在物聯網技術強調萬物皆可上網的時代，手持工具目前在市面上卻仍未有相關智慧產品出現，然而手持工具使用的效果非常地依賴使用經驗，特別是有動力的工具如電鑽或是電鋸等，若能藉由物聯網技術提供給使用者相關使用資訊，相信對於使用者是一大福音，特別是目前台灣的藍領階級正在減少當中，日後需要用到這一類工具的工作如土木工程、裝修、以及水電等都可能會面臨到人員不足的狀態，這時如何提高工作效率和縮短從業人員的經驗差距就成了重要的課題。
本研究提出了一套基於手持無線動力鑽頭和隨機森林的操作條件辨認系統，操作條件為鑽削刀具與工件材料之組合。在實驗室環境下收集不同操作條件下電鑽之馬達電壓、馬達電流與二軸加速度，分別使用金屬用途、木材用途以及水泥用途之鑽頭鑽削鋁塊和木材，共6種工件鑽頭組合，並使用不同資料預處理方法建立操作條件分類模型，以建立效率最佳的資料預處理流程。進一步透過當前操作條件辨認結果確認使用者是否使用合適的工具進行工作。本研究使用訓練資料分別以取樣頻率2048 Hz以及256 Hz進行分類模型建立。取樣頻率2048 Hz資料針對鑽削成功與否辨認準確率為98%，針對各工件鑽頭組合辨認準確率為96%；而取樣頻率256 Hz資料針對鑽削成功與否辨認準確率為93%，針對各工件鑽頭組合辨認準確率為92%。

The technology of automation has been widely adopted in the industry. However, the handheld cordless power tool still requires the human dexterity to operate it, and the performance depends on the skill of the operators. The advance of IoT and Industry 4.0 technology enables the data collection from the handheld tool and the development of intelligent functions. Nevertheless, there are few studies on this topic, and the progress is relatively slow compared to other automatic machine tools. In this study, the drilling conditions of using different drill bits to drill different materials were studied. The experiment of six drilling combinations were performed with three drill bits designed for drilling metal, wood, and concrete, respectively, to drill two working materials, aluminum alloy and wood blocks. Classification models of these drilling conditions of handheld cordless power drill were investigated using the collected data and Random Forest algorithm. The motor voltage, current and two-axis acceleration signals generated by different operating conditions were collected. The filter and embedded feature selection methods were applied to reduce the number of feature inputs from 31 to 7 without scarification of the classification accuracy. The accuracy of classifying the six drilling combinations was 96% with the sampling rate of 2048 Hz, and it was 92% at 256 Hz. In addition, the accuracy of 86% was achieved without the acceleration signals. The results show the promising of developing intelligent functions of the handheld power tools with the concept of Industry 4.0.

[1] 國家發展委員會, 2020, “人口成長趨勢,” from https://www.ndc.gov.tw/Content_List.aspx?n=0F11EF2482E76C53, last access：June 11, 2021
[2] 吳雅樂, 2020, “台積電發「做6給7」日薪搶人 南台灣驚爆缺工潮！,” from https://www.wealth.com.tw/home/articles/27919, last access：June 11, 2021
[3] Fery, Chr. W., Jacubasch, A., Kuntze, H.-B., and Plietsch, 2003, “R. Smart Neuro-Fuzzy Based Control of a Rotary Hammer Drill,” Proceedings of the 2003 IEEE International Conference on Robotics and Automation, Taipei, Taiwan, September 14-19.
[4] Heinis, T. B., Loy, C. L. and Meboldt, M., 2018, “Improving Usage Metrics for Pay-per-Use Pricing with IoT Technology and Machine Learning,” Research-Technology Management, 61(5), pp. 32-40.
[5] Dorr, M., Ries, M., Gwosch, T. and Matthiesen, S., 2019, “Recognizing Product Application based on Integrated Consumer Grade Sensors：A Case Study with Handheld Power Tools,” Procedia CIRP, 84, pp. 798-803.
[6] Dorr, M., Peters, J. and Matthiesen, S., 2021, “Data-Driven Analysis of Human-Machine Systems – A Data Logger and Possible Use Cases for Field Studies with Cordless Power Tools,” Advances in Intelligent Systems and Computing, 1253, pp. 56-62.
[7] Joshua, L., and Varghese, K., 2020, “Accelerometer-Based Activity Recognition in Construction,” Journal of Computing in Civil Engineering, 25(5), pp. 370–379.
[8] Koskimaki, H., Huikari, V., Siirtola, P., Laurinen, P., and Roning, J., 2009, “Activity recognition using a wrist-worn inertial measurement unit: A case study for industrial assembly lines,” 2009 17th Mediterranean Conference on Control and Automation, pp. 401–405.
[9] Ghosh, D., Olewnik, A., and Lewis, K., 2018, “Application of Feature-Learning Methods Toward Product Usage Context Identification and Comfort Prediction,” ASME. J. Comput. Inf. Sci. Eng, 8(1).
[10] Maekawa, T., Kishino, Y., Yanagisawa, Y., and Sakurai, Y., 2012, “Recognizing Handheld Electrical Device Usage with Hand-Worn Coil of Wire,” Pervasive 2012. Lecture Notes in Computer Science, 7319, pp. 234–252
[11] Uhl, M., Bruchmuller, T. and Matthiesen, S., 2019, “Experimental Analysis of User Forces by Test Bench and Manual Hammer Drill Experiments with Regard To Vibrations and Productivity,” International Journal of Industrial Ergonomics, 72, pp. 398-407.
[12] Das, J., Bales, G. L., Kong, Z. and Linke, B., 2018, “Integrating Operator Information for Manual Grinding and Characterization of Process Performance Based on Operator Profile,” Journal of Manufacturing Science and Engineering, Transations of the ASME, 140(8), Paper No. 081011.
[13] Kamath, A., Linke, B. and Chu, C., 2020, “Enabling Advanced Process Control for Manual Grinding Operations,” Smart and Sustainable Manufacturing Systems, 4(2), pp. 210-230.
[14] Voet, H., Altenhof, M., Ellerich, M., Schmitt, R. H. and Linke, B., 2019, “A Framework for the Capture and Analysis of Product Usage Data for Continuous Product Improvement,” Journal of Manufacturing Science and Engineering, Transactions of the ASME, 141(2), Paper No. 021010.
[15] Matthiesen, S. and Germann, R., 2018, “Meaningful Prediction Parameters for Evaluating the Suitability of Power Tools for Usage,” Procedia CIRP, 70, pp. 241-246.
[16] Germann, R., Jahnke, B. and Matthiesen, S., 2019, “Objective Usability Evaluation of Drywall Screwdriver under Consideration of the User Experience,” Applied Ergonomics, 75, pp. 170-177.
[17] Rokach, L., 2010, “Ensemble-based classifiers,” Artificial Intelligence Review, 33(1-2), pp. 1–39.
[18] Chen, J., 2019, “【機器學習】偏差與方差之權衡 Bias-Variance Tradeoff,” from https://jason-chen-1992.weebly.com/home/-bias-variance-tradeoff, last access：June 1, 2021
[19] RPubs, 2018, “R筆記 – (16) Ensemble Learning(集成學習),” from https://rpubs.com/skydome20/R-Note16-Ensemble_Learning, last access：June 1, 2021
[20] Breiman, L., Friedman, J. H., Olshen, R. A., and Stone, C. J., 1984, “Classification and Regression Trees,” Brooks/Cole Publishing, Monterey.
[21] Breiman, L., 2001, “Random Forests,” Machine Learning volume, 45, pp. 5–32.
[22] Friedman, J. H., 2001, “Greedy Function Approximation：A Gradient Boosting Machine,” The Annals of Statistics, 29(5), pp. 1189-1232.
[23] Chen, T. and Guestrin, C., 2016, “XGBoost：A Scalable Tree Boosting System,” KDD '16：Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp.785-794
[24] Precision Microdrives, 2016, “DC Motors : Voltage Vs. Output Speed Vs. Torque - AB-032 Released,” from https://www.precisionmicrodrives.com/content/dc-motors-voltage-vs-output-speed-vs-torque-ab-032-released/, last access：June 8, 2021
[25] Myers, J. L. and Well, A. D., 2003, “Research design and statistical analysis,” Lawrence Erlbaum Associates Inc. publisher, pp. 563-564
[26] Battiti, R., 1994, “Using mutual information for selecting features in supervised neural net learning,” IEEE Transactions on Neural Networks, 5(4), pp. 537-550.
[27] Menze, B. H, Kelm, B M., Masuch, R., Himmelreich, U., Bachert, P., Petrich, W. and Hamprecht, F. A, 2009, “A comparison of random forest and its Gini importance with standard chemometric methods for the feature selection and classification of spectral data,” BMC Bioinformatics, 10, pp. 213.
[28] Altmann, A., Toloşi, L., Sander, O. and Lengauer T., 2010, “Permutation importance: a corrected feature importance measure,” Bioinformatics, 26(10), pp. 1340-1347.
[29] Toulas, B., 2017, “6061-T6 Aluminium – The Ultimate Guide,” from https://www.engineeringclicks.com/6061-t6-aluminum/, last access：June 16, 2021
[30] Mai, T. H., Militz, H. and Mai, C., 2010, “Modification of beech veneers with N-methylol-melamine compounds for the production of plywood,” European Journal of Wood and Wood Products, 70(4), pp. 1-12.
[31] DHL, 2017, “Supply Chain Makes Smart Glasses New Standard in Logistics,” from http://www.dhl.com/en/press/releases/releases_2017/all/logistics/dhl_supply_chain_makes_smart_glasses_new_standard_in_logistics.html