| In the field of competitive sports,increasing the exercise intensity to improve the competition performance is the main training method,but if there is no reasonable scientific control of the exercise intensity,it is easy to cause fatigue-induced muscle damage to athletes.After a large amount of exercise,the human body will experience a phenomenon of decreased muscle function,which is called muscle fatigue.Once signs of muscle fatigue appear,athletes should adjust training intensity in time to avoid muscle damage.Therefore,the use of scientific and technological means to monitor the muscle fatigue state of athletes has important practical application value.Surface electromyogram（s EMG）,as a bioelectrical signal reflecting the temporal and spatial aggregation of skeletal muscle action potentials,is commonly used in the fields of gesture recognition,rehabilitation training,and prosthetic control.Aiming at the problem of low accuracy of muscle fatigue detection methods,this paper proposes a method for optimizing the parameters of the back-propagation neural network（BPNN）.Finally,a muscle fatigue detection system for athletes training is designed.The paper first designs a three-channel EMG acquisition module for collecting triceps brachii,deltoid muscle,and brachioradialis muscle,and transmits the s EMG signals to the host computer through Bluetooth and stores it in the TF card;secondly,the s EMG signals are preset.Processing,including the use of extended Kalman filtering method filtering and sliding window method data segmentation;then 16 features of the s EMG signals were extracted,and the Adaboost algorithm was used to determine the most significant 13 features as the training and testing samples of the model.The muscle fatigue detection model based on fruit fly（FOA）-genetic（GA）algorithm is optimized by BP neural network（FG-NN）.Finally,in order to evaluate the detection effect of the model,the FG-NN model and the comparison test of traditional BP neural network,FOA-NN and GA-NN model,the test results show that the accuracy rate of FG-NN model is 96.7%,and the accuracy rates of the other three models are 85.8%,89.3% and 90.2% respectively.In this paper,a muscle fatigue detection model based on FG-NN is constructed on the basis of BP neural network.The designed muscle fatigue detection system has practical application value for human fatigue detection in sports training. |
