發展傳統機台儀表之光學IOT技術

Development of an Optical IOT Technique for Instruments and Gauges on Traditional Machines

羅俊翔(Jun-xiang Luo)

指導教授 ： 陳冠宇

中原大學/工學院/機械工程研究所/碩士(2019年)

https://doi.org/10.6840/cycu201900646

摘要

近年來隨著工業4.0的浪潮，許多企業積極將其原有生產線提升成自動化或智慧化，其中將機台或設備上的控制器或儀表之數據利用光學影像技術判讀轉成數位化資訊，成為必要的工作，此即稱為光學物聯網(optical Internet of Things, optical IoT)。本文乃結合機器視覺、機器學習演算法及圖形使用者介面三項技術，開發舊式生產機台儀錶資訊之自動監控系統。此監控系統可降低工廠的人力成本，提升工作流暢度。此外，此監控系統尚包含圖形使用者介面，提供管理人員在遠端即可獲知機台一切狀況。本文利用深度學習之卷積神經網路架構訓練LED數字0-9的辨識機制，以判讀設備控制箱上的LED壓力顯示器之數據，最後的實驗結果顯示，本文所發展的光學IOT監控系統可達90%以上的正確率。

關鍵字

深度學習 ； 機器學習 ； 機器視覺 ； 卷積神經網路 ； 光學IoT

並列摘要

In recent years, with the wave of Industry 4.0, many companies have actively upgraded their original production lines into automation or intelligence. Among them, using optical imaging technology for interpreting the data of the controller or instrument on the machine or equipment into digital data is one of the necessary tasks. It is also called optical Internet of Things (optical IoT). This thesis combines three technologies of machine vision, machine learning algorithm and graphical user interface to develop an automatic monitoring system for instrument information of old production machines. This monitoring system can reduce labor costs and improve workflow. In addition, the monitoring system also includes a graphical user interface that provides administrators with complete visibility of the machine at the remote end. This thesis builds a recognition mechanism for LED digital 0-9 by using a convolutional neural network based deep learning architecture to interpret the data of the LED pressure display on the device control box. The experimental results show that the optical IOT monitoring system developed in this thesis can achieve more than 90% accuracy.

並列關鍵字

deep learning ； machine learning ； machine vision ； convolutional neural network ； optical IoT

參考文獻

[1] N. Dalal and B. Triggs, “Histograms of Oriented Gradients for Human Detection,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 886-893, Jun. 20-26, 2005, San Diego, USA.

[2] D. Lowe, “Object recognition from local scale-invariant features,” in Proceedings of the 7th IEEE International Conference on Computer Vision, Vol. 2, pp. 1150-1157, Sep. 20-25, 1999, Kerkyra, Greece.

[3] H. Bay, T. Tuytelaars, and L. Van Gool, “Surf: Speeded up robust features,” in Proceedings of the 9th European Conference on Computer Vision, pp. 404-417, May 7-13, 2006, Graz, Austria.

[4] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” in Proceedings of the IEEE, Vol. 86, pp. 2278-2324, 1998.

[5] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, Vol. 521, pp. 436-444, 2015.