Abstract
Power generation technologies are essential for modern economies. Modal Analysis (MA) is advanced but well-established method for monitoring of structural integrity of critical assets, including power ones. Apart from classical MA, the Operational Modal Analysis approach is widely used in the study of dynamic properties of technical objects. The principal reasons are its advantages over the classical approach, such as the lack of necessity to apply the excitation force to the object and isolate it from other excitation sources. However, for industrial facilities, the operational excitation rarely takes the form of white noise. Especially in the case of rotating machines, the presence of rotational speed harmonics in the response signals causes problems with the correct identification of the modal model. The article presents a hybrid approach where combination of results of two Operational Modal Analyses and Experimental Modal Analysis is performed to improve the models’ quality. The proposed approach was tested on data obtained from a 215 MW turbogenerator operating in one of Polish power plants. With the proposed approach it was possible to diagnose the machine’s excessive vibration level correctly.
Highlights
The more participation of renewable energy sources there is in the system, the more reliable sources are needed to stabilize the system in low-insolation or low-wind periods [2,3]
In the case of analyses performed in the time domain, there can be a distinguished group of methods based on the knowledge of system model in the form of a regression series, most often of the AutoRegressive Moving Average with eXogenous input (ARMAX)
Results from Operational Modal Analysis (OMA) in standard operational conditions, OMA during
Summary
Power generation systems are fundamental for modern societies in all countries across the world. In industrial conditions dynamic analysis is performed in most cases using OMA [23,24] This is because shutting down the machine in plants of continuous production leads to significant financial losses. The first method for separating deterministic and random components is Time Synchronous Averaging (TSA) [29] This allows a periodic signal to be extracted from the signal. In 1975, Widrow et al [32] presented the Adaptive Noise Cancelation (ANC) method which was able to separate a signal into deterministic and random components using an adaptive filter. The results of two independent OMAs and EMA are combined to identify the root cause of the defect It results in a much broader and thorough analysis than any single approach and can reduce the influence of operational excitations.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
