A tool wear monitoring method based on improved DenseNet and GRU

doi:10.23919/JSEE.2025.000124

Journal of Systems Engineering and Electronics ›› 2025, Vol. 36 ›› Issue (6): 1562-1578.doi: 10.23919/JSEE.2025.000124

• SYSTEMS ENGINEERING • Previous Articles

A tool wear monitoring method based on improved DenseNet and GRU

Yue WANG¹(), Yajie MA²(), Jiangnan ZHOU²(), Yanxia WU¹^,*()

¹ School of Computer Science and Technology, Harbin Engineering University, Harbin 150006, China
² College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China

Received:2024-09-19 Online:2025-12-18 Published:2026-01-07
Contact: Yanxia WU E-mail:563669810@qq.com;yajiema@nuaa.edu.cn;zhoujiangnan0899@163.com;wuyanxia@hrbeu.edu.cn
About author:
WANG Yue was born in 1987. He received his B.E. degree and M.S. degree in control engineering from Nanjing University of Aeronautics and Astronautics in 2010 and 2013, respectively. Currently, he is pursuing his Ph.D. degree from Harbin Engineering University. His research interests are intelligent manufacturing, industrial software and information security. E-mail: 563669810@qq.com

MA Yajie was born in 1987. He received his Ph.D. degree from Nanjing University of Aeronautics and Astronautics (NUAA) in 2015. He is currently a professor with the College of Automation Engineering, NUAA. His research interests are fault diagnosis and fault tolerant control. E-mail: yajiema@nuaa.edu.cn

ZHOU Jiangnan was born in 1999. She received her B.S. degree from the North China University of Water Resources and Electric Power in 2021. She is currently pursuing her M.S. degree at Nanjing University of Aeronautics and Astronautics. Her research interests are deep learning and mechanical equipment status detection. E-mail: zhoujiangnan0899@163.com

WU Yanxia was born in 1979. She received her M.S. and Ph.D. degrees from Harbin Engineering University (HEU) in 2005 and 2008, repectively. She is currently a professor with the College of Computer Science and Technology, HEU. Her research interests are embedded systems and the Internet of Things, compilation technologies, big data processing, and reconfigurable computing. E-mail: wuyanxia@hrbeu.edu.cn
Supported by:
This work was supported by the National Natural Science Foundation of China (62020106003; 62273177; 62233009), the Natural Science Foundation of Jiangsu Province of China (BK20222012), the Programme of Introducing Talents of Discipline to Universities of China (B20007), the Fundamental Research Funds for the Central Universities (NI2024001), and the National Key Laboratory of Space Intelligent Control (HTKJ2023KL502006).

Abstract

Abstract:

The precision and quality of machining in computer numerical control (CNC) machines are significantly impacted by the state of the tool. Therefore, it is essential and crucial to monitor the tool’s condition in real time during operation. To improve the monitoring accuracy of tool wear values, a tool wear monitoring approach is developed in this work, which is based on an improved integrated model of densely connected convolutional network (DenseNet) and gated recurrent unit (GRU), which incorporates data preprocessing via wavelet packet transform (WPT). Firstly, wavelet packet decomposition (WPD) is used to extract time-frequency domain features from the original time-series monitoring signals of the tool. Secondly, the multidimensional deep features are extracted from DenseNet containing asymmetric convolution kernels, and feature fusion is performed. A dilation scheme is employed to acquire more historical data by utilizing dilated convolutional kernels with different dilation rates. Finally, the GRU is utilized to extract temporal features from the extracted deep-level signal features, and the feature mapping of these temporal features is then carried out by a fully connected neural network, which ultimately achieves the monitoring of tool wear values. Comprehensive experiments conducted on reference datasets show that the proposed model performs better in terms of accuracy and generalization than other cutting-edge tool wear monitoring algorithms.

Key words: tool wear monitoring, densely connected convolutional network (DenseNet), asymmetric convolutional kernel, dilated convolutional kernel, gated recurrent unit (GRU)

Yue WANG, Yajie MA, Jiangnan ZHOU, Yanxia WU. A tool wear monitoring method based on improved DenseNet and GRU[J]. Journal of Systems Engineering and Electronics, 2025, 36(6): 1562-1578.

Figures/Tables 18

Fig 1

Fig 2

Fig 3

Table 1

Fig 4

Fig 5

Fig 6

Fig 7

Fig 8

Fig 9

Table 2

Table 3

Table 4

Table 5

Fig 10

Fig 11

Fig 12

Table 6

References 31

1	BAROROH D K, CHU C H, WANG L Systematic literature review on augmented reality in smart manufacturing: collaboration between human and computational intelligence. Journal of Manufacturing Systems, 2021, 61, 696- 711. doi: 10.1016/j.jmsy.2020.10.017
2	ANTIC A, POPOVIC B, KRSTANOVIC L, et al Novel texture-based descriptors for tool wear condition monitoring. Mechanical Systems and Signal Processing, 2018, 98, 1- 15. doi: 10.1016/j.ymssp.2017.04.030
3	LIU X L , ZHANG B W, LI X B, et al. An approach for tool wear prediction using customised DenseNet and GRU integrated model based on multi-sensor feature fusion [J]. Journal of Intelligent Manufacturing, 2022, 34: 885−902.
4	GUO J C, LI A H Advances in monitoring technology of tool wear condition. Tool Engineering, 2019, 53 (5): 3- 13.
5	CAO D L, SUN H B, ZHANG J D, et al In-process tool condition monitoring based on convolution neural network. Computer Integrated Manufacturing Systems, 2020, 26 (1): 74- 80.
6	REHORN A G, JIANG J, ORBAN P E State-of-the-art methods and results in tool condition monitoring: a review. The International Journal of Advanced Manufacturing Technology, 2005, 26, 693- 710. doi: 10.1007/s00170-004-2038-2
7	ELSHEIKH A, YACOUT S, OUALI M S Bidirectional handshaking LSTM for remaining useful life prediction. Neurocomputing, 2019, 323, 148- 156. doi: 10.1016/j.neucom.2018.09.076
8	WANG G F, LI Q M, QIN X D, et al Application of support-vector-machine in tool wear of multi-state monitoring. Journal of Tianjin University, 2011, 44 (1): 35- 39.
9	WANG X Q, ZHANG Y, ZHOU H M, et al Continuous tool wear based on hidden Markov model. Modular Machine Tool Automatic Manufacturing Technique, 2016, (10): 87- 90.
10	GUO H, REN B C, YAN X G, et al Research on dynamic prediction of tool life based on deep convolutional neural netwok. Control and Decision, 2022, 37 (8): 2119- 2126.
11	ZHAO R, YAN R Q, CHEN Z H, et al Deep learning and its applications to machine health monitoring. Mechanical Systems and Signal Processing, 2019, 115, 213- 237. doi: 10.1016/j.ymssp.2018.05.050
12	HINTON G E, SALAKHUTDINOV R R Reducing the dimensionality of data with neural networks. Science, 2006, 313 (5786): 504- 507. doi: 10.1126/science.1127647
13	AGHAZADEH F, TAHAN A, THOMAS M Tool condition monitoring using spectral subtraction and convolutional neural networks in milling process. The International Journal of Advanced Manufacturing Technology, 2018, 98, 3217- 3227. doi: 10.1007/s00170-018-2420-0
14	HUANG Z W, ZHU J M, LEI J , et al Tool wear predicting based on multi-domain feature fusion by deep convolutional neural network in milling operations. Journal of Intelligent Manufacturing, 2020, 31, 953- 966. doi: 10.1007/s10845-019-01488-7
15	CHEN Q P, XIE Q S, YUAN Q N, et al Realtime monitoring method for wear state of tool based on deep bidirectional GRU model. Computer Integrated Manufacturing Systems, 2020, 26 (7): 1782- 1793.
16	SUN H B, ZHANG J D, MO R, et al In-process tool condition forecasting based on a deep learning method. Robotics and Computer-Integrated Manufacturing, 2020, 64, 101924. doi: 10.1016/j.rcim.2019.101924
17	HUANG G, LIU Z, LAURENS V, et al. Densely connected convolutional networks. Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 4700−4708.
18	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition. Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770−778.
19	KRIZHEVSKY A, SUTSKEVER I, HINTON G E Imagenet classification with deep convolutional neural networks. Communications of the ACM, 2017, 60 (6): 84- 90. doi: 10.1145/3065386
20	CHEN X Y, XU P Research on tool wear monitoring method based on long short-term memory networks. Mechanical Manufacturing and Automation, 2023, 52 (3): 106- 111.
21	CHENG M S , WANG Y J , ZHANG X C, et al. Research on milling cutter wear using residual structure and convolutional neural networks. Mechanical Science and Technology, 2024, 43(8): 1403−1410. (in Chinese)
22	YAN R Q, GAO R X, CHEN X F Wavelets for fault diagnosis of rotary machines: a review with applications. Signal Processing, 2014, 96, 1- 15. doi: 10.1016/j.sigpro.2013.04.015
23	COIFMAN R R, WICKERHAUSER M V Entropy-based algorithms for best basis selection. IEEE Trans. on Information Theory, 1992, 38 (2): 713- 718. doi: 10.1109/18.119732
24	DING X H, GUO Y C, DING G G, et al. ACNET: strengthening the kernel skeletons for powerful CNN via asymmetric convolution blocks. Proc. of the IEEE/CVF International Conference on Computer Vision, 2019: 1911−1920.
25	IOFFE S, SZEGEDY C. Batch normalisation: accelerating deep network training by reducing internal covariate shift. Proc. of the International Conference on Machine Learning, 2015: 448−456.
26	CHADHA G S, PANARA U, SCHWUNG A, et al Generalised dilation convolutional neural networks for remaining useful lifetime estimation. Neurocomputing, 2021, 452, 182- 199. doi: 10.1016/j.neucom.2021.04.109
27	ZHAO D Z, WANG T Y, CHU F L Deep convolutional neural network based planet bearing fault classification. Computers in Industry, 2019, 107, 59- 66. doi: 10.1016/j.compind.2019.02.001
28	PHM Society. phm data challenge 2010. https://www.phmsociety.org/competition/phm/10.
29	QIAO H H, WANG T Y, WANG P, et al A time-distributed spatiotemporal feature learning method for machine health monitoring with multi-sensor time series. Sensors, 2018, 18 (9): 2932. doi: 10.3390/s18092932
30	LIU D H , WANG Y C , HE H Online tool wear monitoring based on temporal convolutional networks. Journal of Machine Tools & Automatic Manufacturing Technology, 2023, (4): 174- 176.
31	LI Z G, ZHOU X L, SUN P L A tool wear monitoring method that combines residual blocks and Swin-Transformer mechanism. Modern Manufacturing Engineering, 2024, (8): 126- 135.

Model	P_MAE			P_RMSE
Model	S1	S2	S3	S1	S2	S3
2-level WPT	5.76	12.91	20.67	6.85	16.51	25.55
3-level WPT	5.13	6.01	5.63	6.46	9.02	8.29
4-level WPT	7.19	16.29	28.43	9.01	19.16	22.52

Cutting tool			Three-flute tungsten carbide ball end milling tool
Data acquisition equipment			NI DAQ PCI 1200
Workpiece material			Stainless steel HRC52
Force sensor			Kistler 9265B 3-way Force Gauge
Vibration acceleration sensor			Kistler 8636C
Acoustic emission sensor			Kistler8152
Wear value measuring equipment			LEICI MZ12 Microscope

Data set symbol	Training set	Test set
S1	C4, C6	C1
S2	C1, C6	C4
S3	C1, C4	C6

Model	Modelling	P_RMSE
Model	Modelling	S1	S2	S3
ResNet	Input -ResNet-Dense-wear value	13.746	16.398	23.051
DenseNet-LSTM	Input-DenseNet-LSTM-Dense-wear value	18.638	20.715	20.185
DenseNet-GRU	Input-DenseNet-GRU-Dense-wear value	18.723	17.741	18.906
WPT-DenseNet	Input-WPT-DenseNet-Dense-wear value	12.670	15.578	16.419
WPDNet-GRU	Input-WPT-DenseNet-GRU-Dense-wear value	9.667	13.134	14.637
WPACDNet-GRU	Input-WPT-ACDNet-GRU-Dense-wear value	6.457	9.021	8.289

Model	Modelling	P_MAE
Model	Modelling	S1	S2	S3
ResNet	Input -ResNet-Dense-wear value	11.098	14.036	20.671
DenseNet-LSTM	Input-DenseNet-LSTM-Dense-wear value	15.606	16.389	18.190
DenseNet-GRU	Input-DenseNet-GRU-Dense-wear value	18.723	17.741	17.265
WPT-DenseNet	Input-WPT-DenseNet-Dense-wear value	10.704	11.247	11.040
WPDNet-GRU	Input-WPT-DenseNet-GRU-Dense-wear value	7.609	9.406	12.920
WPACDNet-GRU	Input-WPT-ACDNet-GRU-Dense-wear value	5.133	6.008	5.629

A tool wear monitoring method based on improved DenseNet and GRU

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

Share this article

Figures/Tables 18

References 31

Related Articles 2

Recommended Articles

Metrics

Comments

Model	P_MAE			P_MSE
Model	S1	S2	S3	S1	S2	S3
HMM	−	−	−	276.3	85.8	121.2
TDConvLSTM	6.99	6.96	7.50	69.39	70.39	104.4
TCN	6.12	7.44	5.08	56.69	93.43	45.37
HACDNet-GRU	5.85	7.80	6.33	51.98	86.68	48.16
Transformer	8.95	5.73	9.47	60.09	75.16	100.17
WPACDNet-GRU	5.13	6.01	5.63	41.69	81.38	68.71

[1]	Chen CHEN, Wei QUAN, Zhuang SHAO. Aerial target threat assessment based on gated recurrent unit and self-attention mechanism [J]. Journal of Systems Engineering and Electronics, 2024, 35(2): 361-373.
[2]	Caihua WU, Jianchao MA, Xiuwei ZHANG, Dang XIE. User space transformation in deep learning based recommendation [J]. Journal of Systems Engineering and Electronics, 2020, 31(4): 674-684.