Wearable piezoelectrical muscle monitoring system for swimming analysis neural network

Abstract

Swimming is one of the most popular worldwide sports, which offers the players healthy body, improved fitness, and endless enjoyment due to its easy accessibility and year-round availability. Monitoring swimmers can aid in the development and participation of swimming sport. However, due to the difficulty of in-water monitoring, the existing methods in tool box are still limited. In this paper, a more accessible wearable monitoring system based on a flexible piezoelectric device is developed. This system collects muscle contracting and relaxing information by measuring its generated force on a Polyvinylidene fluoride (PVDF) sensor. Then the information is transmitted via a low-frequency radio frequency (RF) signal to the receiver on land. This information is analyzed using a convolutional neural network and the classification of different strokes is realized. The system also supports monitoring at two positions by wearing multiple devices on the body. By integrating data from multiple positions, the analysis algorithm achieves a higher prediction accuracy of over 95%. This work demonstrates that the combination of wearable monitoring devices, the Internet of Things, and Artificial Intelligence, which holds significant promise for sports education, research, promotion, and development.