Research on Risk Assessment Technology for Military Physical Training Based on Deep Learning
Hao Li, Chi Zhang · 2024
To solve the problem of timely and accurate risk assessment in military physical training, a deep learning-based risk assessment method is developed. This method specifically takes into account the biological signals generated by participants during military physical training, which include both electrical and non-electrical signals. The electrical signals are surface electromyography (sEMG) signals from the muscles being trained, while the non-electrical signals encompass the participants' current heart rate, blood oxygen levels, body temperature, respiratory rate, and blood pressure. The processed sEMG signals are analyzed using Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks to assess muscle fatigue in participants. Subsequently, the processed physiological parameters and muscle fatigue metrics are fed into a Fuzzy Neural Network (FNN) for analysis, yielding the final training risk assessment.