Design And Construction Of Recognition System For Leg Fatigue Based On MMG

In recent years,with the strengthening of residents’health awareness,the call for national fitness has been increasing.However,in order to achieve good effect of physical exercise,it is still necessary to arrange fitness scientifically and reasonably according to the athlete’s muscle fatigue state.Therefore,the precise control of muscle fatigue is very important for physical exercise.In view of the difference in exercise intensity between different people,this topic proposes a generic fatigue degree determination rule,and designs a portable leg muscle fatigue recognition system,including hardware and software.The hardware designs and produces a wireless MMG collector and a portable digital signal processing platform based on an embedded microprocessor,which are used to collect and process signals respectively;The software uses a muscle fatigue recognition algorithm to predict the muscle fatigue state of leg,and the results are displayed in real time through graphical user interface.In order to explore the relationship between MMG and muscle fatigue,this topic utilizes MMG of the leg as analysis objects,using time domain analysis and frequency domain analysis to perform statistical analysis respectively.Among them,the time domain analysis method uses the root mean square as a feature parameter to regression analysis,and the results show that R~2 between the root mean square and fatigue is 0.478,which is less than 1,indicating that the time domain characteristics cannot be reflected in the muscle fatigue.The frequency domain analysis method uses the mean power frequency and the median frequency as a feature parameter for regression analysis.The results indicate that R~2 between the mean power frequency and the fatigue is 0.8657,the R~2 between the median frequency and the fatigue is 0.6984,which is much higher than the result of time domain analysis,indicating that the frequency domain analysis method can better characterize muscle fatigue.In addition,the analysis results also indicate that as the degree of muscle fatigue is deepens,the mean power frequency and the median frequency have a significant decline,and the degree of median frequency is more intense.Using the results of the above analysis,combined frequency domain analysis and machine learning methods,this topic proposes a method for identifying leg muscle fatigue,including signal acquisition,preprocessing,STFT transformation,and model construction.Then three fatigue prediction models were constructed through different machine learning algorithms and tested with cross-validation tools.In the end,the verification results show that the model constructed by the random forest algorithm is the best.The model score is 0.958,which is close to 1,indicating that its prediction accuracy is high.The model is transplanted to the portable digital signal processing platform,thus improving the construction of the entire system.Finally,this topic also designed an experiment to test the classification performance of the system.The experimental results show that the average prediction accuracy of the system is 71.5%,which meets the requirements of system design,but the prediction accuracy of different fatigue levels is different.In addition,this experiment also uses model constructors and other personnel to test system classification accuracy,and results show that the fatigue prediction accuracy of model constructors is higher than others,indicating that the versatility of the system needs to be improved.According to the research content of this subject,it can be made into a portable leg muscle fatigue detector,which can be used for real-time monitoring of leg muscle fatigue and has broad application prospects.