Frequency-dependent reconstruction of imbalances. ETNA - Electronic Transactions on Numerical Analysis

Imbalances in rotating machines cause vibrations of the system and may lead to an early wearout of the machine. In this paper, we consider the development of an algorithm for the detection (and subsequent correction) of imbalances from vibrational measurements at certain nodes of the system. Since, e.g., modern wind turbines operate with variable speed, the vibration data are usually collected during changing rotational speed. Based on a mathematical model that connects the measured vibrations at different rotational speeds with the imbalance distribution, we propose an algorithm for its reconstruction. The reconstruction algorithm is based on a tensor product formulation of the forward model. Test examples with artificial data are used to verify our approach