Deep learning-based instantaneous cutting force modeling of three-axis CNC milling

Xie, Jiejun; Hu, Pengcheng; Chen, Jihong; Han, Wei; Wang, Ronghua

doi:10.1016/j.ijmecsci.2023.108153

Cited by 7 publications

(3 citation statements)

References 60 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Guo et al [15] established a multi-scale attention network and its interpretability was introduced from the aspects of structure design and feature extraction to realize the tool wear monitoring in high-speed milling. Xie et al [16] build the instantaneous cutting force model of three-axis CNC milling by using a deep learning network called Milling Force Convolutional Neural Network, which was able to be explained and interpreted well by involving comprehensive geometric information and mathematical operations in the mechanistic force model. Li et al [17] proposed a physics-informed meta learning for machining tool wear prediction by employing the empirical equations' parameters to improve the interpretability of the modeling.…”

Section: Introductionmentioning

confidence: 99%

Physics-informed interpretable machine learning method for DOC monitoring in peripheral milling

Li,

Zheng,

Jiang

et al. 2024

Int J Adv Manuf Technol

View full text Add to dashboard Cite

On-line monitoring of depths of cut (DOC) is an essential way to avoid machining defects, such as over-cutting and machining chatter. Data-driven machine learning algorithm play a principal role to realize the monitor. However, deficient training data and un-interpretable algorithm made it difficult to application. Therefore, a physics-informed interpretable machine learning algorithm was proposed. Firstly, a physics simulation model was established with DOC, rotation speed and feed speed as its inputs and time-domain signals of milling force as its outputs. The output force signals were quantitatively presented by the characteristic values of time domain, frequency domain and waveform parameters. Secondly, six dimensionless characteristics, namely the kurtosis, skewness, waveform factor, peak factor, pulse factor and margin factor of the resultant milling force, were explored through sensitivity analysis method. They were sensitive to DOC but insensitive to milling force coefficient, speed and feed speed. Then, a quantitative relationship model between features and DOC was established by using the least squares linear regression algorithm, which has an intrinsic interpretability. The model was trained by a labeled 100 groups test data. The experiments show that the accuracy of the proposed model for DOC monitoring is greater than 90%.

show abstract

Section: Introductionmentioning

confidence: 99%

Physics-informed interpretable machine learning method for DOC monitoring in peripheral milling

Li,

Zheng,

Jiang

et al. 2024

Int J Adv Manuf Technol

View full text Add to dashboard Cite

show abstract

“…, 2022). The quality of CNC machining is still favorable among other machining; it can achieve high accuracy and high quality of surface roughness (Xie et al. , 2023; Yadav et al.…”

Section: Introductionmentioning

confidence: 99%

“…CNC machining is widely used in many industrial sectors due to its high capability for machining materials such as plastic, wood and metals (Meher et al, 2022). The quality of CNC machining is still favorable among other machining; it can achieve high accuracy and high quality of surface roughness (Xie et al, 2023;Yadav et al, 2022).…”

Section: Introductionmentioning

confidence: 99%

Extreme roughness reduction and ultrafine quality of innovative dual function material extrusion 3D printer

Rashyid,

Jaya,

Muflikhun

2024

RPJ

View full text Add to dashboard Cite

Purpose This paper aims to use hybrid manufacturing (HM) to overcome several drawbacks of material extrusion three-dimensional (3D) printers, such as low dimension ranging from 0.2 to 0.5 µm, resulting in a noticeable staircase effect and elevated surface roughness. Design/methodology/approach Subtractive manufacturing (SM) through computer numerical control milling is renowned for its precision and superior surface finish. This study integrates additive manufacturing (AM) and SM into a single material extrusion 3D printer platform, creating a HM system. Two sets of specimens, one exclusively printed and the other subjected to both printing and milling, were assessed for dimension accuracy and surface roughness. Findings The outcomes were promising, with postmilling accuracy reaching 99.94%. Significant reductions in surface roughness were observed at 90° (93.4% decrease from 15.598 to 1.030 µm), 45° (89% decrease from 26.727 to 2.946 µm) and the face plane (71% decrease from 12.176 to 3.535 µm). Practical implications The 3D printer was custom-built based on material extrusion and modified with an additional milling tool on the same gantry. An economic evaluation based on cost-manufacturing demonstrated that constructing this dual-function 3D printer costs less than US$560 in materials, offering valuable insights for researchers looking to replicate a similar machine. Originality/value The modified general 3D printer platform offered an easy way to postprocessing without removing the workpiece from the bed. This mechanism can reduce the downtime of changing the machine. The proven increased dimension accuracy and reduced surface roughness value increase the value of 3D-printed specimens.

show abstract

Deep learning based automated fracture identification in material characterization experiments

Karathanasopoulos,

Hadjidoukas

2024

Advanced Engineering Informatics

View full text Add to dashboard Cite

Deep learning-based instantaneous cutting force modeling of three-axis CNC milling

Cited by 7 publications

References 60 publications

Physics-informed interpretable machine learning method for DOC monitoring in peripheral milling

Physics-informed interpretable machine learning method for DOC monitoring in peripheral milling

Extreme roughness reduction and ultrafine quality of innovative dual function material extrusion 3D printer

Deep learning based automated fracture identification in material characterization experiments

Contact Info

Product

Resources

About