Tool Wear Prediction Based on DRSN-BiLSTM Model
Release time:2025-02-14
Hits:
- DOI number:
- 10.3901/JME.2024.24.066
- Journal:
- JOURNAL OF MECHANICAL ENGINEERING
- Key Words:
- tool wear；multi-channel fusion；deep learning；DRSN；BiLSTM
- Abstract:
- he prediction of tool wear in CNC machine tools is of great significance to improve the safety of tool processing and product processing quality. With advances in sensing technology, the amount of sensor data used for equipment condition monitoring in the manufacturing industry has exploded. This has led to a high priority for a deep learning-centric, data-driven approach in the field of tool wear prediction. However, it remains a challenge to accurately identify and extract features closely related to tool degradation and make the most of this information to improve the performance of predictive models. In order to solve the above problems, a tool wear prediction method based on deep learning was studied, and the multi-channel screening mechanism was applied to the prediction model, and a tool wear prediction method based on the deep residual shrinkage network-bidirectional long short term memory (DRSN-BiLSTM) model was proposed. According to the fluctuation degree of the monitoring signal, multiple channels related to the tool degradation height are selected for fusion, the convolutional channel attention mechanism is used to fuse the multi-channel data and efficiently mine the abstract feature information of each channel, and then the BiLSTM regression model is established to extract the time series information related to the tool wear to accurately predict the tool wear. Experiments verify the effectiveness of the channel screening mechanism and the accuracy of the prediction model.
- Co-author:
- LI Jinyang,LIU Yaxiong
- First Author:
- wanghaitao,SHI Weichun
- Indexed by:
- Journal paper
- Correspondence Author:
- Lichen Shi
- Discipline:
- Engineering
- Document Type:
- J
- Volume:
- 60
- Issue:
- 24
- Page Number:
- 66-74
- Translation or Not:
- no
- Date of Publication:
- 2024-01-01
- Included Journals:
- EI