A Machine Learning-Enabled Real-Time temperature response system based on Polymer-Filler interactions for conductive network assembly

Abstract

Temperature sensing is essential for the human body’s interaction with the environment, and electronic skin mimicking human perception is crucial for developing smart wearable devices. Wearable sensors based on conductive polymer composites (CPCs) possess large sensitive, simple, and low-cost preparation characteristics. However, establishing the conductive networks necessitates sufficient filler doping, posing processability and cost control challenges. Herein, we report a susceptible thermo-sensor (TS) that utilizes the secondary polymer thermoplastic polyurethane (TPU) to connect carbon black (CB) particles, facilitating the assembly of a conductive network at low concentrations, thereby improving their electrical conductivity. The TS can defect temperatures in the range of 15 – 45 °C with a sensitivity of 1200 %, a positive temperature coefficient (PTC) intensity of approximately 5, and a response time of less than 10 s. By machine learning to identify the output signal of TS, the recognition accuracy reaches 99.8 %, then the real-time temperature display can be successfully realized. This approach provides a simple preparation method for personalized medicine and soft robotics.