Research On ECG And Environment Based Exercise Load Assessment Technology

As the number of athletes in China continues to increase,the number of sports injuries caused by over-exercise is increasing year by year.The main reason for this is that athletes are unable to correctly assess their own exercise load.More and more athletes are using exercise load assessment equipment for their own exercise guidance.However,traditional exercise load assessment ignores the influence of environmental factors and exercise type(aerobic/anaerobic)on its performance,and the output of exercise load grading strength is coarse.In order to solve the above problems,this paper introduces environmental factors(temperature,relative humidity)and exercise type into the exercise load assessment model,and implements an exercise load assessment system based on ECG signals using the industry standard RPE as the grading basis.The main research elements are as follows:1.An R-peak detection algorithm based on empirical mode decomposition,grey relational analysis,and Shannon entropy envelop.Traditional R-peak recognition algorithm has low accuracy in recognizing R-peaks of motion ECG signals.To solve the problem,this paper uses gray correlation to remove the signal components affecting Rpeak recognition and accurately identifies the R-peak positions in ECG signals by Shannon entropy envelope algorithm.This paper conducts comparison experiments with traditional algorithms on authoritative ECG data sets based on this algorithm,and the results show that this algorithm has comparable R-peak recognition performance for resting ECG signals and better R-peak recognition performance for exercise ECG signals compared with traditional algorithms.2.An exercise load assessment model that considers environmental information.In order to introduce environmental factors into the exercise load assessment model,this paper uses multiple regression method to construct a relationship model between RPE and heart rate difference considering environmental factors,and sets up comparison experiments to verify the importance of environmental information for exercise load assessment.It is the first time to explore the effects of different temperatures and relative humidity on the relationship between heart rate difference and RPE.Based on this conclusion,this paper constructs an exercise load assessment model based on random forest algorithm with environmental factors and ECG features as input features,and has higher exercise load recognition accuracy when comparing the related algorithm on the environmental data set constructed using exercise experiments,which proves that the model in this paper has good exercise load recognition ability.3.A sports load assessment system with sports type recognition capability.Considering the importance of exercise type to exercise load assessment,and also in order to realize the function of automatic exercise type recognition,this paper constructs an anaerobic threshold classification model based on LSTM neural network with ECG signal as the basis,and tests the model performance on the exercise type dataset constructed using exercise experiments,and achieves 89.80% accuracy rate of exercise type recognition.Further,this paper integrates the anaerobic threshold recognition model with the exercise load assessment model considering environmental information,and tests the model on the environmental dataset to achieve 87.46% exercise load recognition accuracy,and the recognition accuracy of each exercise load is greater than80%,which further improves the recognition performance and stability of the exercise load assessment model.In summary,compared to existing exercise load assessment tools,this paper uses environment and exercise type to make the exercise load assessment system more capable of identifying exercise load,which is a reference value for future research.