輻射熱為熱舒適性指標中非常重要的參數，對於一般的建築使用者而言，溫度以及濕度為主要重點觀測的數據。在大力推動淨零碳排的現在，空調的設定溫度越來越保守，也導致在某些輻射熱沒有控制好的環境，使用者會被迫在熱舒適不佳的環境下進行工作或活動，本研究希望推廣大眾對於熱輻射的了解，並且提升對環境輻射熱的掌握，簡易黑球溫度計可以透過物聯網與空調設備連接，更精密的掌握體感溫度達到節能效果，更能透過3D列印技術將既有的溫度計透過外殼包覆，進行改造成為黑球溫度計。
研究方向主要從製作黑球的過程中，依照各個階段所需進行統整，並且聚焦於黑球的外殼將之作為研究核心。在初步的試作過程中，發現外殼對於太陽直射的耐受度有限，會造成黑球外殼的變形以及破損，因此進而產生第一階段的曝曬實驗；黑球的尺寸也會影響著溫度計的讀數，分別可以分為直徑的影響以及厚度的影響，這個部份透過3D列印黑球最佳化實驗進行驗證；最後在數據校正的部分先透過應用組實驗檢驗數據的吻合度，再透過機械學習中的長短期記憶模型進行校正。
實驗結果在壁面外殼破裂的情形下PLA材質有1.5毫米厚度的限制，平均統計有2%的變形量，ABS材質在材料上較為堅硬，1毫米厚度外殼可以順利產出，在變形量的統計上則是0.5%，最後則是光固樹脂，在列印支撐適當的情況下可以順利製作出1毫米厚度的黑球。3D列印黑球最佳化實驗最終的數據統計後，因黑球表面製作的精細度，會容易產生誤差，實驗結果顯示ABS具有最好的性能，且球體越薄延遲越短，5公分直徑可以帶來相較3.1公分直徑優秀的表現。最後戶外環境的數據透過機械學習修正完成後可以從最大溫度差為5.92°C，MAPE為2.86%最佳化到4.4°C而MAPE則是1.88%，但還有許多進步空間，也希望後續研究能完善這個部分。

Radiant heat is a critical factor in thermal comfort assessment, yet for general building users, temperature and humidity are often the primary observed indicators. As net-zero carbon emissions become a prominent goal, air conditioning settings are increasingly conservative, which can lead to uncomfortable indoor environments when radiant heat is not adequately managed. This study aims to raise public awareness of thermal radiation and enhance environmental monitoring by promoting the use of simplified black globe thermometers. Through integration with IoT systems and HVAC equipment, these devices can improve perception of operative temperature and contribute to energy savings. Furthermore, by applying 3D printing techniques, existing thermometers can be modified with external shells to function as black globe thermometers.
The research focuses on the development process of the black globe, particularly the optimization of its external shell. Initial trials revealed that under direct sunlight, certain shell materials suffered from deformation and damage, leading to an exposure test as the first phase. The size of the globe—specifically its diameter and shell thickness—was found to significantly influence temperature readings. This relationship was examined through a series of 3D printing optimization experiments. Calibration was conducted first via experimental validation and subsequently refined using a Long Short-Term Memory (LSTM) machine learning model.
Experimental results indicated that for PLA material, a minimum shell thickness of 1.5 mm is necessary to prevent damage, with an average deformation rate of 2% across samples. ABS, being more rigid, allowed for successful production of 1 mm thick shells with a lower deformation rate of 0.5%. Resin-based globes, when printed with proper supports, exhibited no visible deformation under sunlight at 1 mm thickness and were identified as the most suitable material. Although the surface precision of 3D-printed globes limited alignment with theoretical values, thinner shells proved advantageous in indoor environments due to faster response times, with minimal impact on overall performance. Outdoor data, once corrected using machine learning, still presents opportunities for improvement, which future research is expected to address.

[1]Andrés, M. S., Chércoles, R., Navarro, E., de la Roja, J. M., Gorostiza, J., Higueras, M., & Blanch, E. (2023). Use of 3D printing PLA and ABS materials for fine art. Analysis of composition and long-term behaviour of raw filament and printed parts. Journal of Cultural Heritage, 59, 181-189. doi:https://doi.org/10.1016/j.culher.2022.12.005
[2]Bangyu Li, Y. Q. (2024). Weather Prediction Using CNN-LSTM for Time Series Analysis: A Case Study on Delhi Temperature Data. doi:/10.48550/arXiv.2409.09414
[3]Bedford, T., & Warner, C. G. (1934). The Globe Thermometer in Studies of Heating and Ventilation. Epidemiology and Infection, 34(4), 458-473. doi:10.1017/S0022172400043242
[4]Boltzmann, L. (1884). Derivation of Stefan's law, concerning the dependency of heat radiation on temperature, from the electromagnetic theory of light. Annalen der Physik und Chemie (in German), 291–294. doi:10.1002/andp.18842580616
[5]Fanger, P. O. (1973). Assessment of man's thermal comfort in practice. British Journal of Industrial Medicine, 30(4), 313-324. doi:10.1136/oem.30.4.313
[6]Fischer, T., & Krauss, C. (2018). Deep learning with long short-term memory networks for financial market predictions. European Journal of Operational Research, 270(2), 654-669. doi:10.1016/j.ejor.2017.11.054
[7]Hasan, M. N., Toma, R. N., Abdullah-Al, N., Islam, M. M. M., & Kim, J. M. (2019). Electricity Theft Detection in Smart Grid Systems: A CNN-LSTM Based Approach. Energies, 12(17), 18. doi:10.3390/en12173310
[8]Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735-1780. Retrieved from <Go to ISI>://WOS:A1997YA04500007
[9]Hong, F., Long, D. T., Chen, J. Y., & Gao, M. M. (2020). Modeling for the bed temperature 2D-interval prediction of CFB boilers based on long-short term memory network. Energy, 194, 12. doi:10.1016/j.energy.2019.116733
[10]HUMPHREYS, M. A. (1977). THE OPTIMUM DIAMETER FOR A GLOBE THERMOMETER FOR USE INDOORS*. The Annals of Occupational Hygiene, 20(2), 135-140. doi:10.1093/annhyg/20.2.135
[11]Jiang, N., Zheng, X. P., Sun, L. E., Zheng, H., & Zheng, Q. H. (2021). Long Short-Term Memory based PM2.5 Concentration Prediction Method. Engineering Letters, 29(2), 765-774. Retrieved from <Go to ISI>://WOS:000652486600046
[12]Kang, Y., Ma, C., Wang, S., Wu, W., & Zhao, K. (2022). A two-segment LSTM based data center temperature prediction model. IEICE Electronics Express, 19(21), 20220291-20220291. doi:10.1587/elex.19.20220291
[13]Moursi, A. S., El-Fishawy, N., Djahel, S., & Shouman, M. A. (2021). An IoT enabled system for enhanced air quality monitoring and prediction on the edge. Complex & Intelligent Systems, 7(6), 2923-2947. doi:10.1007/s40747-021-00476-w
[14]Planck, M. (1901). Ueber das Gesetz der Energieverteilung im Normalspectrum. Annalen der Physik und Chemie (in German), 309, 553-563. doi:10.1002/andp.19013090310
[15]Shuo, Q., & Yinglai, Q. (2022, 2022/12/29). How Machine Learning Methods Unravel the Mystery of Bitcoin Price Predictions. Paper presented at the Proceedings of the 2022 International Conference on mathematical statistics and economic analysis (MSEA 2022).
[16]Stefan, J. (1879). On the relationship between heat radiation and temperature. 391-428.
[17]Wang, S., & Li, Y. (2015). Suitability of acrylic and copper globe thermometers for diurnal outdoor settings. Building and Environment, 89, 279-294. doi:https://doi.org/10.1016/j.buildenv.2015.03.002
[18]Wang, W., Liu, K., Zhang, M. X., Shen, Y. C., Jing, R., & Xu, X. D. (2021). From simulation to data-driven approach: A framework of integrating urban morphology to low-energy urban design. Renewable Energy, 179, 2016-2035. doi:10.1016/j.renene.2021.08.024
[19]Zhang, Y., Wang, S., Sun, G., & Mao, J. (2022). Aerodynamic surrogate model based on deep long short-term memory network: An application on high-lift device control. Proceedings of the Institution of Mechanical Engineers, Part G, 236(6), 1081-1097. doi:10.1177/09544100211027023