Research On Muscle Force Estimation Method Based On Microscopic Neural Drive Information

Human limb movement is produced by the movement of skeleton around joint under the traction of muscle contraction.Measuring or estimating the skeletal muscle contraction force is the basis for exploring motor behavior and health.When skeletal muscle contracts,both the force and electromyogram(EMG)are produced simultaneously.Using EMG to estimate muscle force is an extensive and practical technique.Through the EMG decomposition technology,the firing sequence and the action potential waveform of a single motor unit(i.e.microscopic nerve drive information)can be obtained,which constitute the neural command from the central nervous system to a single motor neuron.Over the years,various macroscopic features extracted from the raw surface EMG(sEMG)data have been widely reported for estimation of the muscle force,but the use of microscopic neural drive information has not been well investigated.Those macroscopic features fail to describe the internal mechanism of muscle force generation and the phenomenon of muscle fatigue in particular,leading to compromised precision in estimating the muscle force.In order to achieve high precision muscle force estimation and to improve its robustness to the muscle fatigue,investigations into muscle force estimation based on the microscopic nerve drive information are presented.The primary achievements and contributions are summarized as follows:(1)A novel method for muscle force estimation is proposed based on the microscopic nerve drive information.In this method,the preprocessed high-density sEMG signals were firstly decomposed using a progressive FastICA peel-off algorithm to obtain the action potential waveform and firing sequence of each motor unit.Then the relationship between the action potential waveform and the twitch force amplitude of each motion unit was established by a machine learning algorithm.The twitch force contributions of individual motor units were calculated through a physiological model of muscle force formation and they were accumulated to obtain the whole muscle force.In order to verify the effectiveness of this method,we designed an experiment to collect the high-density sEMG and true force data.In addition,by comparing with four common methods based on the sEMG envelope,the sEMG root mean square,the motor unit firing rate and recruitment order of motor units respectively,the superiority of this method in the performance of muscle force estimation was confirmed.(2)The applicability of the proposed method under the condition of muscle fatigue was investigate.Muscle fatigue can be presented with nerve regulation alterations that may affect the characteristics of raw sEMG signals,thus causing bias in muscle force estimation using sEMG features.The proposed method directly uses the microscopic nerve drive information after high-density sEMG decomposition.With sufficient and direct reflection of the changes in nerve regulation,the proposed method is hypothesized not to be affected by the muscle fatigue condition.In order to verify this hypothesis,a fatiguing experiment was carried out.The results of muscle force estimation by using the proposed method and the envelope-based method demonstrated that the performance of the muscle force estimation based on EMG envelope method was significantly reduced under the fatigue condition,but the proposed method was not affected by fatigue factor.This research verified the hypothesis and demonstrated the applicability of this method under the fatigue condition.The muscle force estimation method based on micro neural driving information follows the physiological process of muscle force generation and considers the different contributions of individual motor units to muscle force formation,achieves high precision muscle force estimation and adaptability in fatigue condition.Meanwhile,it exhibits great application potential in precise motor control and rehabilitation medicine.