Department of Electrical and Energy, Nevsehir Haci Bektas Veli University, Nevsehir, Merkez, Turkiye.
Graduate School of Natural and Applied Sciences, Erciyes University, Kayseri, Turkiye.
Proc Inst Mech Eng H. 2024 Jul;238(7):827-836. doi: 10.1177/09544119241266375. Epub 2024 Aug 6.
A real-time hypothermia and hyperthermia monitoring system with a simple body sensor based on a Convolutional Neural Network (CNN) is presented. The sensor is produced with 3D-printed thermochromic material. Due to the color change feature of thermochromic materials with temperature, 3D-printed thermochromic Polylactic Acid (PLA) material was used to monitor temperature changes visually. In this paper, we have used the transfer learning technique and fine-tuned the AlexNet CNN. Thirty images for each temperature class between 28-44°C and 510 image data were used in the algorithm. We used 80% and 20% of the data for training and validation. We achieved 96.1% accuracy of validation with a fine-tuned AlexNet CNN. The material's characteristics suggest that it could be employed in delicate temperature sensing and monitoring applications, particularly for hypothermia and hyperthermia.
提出了一种基于卷积神经网络（CNN）的具有简单体传感器的实时低温和高温监测系统。该传感器采用 3D 打印的热敏材料制成。由于热敏材料随温度变化的颜色变化特征，使用 3D 打印的热敏聚乳酸（PLA）材料进行直观的温度变化监测。在本文中，我们使用了迁移学习技术并对 AlexNet CNN 进行了微调。在算法中，使用了每个温度类别（28-44°C）之间的 30 张图像和 510 张图像数据。我们使用 80%和 20%的数据进行训练和验证。使用经过微调的 AlexNet CNN，我们在验证中达到了 96.1%的准确率。该材料的特性表明，它可以用于精密的温度感应和监测应用，特别是用于低温和高温。
