Research on muscle fatigue classification based on principal component analysis-random forest algorithm

In order to solve the problem of low accuracy of muscle fatigue classification using surface electromyography (sEMG), a fatigue classification model based on principal component analysis of surface electromyography and random forest algorithm was proposed. Firstly, the surface EMG signal is pre-processed by de-noising, and then the time domain, frequency domain and time-frequency domain features are extracted. Principal component analysis in feature extraction was used for feature dimensionality reduction, three effective principal components were extracted, and features with high contribution to the model were screened combined with the variable importance ranking results of random forest algorithm. The screened joint feature parameters were input into the classifier, and 8 subjects were tested to verify the rationality of the proposed method. The experimental results showed that: By optimizing the parameters, the classification accuracy of the random forest algorithm based on principal component analysis is increased by 5% to 93.3% compared with the simple random forest algorithm, which proves that the proposed method has excellent effect in the classification of muscle fatigue.