Machine Learning Optimization Assignment Question

Topic Choice 2. Machine Learning Optimization (MLO) 2018 to present – Choose a business, a hobby, or an operation strategy (you are very familiar with), that is not utilizing machine learning optimization.
Research the theory and practice used by managers. Define the specific function/operation. Then design the concept paper and MLO strategy for the topic you selected. Your week 7 MLO Model must align with the strategy

Solution

Applying Machine Learning to Spacing in Punching of Sheets

Introduction

What makes machine learning (ML) so fascinating is its capacity to learn from prior experience. The algorithms can learn to grasp intricate relationships between the many parameters and their impact on production by analyzing enormous volumes of previous data collected from the platform’s sensors. In theory, the way operators learn to control a process is similar to how algorithms learn from experience. The machine learning algorithms, in contrast to a human operator, have no trouble assessing the complete historical records for hundreds of sensors over a number of years. Unlike a human brain, they have an infinite capacity for experience accumulation.

Machine learning optimization can therefore be applied to the spacing during the punching of sheets to allow for workers to walk between the heaps of punched sheets to pick up the small pieces that get left behind after punching. Such pieces usually promote workplace hazards and jeopardize the efficiency of the punching process.

Need for the Study

Competent designers are expected to have a high level of knowledge and expertise especially in the sheet metal industry. Diverse artificial intelligence techniques are being used in this industrial sector to acquire this level of competence with the hopes of lowering complexity, reducing the need for human labor, and enhancing operational excellence. One of the most effective methods for addressing engineering challenges, lowering complexity, limiting the need for human expertise, and shortening the time required for manufacturing processes is the machine learning algorithm, which is a component of artificial intelligence.

In the sheet metal sector, component quality is a crucial factor, yet it can be time-consuming to manually check for punching-related component flaws. Utilizing an image-based component inspection technique can help you prevent this. This approach of vision-based examination is regarded as a potent and long-lasting remedy (Ghatnekar 2018). The difficulty of identifying and categorizing component problems can be overcome by utilizing computer vision and machine learning. Using a picture to identify the sort of defect the components or parts have is thought to be a crucial method. The image recognition approach is commonly employed in industries for real-time applications (Ghatnekar 2018). The machine-learning approach can help in the identification of parts and the prediction of flaws in sheet metal forming operations (Marques et al., 2020). However, during the punching of sheets, a lot of small pieces of metal get left behind, which not only pose security risks to the workers but also threaten the accuracy of the process. Human workers need to enter through the space between the punched sheets and pick the pieces and safely dump them in the garbage cans. However, the space between the punched sheets is sometimes too small for a human to fit in, which makes the picking both hazardous and sometimes impossible. It is therefore necessary to use machine learning to determine a safe distance between the heaps of punched sheets through which the workers can squeeze in.

Background

There aren’t many machine learning optimization applications in the sheet metal industry. The sheet metal industry has successfully used machine learning and deep learning algorithms to choose an appropriate production process and to produce the final shape of a metal item that is unstructured and significantly dependent on human skills (Hamouche and Loukaides 2018; Chiu, Tsai, and T-l 2020). A fresh approach was put out by Stoerke et al. (2016) to improve the geometric correctness of sheet metal work parts. In order to improve the geometric accuracy of incremental sheet forming operations, they suggested using a machine learning model that utilizes reinforcement learning.

Additionally, sheet metal forming process flaws are predicted using machine learning (Dib, Ribeiro, and Prates 2018; Tsai and Chang 2018). In order to achieve a competitive edge, Kwitek (2016) showed how sheet metal manufacturing equipment can apply machine learning optimization. He concentrated on predictive maintenance using machine learning, which improved operations management. Sheet metal forming, which is regarded as an essential part of contemporary production, was briefly addressed in Hamouche and Loukaides’s (2018) application of machine learning. To identify the manufacturing process that created a part merely from the final shape, a machine learning approach was applied in their study for the first time. A high accuracy rate was established by implementing a mapping of the mean and Gaussian curvatures through machine learning. This automated the operational processes from the initial stage of design to manufacture, obviating the need for human experts to match each product with an appropriate forming method.

A survey of the uses of artificial neural networks in sheet metal work was done by Kashid and Kumar (2012). Pre-process supervised machine learning was used by Zwierzycki, Nicholas, and Thomsen (2018) to forecast and enhance sheet-forming tolerance and produce corrected fabrication models. In an expert system for sheet metal bending tooling, Lin and Chang (1995) proposed a model based on machine learning using neural networks. The authors of this study created a learning model using conditional features to create an expert system for choosing sheet metal bending tools. Wu et al. (1999) used a combination of machine learning and artificial neural network techniques to investigate surface flaws in sheet metal forging.

ML has shown success in a number of manufacturing-related applications. Monitoring is the primary use of ML, particularly in the areas of quality control, machine condition monitoring, fault detection, tool wear, optimization, etc (Wuest eal.2016). Additionally, ML is applied to image recognition in manufacturing, where products with damage are detected using the photos. These ML applications to various manufacturing and optimization issues show how broadly adaptable and applicable the ML technique is.

In the sheet metal sectors, in addition to machine learning algorithms, other algorithms are utilized for a variety of applications. For instance, when batching orders, linear programming can be utilized to reduce the cost of sheet metal punching (Herrmann and Delalio 2001). In order to optimize the geometrical parameters (such as die design, hammering sequence, blank holder pressure, etc.) in the sheet metal business, Kakandikar, Darade, and Nandedkar (2009) used genetic algorithms. The use of ranking algorithms to improve the quality of sheet metal formed parts was suggested by Ashokkumar et al. in 2020. Jiao and Xing (2018) employed a heuristic approach to analyze how parts, clamps, and supporting locators bend during assembly in the sheet metal sector. However, the technology has not been applied in the determination and manipulation of space between sheets during the punching and stamping process.

Objectives

This research will be conducted to ascertain whether machine learning can be applied in the maximization of the space between the punched sheets so that human workers can squeeze in and remove the scraps for safe disposal.

Research Questions and Hypothesis

This research will attempt to answer the question on whether machine learning can help widen the space between the arrangements of the punched sheets enough for humans to walk between them.

Research timeline

The research will be conducted over a period of three weeks. The first one week will be taken in developing an introduction and gathering enough background information. The background information will then be used in guiding the literature review which will take another three days. The gaps identified during the literature review will then be answered through a survey through a questionnaire which will take one week. The remaining two days will be used to write up a results and discussion section.

References

Ashokkumar, S. P., S. S. Sivam, S. Balasubramanian, and R. V. Nanditta. 2020. “Effects of Process Variables Optimization on the Quality of Parts Processed in High Speed Single Point Incremental Sheet Metal Forming by Ranking Algorithm.” Materials Today: Proceedings. https://doi.org/10.1016/j.matpr.2020.08.614. 

Chiu, M.-C., C.-D. Tsai, and L. T-l T-L. 2020. “An Integrative Machine Learning Method to Improve Fault Detection and Productivity Performance in a Cyber-physical System.” Journal of Computer Information Science and Engineering 20(2): 12. pages. https://doi.org/10.1115/1.4045663

Dib, M., B. Ribeiro, and P. Prates. 2018. “Model Prediction of Defects in Sheet Metal Forming Processes.” In: Pimenidis E., Jayne C. (eds) Engineering Applications of Neural Networks. EANN 2018. Communications in Computer and Information Science, vol 893. Springer, Cham. https://doi.org/10.1007/978-3-319-98204-5_14 

Ghatnekar, S. 2018. “Use Machine Learning to Detect Defects on the Steel Surface.” Available at: https://software.intel.com/content/www/us/en/develop/articles/use-machine-learning-to-detect-defects-on-the-steel-surface.html 

Hamouche, E., and E. G. Loukaides. 2018. “Classification and Selection of Sheet Forming Processes with Machine Learning.” International Journal of Computer Integrated Manufacturing 31: 921–932. doi:https://doi.org/10.1080/0951192X.2018.1429668

Herrmann, J. W., and D. R. Delalio. 2001. “Algorithms for Sheet Metal Nesting.” IEEE Transactions on Robotics and Automation 17: 183–190. doi:https://doi.org/10.1109/70.928563.  

Jiao, Z., and Y. Xing. 2018. “Clamping-sequence Optimisation Based on Heuristic Algorithm for Sheet-metal Components.” International Journal of Production Research 56: 7190–7200. doi:https://doi.org/10.1080/00207543.2017.1410245.  

Kakandikar, G. M., P. D. Darade, and V. M. Nandedkar. 2009. “Applications of Evolutionary Algorithms to Sheet Metal Forming Processes: A Review.” International Journal of Machine Intelligence 1: 47–49. doi:https://doi.org/10.9735/0975-2927.1.2.47-49.  

Kashid, S., and S. Kumar. 2012. “Applications of Artificial Neural Network to Sheet Metal Work – A Review.” American Journal of Intelligent System 2: 168–176. doi:https://doi.org/10.5923/j.ajis.20120207.03

Kwitek, M. 2016. “A Feasibility Study of Azure Machine Learning for Sheet Metal Fabrication.” Master’s thesis, Industrial Management Unit, University of Vaasa, Finland. 

Lin, Z. C., and D. Y. Chang. 1995. “Application of a Neural Network Machine Learning Model in the Selection System of Sheet Metal Bending Tooling.” Artificial Intelligence and Engineering 10: 21–37. doi:https://doi.org/10.1016/0954-1810(95)00013-5.  

Stoerkle, D. D., P. Seim, L. Thyssen, and B. Kuhlenkoetter. 2016. “Machine Learning in Incremental Sheet Forming.” Proceedings of 47th International Symposium on Robotics, Munich, Germany, Germany, 667–673

Tsai, S.-Y., and J.-Y. Chang. 2018. “Parametric Study and Design of Deep Learning on Leveling System for Smart Manufacturing.” 2018 IEEE International Conference on Smart Manufacturing, Hsinchu, Taiwan

Wu, X., J. Wang, A. Flitman, and P. Thomson. 1999. “Neural and Machine Learning to the Surface Defect Investigation in Sheet Metal Forming.” Proceedings of 6th International Conference on Neural Information Process Perth, Australia, November 16-20, 1088–1093. [Google Scholar]

Wuest, T., D. Weimer, C. Irgens, and K.-D. Thoben. 2016. “Machine Learning in Manufacturing: Advantages, Challenges, and Applications.” Production & Manufacturing Research: An Open Access Journal 23–45. doi:https://doi.org/10.1080/21693277.2016.1192517. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]

Zwierzycki, M., P. Nicholas, and M. Ramsgaard Thomsen. 2018. “Localised and Learnt Applications of Machine Learning for Robotic Incremental Sheet Forming.” In: De Rycke K. et al.(eds) Humanizing Digital Reality. Springer, Singapore. 382, https://doi.org/10.1007/978-981-10-6611-5_32 [Crossref], [Google Scholar]