| As an important physiological function index of human body,balance ability plays a very important role in daily activities.The regulation of balance ability is a complicated process,which requires the participation of sensory input,nerve center integration and limb posture control.The decline of balance ability may lead to an increase in the risk of falls and the inability to take care of daily life,which seriously affects the quality of life.With the aging of society,the decline of body function and the increase of related diseases have led to more and more balance disorders.Therefore,the assessment of balance ability is particularly important for rehabilitation training or treatment of patients with balance disorders.At present,most of the assessment methods of balance ability are relatively single,and scientific quantification is insufficient,which restricts further analysis.How to obtain more comprehensive experimental data and means and get more accurate and objective evaluation results is of great significance to improve the balance ability evaluation level.In this paper,based on the balance detection platform which combines dynamometer board,Electromyogram(EMG)acquisition instrument and Electroencephalogram(EEG)acquisition instrument,the multi-modal information related to balance is studied extensively.By blocking proprioception and vision,the changes of plantar pressure signal and corticomuscular coherence were analyzed,and the relationship between biological information and plantar pressure information of human balance regulation was explored,and the human balance ability was evaluated.The main work and innovations of this paper are as follows:(1)The method of collecting the kinematics and bioelectric signals of the sole of human foot,lower limb muscles and cerebral cortex at the same time to carry out the research of human balance is put forward These signals contain a lot of important information related to human balance control.Compared with the traditional balance research based on plantar pressure or EMG signals,the EEG signals introduced in this paper can further explore and study the activity information of nerve center in the process of balance regulation,so as to make the assessment of balance ability more comprehensive and accurate.(2)An automatic removal method of ocular artifacts in single channel EEG signals based on complete empirical mode decomposition with adaptive noise(CEEMDAN)and independent component analysis(ICA)is studied.This method first adaptively decomposes the EEG signal with artifacts into multi-dimensional intrinsic mode functions(IMF)to satisfy the requirement of blind source separation for positive or over-determined signals,then uses ICA method to construct the multi-dimensional source signal for the intrinsic mode functions,uses fuzzy entropy threshold criterion to discriminate the artifacts in the multi-dimensional source signal,completes filtering and reconstructs the EEG signal,This method solves the undetermined problem and the uncertain problem of separated source signals of traditional blind source separation method.In the denoising of EMG signal,the wavelet soft threshold method is used to complete the pre-processing of EMG signal.(3)Combining the balance regulation of nerve center from cortex to muscle and the change of plantar pressure during the period,a new feature describing the balance coordination response ability of human body is proposed.Based on the analysis of the characteristics of the balance-related plantar pressure center(COP)under the condition of visual and proprioceptive blockade,the corticomuscular coherence in the process of human balance is analyzed.It is found that the corticomuscular coherence is mainly located in the β frequency band and y frequency band.With the decrease of the stability of human balance,the coherence frequency band tends to shift to the high frequency band,it is concluded that coherence band transfer is a physiological inherent mechanism of self-balancing regulation,and the ratio of COP comprehensive index to coherence frequency CR is regarded as a new feature of human balance ability.Based on the characteristics of CR,COP,multivariate and multiscale sample entropy of EEG signal and EMG signal,the feature vectors for evaluating human balance ability are constructed..(4)Extreme Learning Machine(ELM)based on Manifold Regularization was used to evaluate human balance ability.The classifier introduces manifold learning idea to mine the non-linear structure in the feature space of the extreme learning machine,which improves the classification accuracy.On the basis of training classifiers with feature vectors,classification and recognition are completed,which verifies the validity of balancing ability feature based on plantar pressure and biological information. |
