Bearing Life Prediction Model for Electromechanical Equipment by Integrating Deep Neural Network and K-Nearest Neighbor Algorithm and Its Application

Current life prediction methods of Electromechanical equipment bearings have issues of low accuracy and lack of stability. To address these problems, firstly, indicators based on life degradation characteristics of bearings are selected. Then, a deep neural network-based life prediction model is constructed. Finally, the K-nearest neighbor algorithm is introduced to correct the deviation of the deep neural network prediction model, and a hybrid life prediction model is designed. Results show that effectiveness of the designed model was better, which was of great practical significance for detecting bearing failures in advance, reducing equipment losses and improving equipment reliability.

---

Opracowanie rekordu ze środków MNiSW, umowa nr POPUL/SP/0154/2024/02 w ramach programu "Społeczna odpowiedzialność nauki II" - moduł: Popularyzacja nauki (2025).