Study On Algorithms For Feature Information Extraction Based On Muscle Tension Perception

Muscle tension is the degrees of strain that characterizes the muscles of the human body.It is a form of abnormal increase in muscle tension.It is a common complication of stroke patients,and its external manifestations are muscle stiffness and continuity or intermittent convulsions accompanied by severe pain,clinical manifestations of muscle tonic,clonic and painful symptoms.At present,the critical characteristic information is mainly extracted by sensor information perception and algorithm processing to extract the type of enthalpy contained in it and its characteristic threshold.Based on the research of sensory acquisition and feature analysis of muscle tension signals,this thesis proposes to use the B4-FFT algorithm(B4-FFT:Base 4 Fast Fourier Transform)to deal with clonic and painful muscle tension signals,using time domain and frequency domain.The combined method was used to analyze the temporal and frequency domain characteristics of the spastic muscle tension.The musculature walking lower limb muscle tension collection experiment and human-computer interaction force excitation experiment were carried out.The B4-FFT algorithm was used to carry out the exoskeleton rehabilitation.Experimental study on the extraction of lower limb muscle tension and interactive force excitation of exoskeleton gait rehabilitation robot.The specific research contents and related work are as follows.(1)According to the sensing mechanism of lower limb muscle spasm,the muscle tension sensor for acquiring muscle tension signal was designed to obtain the spasm detection flow method.By analyzing the commonly used filtering methods,the characteristics of muscle tension signal were proposed.The mobile mean filtering method is used for preprocessing,which lays a foundation for the research of spasm feature information extraction and the experimental research work of rehabilitation robots.(2)Time domain feature information extraction and expression of spastic muscle tension.By completing the experimental collection of stroke patients,the muscle tension signals reflected by the three pathological underarms of tonic,clinic and painful spasm were analyzed,and the time domain occurrence threshold and time domain characteristic expression of each type of spasm were given.It shows that the time domain statistical method has great limitations,which is not conducive to the feature extraction and judgment of clonic and painful spasm.The experimental study on the time domain feature extraction of lower limb muscle tension by gait walking points out the time domain under gait walking.Feature extraction has the influence of operational errors and other factors,and further requirements are proposed for the study of frequency domain feature extraction of spastic muscle tension in the following chapters.(3)Extraction and expression of frequency domain characteristic information of spastic muscle tension.Firstly,the Fourier mathematical theory is used to analyze the characteristics of spasm muscle tension,and the corresponding expressions of three types of spasm are given.Then,the DFT(DFT:Discrete Fourier Transform)direct algorithm and B2-FFT(B2-FFT:Base 2 Fast Fourier Transform)algorithm are compared and analyzed.And the difference in performance between the three methods of B4-FFT algorithm.According to the advantages,the B4-FFT algorithm is used to deal with the spastic muscle tension signal,and the frequency interval,amplitude threshold and frequency domain characteristics of spasm occur.The B4-FFT algorithm is used to process the steps.The lower limb muscle tension signal of the walking state shows that the gait walking has little effect on the disturbance of the frequency domain feature extraction.(4)The experimental study on human-computer interaction force excitation was carried out to verify the effectiveness of the feature extraction algorithm.The muscle tension feature extraction experiment stimulated by human-computer interaction force,the frequency domain threshold is effectively extracted when the spasm occurs,and the effectiveness of the extraction algorithm is validated;the interaction force excitation experiment in the assisting process of the exoskeleton gait rehabilitation robot.The effectiveness of the extraction method on the dynamic extraction of muscle tension information was verified and used to evaluate the assist effect of exoskeleton rehabilitation robots.In this thesis,based on the research results of the spasm feature information extraction algorithm,the algorithm can extract the spasm characteristic thresholds of the spastic muscle tension signals in the time domain and the frequency domain in a stable and real-time manner,and provide a method for accurately extracting the threshold of spasm occurrence and the identification of spasm type.An effective technical means;In order to prevent stroke patients from providing critical information acquisition channels for lower limb muscle spasm during rehabilitation training,the safety and intelligence level of rehabilitation robots can be effectively improved.