میزان‌سازی روتور صلب چهار درجه آزادی با استفاده از یادگیری ماشین

There are several methods for balancing rotating equipment, each with various advantages and disadvantages. This research introduces a method for rotor balancing using machine learning. To this aim, the governing equations of motion for a rigid rotor with four degrees of freedom (DoF) under the influence of an unbalanced force have been extracted, and a simulator has been developed to study the dynamic behavior of the rotor. Subsequently, by employing this simulator in 2000 iterations, the vibration effect caused by the applying a random initial unbalance and then adding a random trial mass to the same rotor has been extracted in four points (at DoFs), thus creating a dataset. Then, using a feedforward neural network (FFNN), the possibility of detecting the value and the angular position of the initial unbalance based on the vibration data as well as the trial mass has been investigated. It has been shown that by using data from at least two out of the four points, a suitable model can be obtained to detect the value of initial unbalance. However, accurately detecting the angular position of unbalance with this data is not possible, and the machine learning model has a limited capacity for estimating the angular position.