Identification of Unbalance in a Rotating System Using Artificial Neural Networks

Rotating systems are one of the most common systems that transfer power in several industrial sectors. Unbalance is a severe fault that contributes to machine downtime and unscheduled maintenance actions and can damage crucial rotary systems such as turbines and compressors. Estimation of unbalance in rotor bearing systems is crucial for safe and efficient operation of the machine. The current work introduces a method that can identify the mass, angle, and location of unbalance using Artificial Neural Networks (ANNs). The inputs of the neural networks are some signal features derived from the bearing displacement signals. To generate data for the ANN training, a finite element model of a multidisc rotor bearing system supported by two journal bearings is used to simulate different unbalances (mass and angle) in the disks along the shaft. Regarding the calculation of the dynamic coefficients (stiffness and damping) of the Fluid Film Bearing, the Finite Difference Method is used for the solution of the Reynolds equation. The training data of the ANN consist of several unbalance masses and angles at two disks for 1500 rpm rotor speed. Feature selection methods used to provide the best signal features for the creation of the ANN. The results show that the ANN can accurately estimate the unbalance mass and angle of rotor. The ANN models of the current work can be part of an online condition monitoring system for fault diagnosis of rotating systems.