A Multi-Scale Temporal Convolutional Network Approach for Remaining Useful Life Prediction of Rolling Bearings
Journal Title: Precision Mechanics & Digital Fabrication - Year 2025, Vol 2, Issue 1
Abstract
Rolling bearings, as key components of rotating machinery, play a crucial role in the reliable operation of equipment. Over time, rolling bearings inevitably experience wear and fatigue, leading to damage. Accurate prediction of their Remaining Useful Life (RUL) is of paramount importance. This paper proposes an RUL prediction model based on the Multi-Scale Temporal Convolutional Network (MSTCN). The model effectively integrates both time-domain and frequency-domain information from bearing vibration signals through a multi-scale feature extraction module, enabling it to capture feature representations at different time scales. Additionally, the MSTCN's powerful temporal modeling capabilities allow it to capture long-term dependencies and short-term fluctuations in the bearing degradation process. Experimental results show that, compared to traditional methods, the proposed MSTCN model significantly improves the accuracy and stability of RUL predictions on the PHM2012 bearing dataset, demonstrating the effectiveness of the method in predicting the RUL of rolling bearings.
Authors and Affiliations
Tichun Wang, Qiji Teng
Keywords
Related Articles
- EP ID EP767844
- DOI https://doi.org/10.56578/pmdf020103
- Views 20
- Downloads 0
Finite Element Analysis of In-Service Loading on Hub Steering Knuckles: A Comparison of A356.0-T6 and Grey Cast Iron
This study investigates the structural response of a hub steering knuckle from a Toyota Camry LE under typical in-service loading conditions, with a focus on material performance comparison. Aluminium alloy A356.0-T6 and...
Design and Performance Analysis of a Torque-Based Optical Fiber Flow Sensor
A torque-based optical fiber flow sensor has been designed and experimentally tested to assess its potential for fluid flow measurement. The sensor utilizes an optical fiber strength modulation principle to achieve flow...
Identification of the Conduit Water Starting Time Constant in Hydropower Plants Using LSTM and MLP Machine Learning Algorithms
The accurate determination of the conduit water starting time constant (Tw) is critical for optimizing hydro turbine performance and dynamic control in hydropower plants. Instead of relying on conventional calculation me...
Reliability Analysis of Complex Repairable Systems Using Artificial Neural Networks: A Case Study on Underground Mining Machinery
The effective utilisation of equipment is essential for achieving the operational goals within production sectors, particularly in industries involving heavy machinery. Throughout its lifecycle, equipment is exposed to d...
Forward Kinematics Solution for Cable-Driven Hyper-Redundant Manipulators Based on BiLSTM
This paper investigates the kinematic solution of cable-driven hyper-redundant manipulators, focusing on the transformation from the cable-driven space to the joint space. Two forward kinematics solution networks based o...