Error assessment of laser cutting predictions by semi-supervised learning

Abstract

Experimentation data of perspex glass sheet cutting, using CO₂ laser, with missing values were modelled with semi-supervised artificial neural networks. Factorial design of experiment was selected for the verification of orthogonal array based model prediction. It shows improvement in modelling of edge quality and kerf width by applying semi-supervised learning algorithm, based on novel error assessment on simulations. The results are expected to depict better prediction on average by utilizing the systematic randomized techniques to initialize the neural network weights and increase the number of initialization. Missing values handling is difficult with statistical tools and supervised learning techniques; on the other hand, semi-supervised learning generates better results with the smallest datasets even with missing values.

Keywords

semi-supervised learning / training algorithm / kerf width / edge quality / laser cutting process / artificial neural network (ANN)

References

[1] ZaidiM. Laser cutting parameters optimization of polystyrene foam using Taguchi method and artificial neural network tools [D]. Computer Science, SZABIST, Karachi, 2010

[2] ZaidiM, AminI, HussainA, YusoffN. Estimation of ANN modelling of laser cutting with missing values [J]. Pensee Journal, 2013, 75(11): 159-171

[3] PandeyK, DubeyA K. Neuro fuzzy modeling of laser beam cutting process [J]. Applied Mechanics and Materials, 2012, 110: 4109-4117

[4] Sharma, YadavaV. Modelling and optimization of cut quality during pulsed Nd:YAG laser cutting of thin Al-alloy sheet for straight profile [J]. Optics & Laser Technology, 2012, 44: 159-168

[5] LugerG F, StubblefieldW AArtificial intelligence: Structures and strategies for complex problem solving [M], 1998Third edUSA, Addion Wesley Longman: 663-668

[6] AndersonD, McneillGArtificial neural networks technology [M], 1992, Utica, New York, Kaman Sciences Corporation: 31

[7] ChavezAArtificial neural networks [M], 2010, Northridge, California, Maths, California State University: 14-15

[8] ShihaniN, et al. . Modeling of extrusion process using response surface methodology and artificial neural networks [J]. Journal of Engineering Science and Technology, 2006, 1(1): 31-40

[9] Sivarao, et al. . Neural network multi layer perceptron modeling of surface quality in laser machining [C]. International Conference on Intelligent and Advance Systems. Kuala Lumpur, Malaysia, 200781-86

[10] NguyenD, WidrowB. Improving the learning speed of 2-layer neural networks by choosing initial values of the adaptive weights [C]. IJCNN International Joint Conference on Neural Networks. San Diego, CA, USA, 199021-26

[11] QianN. On the momentum term in gradient descent learning algorithms [J]. Neural Networks, 1999, 12: 145-151

[12] HayatiM, ShirvanyY. Artificial neural network approach for short term load forecasting for illam region [J]. International Journal of Electrical, Computer and Systems Engineering, 2007, 1(4): 659-663

[13] HarshanandK, Ghuge, RegulwarD G. Artificial neural network method for estimation of missing data [J]. International Journal of Advanced Technology in Civil Engineering, 2013, 2(1): 1-4

[14] ZaidiM, AmznI, YusoffN B, HaiderN. Laser cutting quality control of melamine using artificial neural networks [C]. APIEMS: Asia Pacific Industrial Engineering and Management Systems Conference. Melaka, Malaysia, 2010

[15] ZaidiM, Bushra SaeedA, AminI, Nukman Yusoff. Experimental data mining techniques (using multiple statistical methods) [J]. IJCSI International Journal of Computer Science Issues, 2012, 3: 132-138

[16] PontesF J, FerreiraJ R, SilvaM B, PaivaA P, BalestrassiP P. Artificial neural networks for machining processes surface roughness modelling [J]. The International Journal of Advanced Manufacturing Technology, 2010, 49: 879-902

[17] ZainA M, HaronH, SharifS. Review of ANN technique for modeling surface roughness performance measure in machining process [C]. Third Asia International Conference on Modelling & Simulation. Bandung, Bali, Indonesia, 200935-39

[18] MontgomeryD CDesign and analysis of experiments [M], 1991Third edNew York, John Wiley & Sons: 11-17

[19] TanS B, NukmanY. A study of laser engraving on Ceramic material [D]. Department of Engineering Design and Manufacture, Faculty of Engineering, University of Malaya, Kuala Lumpur, 2008

[20] Zaman, ZaidiM. Evaluation of training algorithms for the laser cutting process [C]. SZABIST 17th National Research Conference on Management, Computing and Social Sciences & Economics, Karachi, Pakistan, 20111-13