Assessment and reduction of uncertainties in operational modal analysis (RUN-OMA)

利用环境激励下的模态参数对机械结构的振动特性进行实验识别是工作模态分析的目的。为此，基于频域和时间域配方的许多可靠且经过充分验证的方法可用，提供各种优点和缺点。然而，由于测量振动信号的随机性以及振动结构、测量过程和分析之间的复杂关系，这些方法识别的模态参数往往具有很大的不确定性。这些原因可以归因于随机效应(情感不确定性)或不完全知识(认知不确定性)，以及两者的组合(多态不确定性)。该研究方案的目的是开发一种方法，用于识别和量化测量和分析过程中连续步骤输出量中的多态不确定性，从测量的振动信号到识别的系统和模态参数。这包括为这些产出量制定评价标准。另一个目标是确定多个因素对识别的模态参数的不确定性的相互作用和影响的量化，并通过优化控制因素来减少它们。方法的开发是基于测量和分析过程的参数模型，并考虑到多态的不确定性。该项目将主要使用合成振动信号。在此背景下，考虑了两种成熟的运行模式分析的参数方法。此外，还运用了信号理论、机器学习、多态不确定性建模、不确定性量化(蒙特卡罗方法和优化方法)和灵敏度分析等方法。通过所提出的方法，支持在基于通过运行模态分析确定的模态参数的进一步调查的背景下的决策。在过程的每个步骤中，从测量到识别模态参数，可以根据客观标准对中间结果进行评估和优化，以获得最小不确定度的结果。该方法的设计方法基本上可以转移到实际的测量和分析过程中。

The experimental identification of the vibrational characteristics of mechanical structures by modal parameters under ambient excitation is the aim of Operational Modal Analysis. For this purpose, a number of reliable and well-proven methods based on formulations both in frequency and time domain are available offering various advantages and disadvantages. However, due to the stochastic nature of measured vibration signals and complex relations between a vibrating structure, measurement process and analyses, the modal parameters identified by these methods are often subject to significant uncertainties. These can causally be attributed to random effects (aleatory uncertainty) or incomplete knowledge (epistemic uncertainty), as well as to combinations of both (polymorphic uncertainty). The aim of this research proposal is to develop a methodology for the identification and quantification of polymorphic uncertainties in the output quantities of consecutive steps in the measurement and analysis process, starting from measured vibration signals to identified systems and modal parameters. This includes the development of evaluation criteria for these output quantities. Another aim is the determination of the interaction and the quantification of influences of multiple factors on the uncertainty of identified modal parameters, as well as their reduction by optimization of the governing factors. The development of methods is based on a parametric model of the measurement and analysis process taking polymorphic uncertainties into account. Primarily, synthetic vibration signals will be used in this project. Within this context, two well-established parametric methods of Operational Modal Analysis are considered. Moreover, methods from signal theory, machine learning, polymorphic uncertainty modelling, uncertainty quantification (Monte Carlo methods and optimization procedures) and sensitivity analysis are applied. By means of the proposed methodology, decisions in context with further investigations based on modal parameters identified by means of an Operational Modal Analysis are supported. In every step of the process, from measurement to the identification of modal parameters, intermediate results can be evaluated and optimized on the basis of objective criteria to achieve results with minimal uncertainties. The methodology is designed in such a way that it can basically be transferred to real measurement and analysis processes.