Surface Roughness Prediction and Process Parameter Optimization for TB6 Titanium Alloy Boring Considering Tool Wear

doi:10.3969/j.issn.1004-132X.2026.08.016

Abstract

Abstract:

To address the issues of severe tool wear， poor surface quality， and low machining efficiency during the boring process of TB6 titanium alloy， has been conducted on the prediction of surface roughness considering tool wear and the optimization of process parameters for TB6 titanium alloy boring. Firstly， the factors influencing the surface roughness of TB6 titanium alloy boring were analyzed. Secondly， a tool wear monitoring model was established by integrating the 1D convolutional neural network， multi-head attention mechanism， and bidirectional gated recurrent unit network. On this basis， a surface roughness prediction model was established using the support vector machine regression algorithm based on tool wear monitoring data and process parameters. Then， a process parameter optimization model was established with surface roughness and machining time as the objectives. And the model was solved using an improved multi-objective Harris hawks optimization（MOHHO） algorithm to obtain the optimal combination of process parameters that comprehensively considers both surface roughness and machining time. The case study results show that the surface roughness value is reduced by 10.68% and the machining time is shortened by 17.14% after the optimization， verifying the effectiveness of the method.

Key words: TB6 titanium alloy, boring process, surface roughness prediction, process parameter optimization, tool wear

摘要：

针对TB6钛合金镗削加工中刀具磨损严重、表面质量差、加工效率低等问题，开展了考虑刀具磨损的TB6钛合金镗削表面粗糙度预测及工艺参数优化研究。首先分析TB6钛合金镗削表面粗糙度的影响因素，然后结合一维卷积神经网络、多头注意力机制和双向门控循环神经网络建立刀具磨损监测模型；在此基础上，基于刀具磨损监测数据和工艺参数，利用支持向量机回归算法建立表面粗糙度预测模型；最后建立以表面粗糙度和加工时间为目标的工艺参数优化模型，并利用改进多目标哈里斯鹰优化算法求解，获取综合考虑表面粗糙度和加工时间的最优工艺参数组合。案例研究结果表明，优化后表面粗糙度减小了10.68%，加工时间缩短了17.14%，验证了所提方法的有效性。

关键词: TB6钛合金, 镗削加工, 表面粗糙度预测, 工艺参数优化, 刀具磨损

CLC Number:

TH161

LE Yuxin, LI Congbo, HUANG Kanghua, WANG Guangping, SU Li, AN Qingqiang. Surface Roughness Prediction and Process Parameter Optimization for TB6 Titanium Alloy Boring Considering Tool Wear[J]. China Mechanical Engineering, 2026, 37(8): 1965-1975.

乐庾忻, 李聪波, 黄康华, 汪广平, 苏栗, 安庆强. 考虑刀具磨损的TB6钛合金镗削表面粗糙度预测及工艺参数优化[J]. 中国机械工程, 2026, 37(8): 1965-1975.

Figures/Tables 25

Fig.1 Influencing factors of surface roughness in TB6 titanium alloy boring

Fig.2 Data acquisition for the TB6 titanium alloy boring experimental

Fig.3 Working principle of the tool wear monitoring model

Fig.4 Schematic architecture of the BiGRU

Fig.5 Flowchart of the TC-MOHHO

Fig.6 Data acquisition equipment setup

Fig.7 Tool wear captured by industrial cameras

Fig. 8 Measurement of boring surface roughness

Tab. 1 Factor level table for the tool wear monitoring experiment

因素水平	n/（r·min $- 1$ ）	f_z/mm	a_p/mm
1	130	0.05	0.025
2	240	0.1	0.055
3	350	0.15	0.085

Tab. 1 Factor level table for the tool wear monitoring experiment

因素水平	n/（r·min $- 1$ ）	f_z/mm	a_p/mm
1	130	0.05	0.025
2	240	0.1	0.055
3	350	0.15	0.085

Tab. 2 Selected tool wear monitoring experiment results

序号	镗孔个数s	n/（r·min $- 1$ ）	f_z/mm	a_p/mm	V_B /μm
1	4	130	0.05	0.085	16.368
2	8	130	0.05	0.085	22.812
3	12	130	0.05	0.085	28.393
︙	︙	︙	︙	︙	︙
53	20	350	0.05	0.055	30.455
54	24	350	0.05	0.055	32.695
55	28	350	0.05	0.055	34.286
︙	︙	︙	︙	︙	︙
106	40	130	0.1	0.085	59.574
107	44	130	0.1	0.085	69.255
108	48	130	0.1	0.085	79.662

Tab. 2 Selected tool wear monitoring experiment results

序号	镗孔个数s	n/（r·min $- 1$ ）	f_z/mm	a_p/mm	V_B /μm
1	4	130	0.05	0.085	16.368
2	8	130	0.05	0.085	22.812
3	12	130	0.05	0.085	28.393
︙	︙	︙	︙	︙	︙
53	20	350	0.05	0.055	30.455
54	24	350	0.05	0.055	32.695
55	28	350	0.05	0.055	34.286
︙	︙	︙	︙	︙	︙
106	40	130	0.1	0.085	59.574
107	44	130	0.1	0.085	69.255
108	48	130	0.1	0.085	79.662

Tab.3 Parameters of the convolutional layers

卷积层	卷积核个数	卷积核大小	输入通道数	激活函数
C1	16	3	6	ReLU
C2	64	3	16	ReLU
C3	128	3	64	ReLU
C4	64	3	128	ReLU

Fig.9 Effectiveness of tool wear monitoring for test set

Fig.10 Comparison of tool wear monitoring results across different models in the test set

Tab.4 Evaluation indexes of each tool wear monitoring model

模型	R²	RMSE值	MAPE值	MAE值
模型M	0.9527	4.0981	0.0836	3.6042
模型1	0.8854	6.3767	0.1746	6.2291
模型2	0.8934	6.1487	0.1625	5.9490
模型3	0.9392	4.6447	0.0873	3.7225

Tab.5 Selected surface roughness prediction experiment results

序号	n/（r·min $- 1$ ）	f_z/mm	a_p/mm	V_BM/μm	Ra/μm
1	130	0.05	0.025	9.832	0.339
2	130	0.05	0.055	11.304	0.363
3	130	0.05	0.085	11.113	0.584
…	…	…	…	…	…
53	130	0.10	0.025	79.534	0.648
54	130	0.10	0.055	82.329	0.652
55	130	0.10	0.085	86.421	0.842
…	…	…	…	…	…
106	350	0.15	0.025	118.234	1.852
107	350	0.15	0.055	125.324	2.298
108	350	0.15	0.085	131.421	2.574

Tab.5 Selected surface roughness prediction experiment results

序号	n/（r·min $- 1$ ）	f_z/mm	a_p/mm	V_BM/μm	Ra/μm
1	130	0.05	0.025	9.832	0.339
2	130	0.05	0.055	11.304	0.363
3	130	0.05	0.085	11.113	0.584
…	…	…	…	…	…
53	130	0.10	0.025	79.534	0.648
54	130	0.10	0.055	82.329	0.652
55	130	0.10	0.085	86.421	0.842
…	…	…	…	…	…
106	350	0.15	0.025	118.234	1.852
107	350	0.15	0.055	125.324	2.298
108	350	0.15	0.085	131.421	2.574

Fig.11 Effectiveness of surface roughness prediction

Fig.12 The influence of tool wear and process parameters on surface roughness

Fig.13 Comparison of surface roughness prediction results across different models in the test set

Tab.6 Evaluation indexes of each surface roughness prediction model

模型	R²	RMSE值	MAPE值	MAE值
本文模型	0.953	0.145	0.100	0.105
SVR（s）模型	0.881	0.231	0.172	0.184
GPR模型	0.917	0.195	0.136	0.143
RFR模型	0.904	0.207	0.175	0.164

Fig.14 Partial algorithm parameter adjustment result

Fig.15 Comparison of Pareto solution sets of various algorithms

Tab.7 The specific values of each Pareto solution

Pareto解	Ra/μm	t/min	欧氏距离D	排名
1	0.292	16	0.775	20
2	0.327	15.4	0.694	19
3	0.357	14.9	0.627	17
4	0.389	14.4	0.561	15
5	0.402	13.9	0.495	13
6	0.409	13.5	0.443	10
7	0.432	12.8	0.354	7
8	0.463	12.4	0.310	4
9	0.527	12.1	0.296	2
10	0.588	11.9	0.303	3
11	0.619	11.6	0.292	1
12	0.686	11.3	0.312	5
13	0.744	11.1	0.340	6
14	0.806	10.9	0.375	8
15	0.879	10.8	0.424	9
16	0.925	10.7	0.454	11
17	0.983	10.6	0.494	12
18	1.042	10.5	0.535	14
19	1.104	10.4	0.578	16
20	1.180	10.3	0.632	18

Tab.8 Comparison of optimization results

优化方案	n/（r·min $- 1$ ）	f_z/mm	a_p/mm	Ra/μm	t/min
综合优化 Ra&T	242	0.068	0.075	0.619	11.6
单独优化Ra	202	0.065	0.025	0.292	16
单独优化T	346	0.146	0.078	1.180	10.3
经验方案	216	0.093	0.03	0.693	14

Tab.8 Comparison of optimization results

优化方案	n/（r·min $- 1$ ）	f_z/mm	a_p/mm	Ra/μm	t/min
综合优化 Ra&T	242	0.068	0.075	0.619	11.6
单独优化Ra	202	0.065	0.025	0.292	16
单独优化T	346	0.146	0.078	1.180	10.3
经验方案	216	0.093	0.03	0.693	14

Fig. 16 The results of IGD and HV metric

Tab.9 Comparison of optimization and experimental

方

案

Ra优化

结果/μm

Ra实验

结果/μm

误差/%

t优化

结果/min

t实验

结果/min

References 22

[1]	张美娟，南海，鞠忠强，等. 航空铸造钛合金及其成型技术发展［J］. 航空材料学报， 2016， 36（3）： 13-19.
	ZHANG Meijuan， Hai NAN， JU Zhongqiang， et al. Aeronautical Cast Ti Alloy and Forming Technology Development［J］. Journal of Aeronautical Materials， 2016， 36（3）： 13-19.
[2]	ZHAO Mingli， XUE Boxi， LI Bohan， et al. Ensemble Learning with Support Vector Machines Algorithm for Surface Roughness Prediction in Longitudinal Vibratory Ultrasound-assisted Grinding［J］. Precision Engineering， 2024， 88： 382-400.
[3]	WANG Jingshu， CHEN Tao， KONG Dongdong. Knowledge-based Neural Network for Surface Roughness Prediction of Ball-end Milling［J］. Mechanical Systems and Signal Processing， 2023， 194： 110282.
[4]	CAO Huajun， LIU Lei， WU Bo， et al. Process Optimization of High-speed Dry Milling UD-CF/PEEK Laminates Using GA-BP Neural Network［J］. Composites Part B： Engineering， 2021， 221： 109034.
[5]	LIU Weichao， WANG Pengyu， YOU Youpeng. Surface Roughness Prediction Using Multi-source Heterogeneous Data and Bayesian Quantile Regression in Milling Process［J］. Journal of Manufacturing Processes， 2023， 95： 446-460.
[6]	CHENG Minghui， JIAO Li， YAN Pei， et al. Prediction and Evaluation of Surface Roughness with Hybrid Kernel Extreme Learning Machine and Monitored Tool Wear［J］. Journal of Manufacturing Processes， 2022， 84： 1541-1556.
[7]	WANG Yahui， WANG Yiwei， ZHENG Lianyu， et al. Online Surface Roughness Prediction for Assembly Interfaces of Vertical Tail Integrating Tool Wear under Variable Cutting Parameters［J］. Sensors， 2022， 22（5）： 1991.
[8]	HE Zhaopeng， SHI Tielin， XUAN Jianping. Milling Tool Wear Prediction Using Multi-sensor Feature Fusion Based on Stacked Sparse Autoencoders［J］. Measurement， 2022， 190： 110719.
[9]	KUNTOĞLU M， SAĞLAM H. Investigation of Signal Behaviors for Sensor Fusion with Tool Condition Monitoring System in Turning［J］. Measurement， 2021， 173： 108582.
[10]	CAGGIANO A. Tool Wear Prediction in Ti-6Al-4V Machining through Multiple Sensor Monitoring and PCA Features Pattern Recognition［J］. Sensors， 2018， 18（3）： 823.
[11]	RUDRAPATI R， PAL P K， BANDYOPADHYAY A. Modeling and Optimization of Machining Parameters in Cylindrical Grinding Process［J］. The International Journal of Advanced Manufacturing Technology， 2016， 82（9）： 2167-2182.
[12]	LEONE C， GENNA S， TAGLIAFERRI F. Multiobjective Optimisation of Nanosecond Fiber Laser Milling of 2024 T3 Aluminium Alloy［J］. Journal of Manufacturing Processes， 2020， 57： 288-301.
[13]	JOSHI M， GHADAI R K， MADHU S， et al. Comparison of NSGA-II， MOALO and MODA for Multi-objective Optimization of Micro-Machining Processes［J］. Materials， 2021， 14（17）： 5109.
[14]	鄢威，王欣怡，张华，等. 考虑切削能耗和表面质量的碳纤维增强树脂基复合材料加工工艺参数优化决策［J］. 中国机械工程， 2024， 35（10）： 1834-1844.
	YAN Wei， WANG Xinyi， ZHANG Hua， et al. Optimization Decision of CFRP Processing Parameters Considering Cutting Energy Consumption and Surface Quality［J］. China Mechanical Engineering， 2024， 35（10）： 1834-1844.
[15]	KIRANYAZ S， AVCI O， ABDELJABER O， et al. 1D Convolutional Neural Networks and Applications： a Survey［J］. Mechanical Systems and Signal Processing， 2021， 151： 107398.
[16]	VOITA E， TALBOT D， MOISEEV F， et al. Analyzing Multi-head Self-Attention： Specialized Heads Do the Heavy Lifting， the Rest Can Be Pruned［C］∥Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. ACL， 2019： 5797-5808.
[17]	XU Ziwei， LI Yanfeng， HUANG Hongzhong， et al. A Novel Method Based on CNN-BiGRU and AM Model for Bearing Fault Diagnosis［J］. Journal of Mechanical Science and Technology， 2024， 38（7）： 3361-3369.
[18]	NAJM S M， PANITI I. Predict the Effects of Forming Tool Characteristics on Surface Roughness of Aluminum Foil Components Formed by SPIF Using ANN and SVR［J］. International Journal of Precision Engineering and Manufacturing， 2021， 22（1）： 13-26.
[19]	ZHU Haitao， GENG Guoqian， YU Yang， et al. Probabilistic Analysis on Parametric Random Vibration of a Marine Riser Excited by Correlated Gaussian White Noises［J］. International Journal of Non-Linear Mechanics， 2020， 126： 103578.
[20]	YÜZGEÇ U， KUSOGLU M. Multi-objective Harris Hawks Optimizer for Multi-objective Optimization Problems［J］. BSEU Journal of Engineering Research and Technology， 2020， 1（1）： 31-41.
[21]	ZHANG S J， TO S， WANG S J， et al. A Review of Surface Roughness Generation in Ultra-precision Machining［J］. International Journal of Machine Tools and Manufacture， 2015， 91： 76-95.
[22]	COELLO C A C， PULIDO G T， LECHUGA M S. Handling Multiple Objectives with Particle Swarm Optimization［J］. IEEE Transactions on Evolutionary Computation， 2004， 8（3）： 256-279.