Research on the remaining useful life prediction method of rolling bearings based on TS-BiGRU
YANG Peng
XUE Hongwei
ZHANG Chao
CHEN Huini
YANG Xiaodong
LI Lu
Abstract：To address the limitations of current remaining useful life(RUL)prediction models for rolling bearings in long-sequence feature extraction and model generalization capability,this paper proposes a remaining life prediction method based on Temporal Convolution Network(TCN),SimAM module,and Bi-directional Gated Recurrent Unit(BiGRU).Firstly,TCN is used for short-term feature extraction,and its dilated convolutions and residual structures are utilized to efficiently model the input sequence.Subsequently,the SimAM module adaptively weights the extracted features to suppress redundant information and enhance key feature representation.Finally,a BiGRU is utilized to capture long-term dependencies within the input sequences by integrating forward and backward information,thereby improving the modeling capability for complex data.The proposed method was validated using the IEEE PHM 2012 dataset.Results demonstrate that it significantly improves both the accuracy of long-sequence prediction and its adaptability to various operating conditions.
Keywords：rolling bearingTCNBiGRUSimAMremaining useful life prediction
Publication Date：2025-09-20
Online Publishing Date：2025-10-28(First online date of this platform, not the publication date of the document)
Pages：9( 34-42 )