Research On Muscle Strength Prediction Method Of Different Posture States Of Upper Limbs Based On SEMG

During the rehabilitation process,the active or passive rehabilitation mode and the added auxiliary force or load should be determined according to the muscle force,so as to complete the specific formulation of the rehabilitation plan.In elbow flexion muscle force of related study,usually by hand at the end of force as a muscle force,but the actual elbow flexion movement of the muscle force is provided by the upper limb muscles together,is not provided by the biceps separately,and elbow flexion Angle changes can also cause the change of muscle share rate,therefore,this method is not accurate.Moreover,the existing evaluation methods of rehabilitation effect have been lacking a quantitative and accurate evaluation index on muscle strength evaluation.Aiming at the above problems,this paper aims to study a model that can predict biceps muscle force online according to sEMG and upper limb posture status,and build an evaluation system of muscle force rehabilitation effect according to the muscle force prediction model.First,the muscle force analysis unit of human body is constructed.The dumbbell load model was imported into the software and combined with the human model to form an interaction model.The set drive enabled the interaction model to complete the elbow flexion movement of the human upper limb.Reverse dynamic analysis was performed on the interactive model,and the muscle force data of biceps under different positions were obtained.Then,the muscle force prediction model based on improved random forest was built.To sEMG integral electrical values,root mean square value,mean power frequency and median frequency signal characteristic value for constructing the independent variable,joint Angle characteristic matrix,derived by muscle force analysis unit of muscle force data as the dependent variable,with the improved random forest model the nonlinear mapping relationship between the two,to complete the training of muscle force prediction model.Through the comparison with BP neural network model,the validity and accuracy of the muscle force prediction model based on improved random forest are verified.Finally,the evaluation system of muscle strength rehabilitation effect was established on PC.The trained muscle force prediction model was integrated into the muscle force rehabilitation effect evaluation system,and the biceps muscle force was predicted online and its rehabilitation effect was evaluated when the human body was doing elbow flexion.Through experiments,the accuracy of online prediction of muscle strength and the usability of rehabilitation effect assessment system are verified.The results show that,with the assistance of software AnyBody,muscle force prediction can be performed accurately using sEMG and joint Angle signals.The muscle force rehabilitation effect assessment system completed in this study not only predicted the muscle force data online,but also assessed the muscle force rehabilitation effect of the biceps muscle of the human elbow flexion movement.