Static Analysis And Evaluation Of Guitar Chord Fingers Based On SEMG Signal

Surface electromyography signal Electromyography(sEMG)is a kind of electrical signal,and it is also a very effective and useful method in the non-invasive muscle inspection of human body.We analyze and study the detection and analysis method of sEMG signal,and this paper mainly uses this relatively perfect and mature method,and improves some parts of it to evaluate the performance of arm recognition under different strength The electrical signal when playing the guitar,so as to evaluate the strength of a person's fingers according to the chord when playing the guitar.In this paper,the strength of this single aspect of in-depth study.In the first step,through the acquisition of EMG signal and preprocessing process,we can further understand the relationship between the electric signal of the movement of the forearm and the signal of the force generating muscle group,also clarify the force generating muscle group of the arm in this paper.The two channels of EMG signals are compared and analyzed,and the appropriate signals are selected and preprocessed,such as filtering,active segment endpoint detection and so on.Then feature extraction and analysis.At present,there are three commonly used methods,namely time-domain method,frequency-domain time-frequency-domain method.Finally,we select the singular value feature,the maximum value feature and the energy coefficient feature of wavelet and wavelet packet in the time-frequency domain,and then make a statistical analysis of some samples,and finally select the singular value of wavelet coefficient as the feature vector for the next recognition and classification training.Finally,pattern recognition and exploration.In this paper,two commonly used algorithms are selected,BP artificial neural network algorithm and DBN algorithm for comparative analysis.The former,for the recognition rate of four different forces,the recognition rate of 1.5n force is up to 94.8%,the latter is up to 72.5%,while the actual intensity of 1.5n according to chord is in line with the standard.Because of the high recognition rate of the former,the algorithm of BP neural network is finally selected.