Wyniki wyszukiwania - BazTech

Ograniczanie wyników

1 Diagnostyka

1 2025

Znaleziono wyników: 1

Liczba wyników na stronie

Wyniki wyszukiwania

Sortuj według:

Ogranicz wyniki do:

Diagnostics of vibrations due to looseness fault and unbalance in rotating machinery with neural network

Saeed Samed, Reddy Babu

Diagnostyka

2025

Vol. 26, No. 4

art. no. 2025403

Current work aims at the development and evaluation of Neural Network (NN) model for diagnosing faults due to unbalanced mass and structural looseness in an induction motor setup. These experiments were conducted with and without structural looseness. This data is processed in MATLAB software and is then used to train NN model to detect the unbalance, and structural looseness faults in the setup. The performance of the model trained is evaluated by model performance metrics, which showed the model predicts the presence of above faults with high accuracy. The, Kruskal Willis algorithm is used in MATLAB software to get the feature importance scores, so that, the number of predictors/features can be reduced. It is found that two mutually perpendicular radial accelerations of the setup have significant importance, and hence, a new NN model is trained with the reduced number of predictors/features. It was found that there is a slight reduction in the model’s performance, therefore, to increase the performance, another model is trained with the two mutually perpendicular radial accelerations, and their resultants. This increased the performance of the model considerably, hence making it suitable to deploy for the detection of unbalance, and structural looseness faults.