropriate to describe system behavior in Non-Gaussian circumstance,the system inverse transfer behavior is gained using by high-order cumulants.LTHR vibration testing is generally implemented using by the electro-hydraulic shaker with high load-bearing and pushing force.However,there is the existence of large waveform distortion and low frequency reponse.Considering these two drawbacks,the modeling method of electro-hydraulic shaker will be researched in this dissertation.After the model built,an adaptive three-variable control algorithm based on generalized predictive control(GPC) will be presented.Three-variable control scheme has been widely used in the vibration control of electro-hydraulic shaker.However,it is not appropriate for unknown time-lagged and time varying system by the reason of strong dependence on its own control parameters.In order to improve the flexibility of determining control parameters and minimize the effect of parameter time-variability in vibration control,the GPC will be applied to tune the control parameters and the recursive least square method will be used in system modeling in this dissertation. Numerical and experimental results indicate the proposed approach is efficient and versatile tool in minimizing the waveform distortion of eletro-hydraulic shaker.更多还原