Time-Domain Experimental Identification Technique of Rotating Short Systems

随着旋转机械的高度越来越高、重量越来越轻，实现其稳定、安静运行的难度也越来越大。为了在这种情况下对机器进行建模，实验辨识技术变得更加重要。在这项研究中，我们试图开发一种新的适用于旋转机械的实验辨识技术。为了识别旋转机械，我们遇到了一些困难，这些困难在通常的非旋转机械中是免费的。首先，旋转机械在高速运转时通常很难被激振。其次，它们包含不平衡和初始缺陷等参数，这些参数在通常的非旋转机械中并不重要。第三，它们往往会变成非线性系统。为了克服这些困难，我们在以前的研究中发展了频域技术。然而，这些技术需要很长的时间来进行识别。为了改善这一缺陷，我们尝试开发一种时域技术。首先，作为基础性研究，我们考虑了带有圆盘的转轴，提出了一种基于最小二乘法和拉格朗乘子法的辨识方法。通过数值模拟和实验验证了该方法的适用性。然后将该方法推广到非线性系统中。最后，我们开发了一种用于连续旋转机械的技术。并通过数值模拟和实验验证了该方法的适用性。

As rotating machines becomes higher and lighter, it becomes more difficult to realize their stable and quiet operation. To model machines in this situation, experimental identification techniques become more important. In this research we attempt to develop a new experimental identification technique suitable for rotating machines. To identify rotating machines, we encounter some difficulties which are free in the usual non-rotating machines. First the rotating machines are usually difficult to be excited during their high-speed operation. Second they contain such parameters as unbalances and initial imperfections, which are not important in the usual non-rotating machines. Third they often become nonlinear systems. To overcome these difficulties, we developed, in a former research, frequency-domain techniques. These techniques, however, require long time to conduct identification. To improve this defect, we attempt to develop a time-domain technique. First, as a fundamental study, we consider a rotating shaft with a disc, and propose an identification technique using the least square method and the Lagrane multiplier method. We showed its applicability by numerical simulation and experiment. Then we generalize the technique so that it is applicable to nonlinear systems. Finally, we developed a technique for continuous rotating machines. We also showed its applicability by numerical simulation and experiment.