Research On Real-time Monitoring System Of Sports Fatigue

With the development of our country's economy and society,continuous development in fields such as aerospace,deep sea exploration and sports competition will be crucial for monitoring the fatigue status of workers and athletes and preventing excessive fatigue and sports injuries.In order to protect health and prevent accidents,this article focuses on wearable sports fatigue monitoring technology and hopes to achieve non-invasive detection technology to detect human body sports fatigue.By performing real-time monitoring of human electrocardiographic signals and bioelectrical signals on related muscles,the body surface bioelectric signals acquired are subjected to some necessary filtering and preprocessing to extract fatigue-related feature indices,and then based on machine learning and based on deep neural The network method classifies and recognizes the state of sports fatigue.The main research work of this article is as follows:(1)This article uses electrocardiographic signals and surface EMG signals to detect human sports fatigue.Designing high-amplification,low-noise physiological signal sensors for physiological signals with weak energy(uV level)and unacceptable characteristics.These include conditioning circuits for front-end faint analog signals,classical filter circuits,multi-stage amplifier circuits,and back-end digital circuits.system.According to the amplitude-frequency characteristics of weak physiological signals for schematic design,device selection,hardware circuit production and debugging,physiological signal acquisition experiments.(2)Using an adaptive filter algorithm to suppress motion artifact noise.Aiming at the feature of strong randomness of motion artifact noise,overlapping of spectrum range and ECG signal,large amplitude and strong interference to ECG signal,which cannot be processed by classical filters,an adaptive filter algorithm is combined with human motion information to design motion artifact from Adapt to the filter.The adaptive filter algorithm continuously optimizes filter parameters and weights based on human motion information to automatically match motion noise and suppress motion interference.(3)Sports fatigue detection algorithm based on pattern recognition.After in-depth investigation and analysis,it has been found that most researchers used thesingle physiological and biochemical indicators to study human sports fatigue.However,the human body is a miscellaneous organism.Its physiological changes are complex and it is impossible to accurately determine the physical state through a single variable.Therefore,the use of pattern recognition algorithm to fuse multiple physiological indicators for sports fatigue detection has strong applicability and higher accuracy.(4)A sports fatigue detection algorithm based on LSTM neural network is proposed.For the pattern recognition SVM algorithm model,its classification optimal decision surface is fixed after the training is completed,it can not effectively use the current input and output optimization model,can not keep the historical input information in the model,the application flexibility and scope are limited.In order to better solve the above problems,the LSTM neural network model based on the movement fatigue detection technology,LSTM neural network can use the current input information and output results,continuous optimization of model parameters and weights through feedback,the use of hidden layer structure to the historical information It is reserved for use in the network and can make full use of all historical information.It is easy to train,has strong learning ability and wide applicability,and it has better detection and classification results.Finally,the experimental results of sports fatigue exhaustion show that with the increase of exercise intensity and exercise time,the subjective fatigue feeling of athletes gradually increases.ECG indicators: heart rate,high frequency power(HF),low frequency power(LF),The surface EMG indicators of thigh muscle: RMS and MPF vary significantly and can be used as indicators of exercise fatigue state recognition.Experiments on multi-class classifiers of sports fatigue show that the classification accuracy of multi-class classifiers based on SVM is lower than that of LSTM multi-class classifiers.Multi-class classifiers of LSTM neural network models can make full use of historical information,and mining the characteristics of physiological signal,then improve the accuracy of multiple fatigue classifiers.