Line-based process duration prediction using the vehicles features for the automotive industry

Öz

The automotive sector is an industry that acts as one of the locomotive among the manufacturing sectors in the world [1]. In these factories, daily, even hourly vehicle production speeds are important. Both production and supply logistics are very important for the vehicles to be produced and delivered to the end user. One of the most important factors in feeding the raw materials and semi-finished products of the vehicles coming off the line in internal logistics is the time they spend on the lines and production times of the vehicles. The accurate determination of the production speed in different lines for each vehicle affects the planning of the entire factory. The traditional approach here is to determine a certain time for each job and to find the line times by summing up manually. However, as a result of unpredictable events, breakdowns and processes that have not been performed before, deviations in periods may occur. It has been observed that the data generated by process times in the automotive sector can take extreme values, therefore, it is important to remove them from the model. In this study, unlike the traditional approach, using regression, regression and classification trees, and fully connected artificial neural networks, which are among statistical and artificial intelligence techniques, the waiting times of the vehicle, process and buffer sections of the source, assembly, dye-house sections were estimated. 61 different models were created for the predictions and it is seen that the classification trees technique performed the lowest mean absolute percent (MAPE), mean absolute (MAE) and mean square errors (MSE) in the lines other than the assembly transit and the welding buffer upper floor. While the estimation of duration in assembly lines was obtained with an average of 7.42% MAPE, it was determined with an average MAPE of 21.24% for dye-house and 22.49% MAPE for welding lines. Since the features specific to the vehicle are used, it has been determined that the time estimation in the buffer zones where the vehicles are kept is realized with an average of 82.89% MAPE except for the assembly lower floor buffer. The determination of a value higher than 1000% MAPE of the assembly lower floor buffer showed that the duration estimation of this zone was not suitable.

Anahtar Kelimeler

• Automotive industry
• regression and classification trees
• multiple linear regression
• artificial neural networks
• process duration prediction

Proje Numarası

1170357

Otomotiv endüstrisi için araç özelliklerini kullanarak proses sürelerinin hat bazlı tahmini

Öz

Dünyadaki üretim sektörleri arasında lokomotif görevi gören sektörlerden biri de otomotivdir [1]. Bu sektöre hizmet eden fabrikalarda günlük, hatta saatlik araç çıkış sayıları önem arz etmektedir. Araçların üretilip son kullanıcıya ulaştırılmasında, hem üretim hem de tedarik lojistiği kullanılmaktadır. Hattan çıkan araçların hammadde ve yarı mamullerinin iç lojistikte beslenmesindeki en önemli etkenlerden biri de araçların hatlarda geçirdikleri zaman ve üretim süreleridir. Her araç için farklı hatlardaki üretim hızının doğru belirlenmesi, fabrikanın planlamasını etkileyen faktörlerden biridir. Burada kullanılan klasik yaklaşım, her iş için belirli bir sürenin belirlenmesi ve bu sürelerin toplanarak hat sürelerinin bulunmasıdır. Fakat öngörülemeyen durumlar, daha önce gerçekleştirilmemiş işlemler ve arızalar nedeniyle, sürelerde sapmalar meydana gelebilmektedir. Beklenmeyen durumların gerçekleşmesi, otomotiv sektöründeki üretim sürelerinin oluşturduğu verilerin uç değerler alabileceği ve bu değerlerin de modelden çıkartılmasının gerekliliğini göstermektedir. Bu çalışmada klasik yaklaşımdan farklı olarak istatistik ve yapay zeka tekniklerinden regresyon, regresyon ve sınıflandırma ağaçları ile tam bağlı yapay sinir ağları kullanılarak kaynak, montaj, boyahane bölümlerinin araç proses ve tampon bölümlerinin bekleme sürelerinin tahmini gerçekleştirilmiştir. Yapılan tahminler için 61 farklı model oluşturulmuş ve montaj transit ile kaynak tampon üst kat dışındaki hatlarda en düşük ortalama mutlak yüzdesel (OMYH), ortalama mutlak ve ortalama karesel hatalara sahip tekniğin sınıflandırma ağaçları olduğu görülmüştür. Montaj hatlarında süre tahmini ortalama %7,42 OMYH ile elde edilirken, boyahane hatları için ortalama %21,24 OMYH ve kaynak hatlarında için ortalama %22,49 OMYH ile belirlenmiştir. Araca mahsus öznitelikler kullanıldığından, araçların bekletildiği tampon bölgelerde süre tahmininin montaj alt kat haricinde ortalama %82,89 OMYH ile gerçekleştiği tespit edilmiştir. Montaj alt kat tampon bölgesinin %1000 OMYH değerinden daha yüksek değer belirlenmesi, bu alanın süre tahmininin uygun olmadığını göstermiştir.

Anahtar Kelimeler

• Otomotiv endüstrisi
• regresyon ve sınıflandırma ağaçları
• çoklu doğrusal regresyon
• yapay sinir ağları
• proses süre tahmini

Destekleyen Kurum

Türkiye Bilimsel ve Teknolojik Araştırma Kurumu, Ford Otosan A.Ş.

Proje Numarası

1170357

Kaynakça

1. [1] T.C. Kalkınma Bakanlığı, “Onbirinci Kalkınma Planı - Otomotiv Sanayii Çalışma Grubu Raporu (2019 - 2023).”
2. [2] A. Cornet et al., “RACE 2050 – a vision for the European automotive industry,” 2019.
3. [3] Turkish Automotive Manufacturer Association, “General and Statistical Information Bulletin Of Automotive Manufacturers,” 2020.
4. [4] I. O. of M. V. Manufacturers, “2018 World Motor Vehicle Production Statistics,” 2019. [Online]. Available: http://www.oica.net/category/production-statistics/2018-statistics/.
5. [5] P. Pavlínek, “The Internationalization of Corporate R&D and the Automotive Industry R&D of East-Central Europe,” Economic Geography, vol. 88, no. 3, pp. 279–310, Jul. 2012.
6. [6] E. R. Badillo, F. L. Galera, and R. Moreno Serrano, “Cooperation in R&D, firm size and type of partnership,” European Journal of Management and Business Economics, vol. 26, no. 1, pp. 123–143, Jul. 2017.
7. [7] W. Chamsuk, W. Fongsuwan, and J. Takala, “The Effects of R&D and Innovation Capabilities on the Thai Automotive Industry Part’s Competitive Advantage: A SEM Approach,” Management and Production Engineering Review, vol. 8, no. 1, pp. 101–112, Mar. 2017.
8. [8] G. Mordue and D. Karmally, “Frontier Technologies in Non-Core Automotive Regions: Autonomous Vehicle R&D in Canada,” Canadian Public Policy, vol. 46, no. 1, pp. 73–93, Mar. 2020.

9. [9] F. Mogge,K. O. Fritz,T. Schlick,C. Söndermann, F. Daniel, “Global Automotive Supplier Study 2019,” Munich, 2019.
10. [10] M.E.Salveson,“On a Quantitative Method in Production Planning and Scheduling,” Econometrica,20,4,554,Oct. 1952.
11. [11] S. Dauzère-Péres and J.-B. Lasserre, An Integrated Approach in Production Planning and Scheduling, vol. 411. Berlin, Heidelberg: Springer Berlin Heidelberg, 1994.
12. [12] D. Terekhov, T. T. Tran, D. G. Down, and J. C. Beck, “Integrating Queueing Theory and Scheduling for Dynamic Scheduling Problems,” Journal of Artificial Intelligence Research, vol. 50, pp. 535–572, Jul. 2014.
13. [13] K. M. M. A. Bukkur, Shukri M.I., and O. M. Elmardi, “A Review for Dynamic Scheduling in Manufacturing,” The Global Journal of Researches in Engineering, vol. 18, no. 5-J, pp. 25–37, 2018.
14. [14] M. Vlk and R. Bartak, “Replanning in Predictive-reactive Scheduling,” in International Conference on Automated Planning and Scheduling, 2015.
15. [15] J. P. MacDuffie, K. Sethuraman, and M. L. Fisher, “Product Variety and Manufacturing Performance: Evidence from the International Automotive Assembly Plant Study,” Management Science, vol. 42, no. 3, pp. 350–369, Mar. 1996.
16. [16] L. Felipe Scavarda, J. Schaffer, A. José Scavarda, A. da Cunha Reis, and H.Schleich, “Product variety: an auto industry analysis and a benchmarking study,” Benchmarking: An International Journal, vol. 16, no. 3, pp. 387–400, May 2009.
17. [17] A. Fysikopoulos, D. Anagnostakis, K. Salonitis, and G. Chryssolouris, “An Empirical Study of the Energy Consumption in Automotive Assembly,” Procedia CIRP, vol. 3, pp. 477–482, 2012.
18. [18] C. Galitsky and E. Worrell, “Energy Efficiency Improvement and Cost Saving Opportunities for the Vehicle Assembly Industry: An ENERGY STAR® Guide for Energy and Plant Managers,” 2008.
19. [19] A. R. Rahani and M. Al-Ashraf, “Production Flow Analysis through Value Stream Mapping: A Lean Manufacturing Process Case Study,” Procedia Engineering, vol. 41, pp. 1727–1734, 2012.
20. [20] J. Mukund Nilakantan, G. Q. Huang, and S. G. Ponnambalam, “An investigation on minimizing cycle time and total energy consumption in robotic assembly line systems,” Journal of Cleaner Production, vol. 90, 311–325, Mar. 2015.
21. [21] R. E. Walpole, R. H. Myers, S. L. Myers, and K. Ye, Probability & Statistics for Engineers & Scientists, Global Edition, 9th ed. Pearson Education Limited, 2017.
22. [22] XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.
23. [23] S. M. Ross, Introduction to Probability and Statistics for Engineers and Scientists, 5th ed. Academic Press, 2014.
24. [24] XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.
25. [25] S. Makridakis, S. C. Wheelwright, and H. R. J., Forecasting: methods and applications. New York: John Wiley, 2008.
26. [26] S. Delurgio and C.Bhame, Forecasting Systems for Operations Management. New York:Irwin Professional Pub, 1991.
27. [27] A. Yılmaz, “Granüler yol malzemeleri için regresyon yöntemiyle Esneklik modülü (Mr) tahmin modeli geliştirilmesi,” Journal of the Faculty of Engineering and Architecture of Gazi University, vol. 35, no. 1, pp. 507–518, Oct. 2019.
28. [28] N. O’Rourke, L. Hatcher, and E. J. Stepanski, A Step-by-Step Approach to Using SAS for Univariate & Multivariate Statistics. North Calorina: SAS Institute Inc., 2005.
29. [29] L. Rokach and O. Maimon, Data Mining with Decision Trees: Theory and Applications, 2nd ed. MA, USA: World Scientific Publishing, 2015.
30. [30] L. Breiman, J. H. Friedman, R. A. Olshen, and C. J. Stone, Classification And Regression Trees. NY: Routledge, 1984.
31. [31] B. Choubin, G. Zehtabian, A. Azareh, E. Rafiei-Sardooi, F. Sajedi-Hosseini, and Ö. Kişi, “Precipitation forecasting using classification and regression trees (CART) model: a comparative study of different approaches,” Environmental Earth Sciences, vol. 77, no. 8, p. 314, Apr. 2018.
32. [32] R. Timofeev, “Classification and Regression Trees (CART) Theory and Applications,” Humboldt University, 2004.
33. [33] J. Han, M.Kamber, and J.Pei, Data Mining: Concepts and Techniques. MA,USA:Morgan Kaufmann Publishers, 2012.
34. [34] A. S. Ahmad et al., “A review on applications of ANN and SVM for building electrical energy consumption forecasting,” Renewable and Sustainable Energy Reviews, vol. 33, pp. 102–109, May 2014.
35. [35] H. Zhao and F. Magoulès, “A review on the prediction of building energy consumption,” Renewable and Sustainable Energy Reviews, vol. 16, no. 6, pp. 3586–3592, Aug. 2012.
36. [36] H. G. Reşat, “Sürdürülebilir enerji yönetimi için yapay sinir ağları ve ARIMA metotları kullanılarak melez tahmin modelinin tasarlanması ve geliştirilmesi: Tütün endüstrisinde vaka çalışması,” Journal of the Faculty of Engineering and Architecture of Gazi University, vol. 35, no. 3, pp. 1129–1140, Apr. 2020.
37. [37] M. Bulut and B. Başoğlu, “Kısa Dönem Elektrik Talep Tahminleri İçin Yapay Sinir Ağları ve Uzman Sistemler Tabanlı Hibrid Tahmin Sistemi Geliştirilmesi,” Journal of the Faculty of Engineering and Architecture of Gazi University, vol. 32, no. 2, Jun. 2017.
38. [38] S. Toraman and İ. Türkoğlu, “Dalgacık dönüşümü ve makine öğrenme teknikleri kullanılarak FTIR sinyallerinden kolon kanseri hastaları ve sağlıklı kişileri sınıflandırmak için yeni bir yöntem,” Journal of the Faculty of Engineering and Architecture of Gazi University, vol. 35, no. 2, pp. 933–942, Dec. 2019.
39. [39] E. Özcan, T. Danışan, and T. Eren, “Hidroelektrik santrallarda bakım çizelgeleme için hibrid bir model önerisi,” Journal of the Faculty of Engineering and Architecture of Gazi University, vol. 35, no. 4, pp. 1815–1827, Jul. 2020.
40. [40] R. J. Schalkoff, Artificial Neural Networks, 1st ed. New York, New York, United States: McGraw Hill, 1997.
41. [41] S. Haykin, Neural Networks: A Comprehensive Foundation, Second Edi. New Jersey: Prentice Hall, 1998.
42. [42] K. Swingler, Applying Neural Networks: A Practical Guide, First Edit. New York: Academic Press, 1996.
43. [43] S. I. Inc., SAS/ETS® 13.2 User’s Guide. North Calorina: SAS Institute Inc., 2014.
44. [44] J. Salmeron and A. Ruiz-Celma, “Elliot and Symmetric Elliot Extreme Learning Machines for Gaussian Noisy Industrial Thermal Modelling,” Energies, vol. 12, no. 1, p. 90, Dec. 2018.
45. [45] D. L. Elliott, “A Better Activation Function for Artificial Neural Networks,” Maryland, 1993.
46. [46] XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.
47. [47] XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.
48. [48] C. D. Lewis, B. Green, and K. Sevenoaks, Industrial and business forecasting methods: A practical guide to exponential smoothing and curve fitting, First. London: Butterworth-Heinemann, 1982.
49. [49] M. Bulut and B. Başoğlu, “Development of a Hybrid System Based on Neural Networks and Expert Systems for Short-Term Electricity Demand Forecasting,” Journal of the Faculty of Engineering and Architecture of Gazi University, vol. 32, no. 2, Jun. 2017.
50. [50] M. Acı, M. Avcı, and Ç. Acı, “Reducing simulation duration of carbon nanotube using support vector regression method,” Journal of the Faculty of Engineering and Architecture of Gazi University, vol. 32, no. 3, pp. 901–907, Sep. 2017.
51. [51] XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.
52. [52] XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX.