CN103886215B

Movatterモバイル変換

Info

Publication number: CN103886215B
Application number: CN201410136557.8A
Authority: CN
Inventors: 刘玲; 李飞; 陈香
Original assignee: University of Science and Technology of China USTC
Current assignee: University of Science and Technology of China USTC
Priority date: 2014-04-04
Filing date: 2014-04-04
Publication date: 2017-01-11
Anticipated expiration: 2034-04-04
Also published as: CN103886215A

Abstract

本发明提供了一种基于肌肉协同的步行能力分析方法和装置。所述装置包括多通道表面肌电采集模块，数据解析存储模块，步态数据分析处理模块；所述计算方法使用非负矩阵分解算法从步态周期表面肌电信号中提取步态肌肉协同，将成人步态肌肉协同模式作为标准模板，利用基于步态肌肉协同模式相似性的协同综合计算(Synergy Comprehensive Calculation，SCC)模型，对受试者的肌肉协同模式进行评分获得反映其步态的特征参数。该方法从大脑中枢神经系统控制的角度揭示不同个体的步行能力差异，通过该方法可以获得一个客观的步行能力参数。该参数反映了中枢神经系统对步态控制的效果。

The invention provides a walking ability analysis method and device based on muscle coordination. The device includes a multi-channel surface electromyography acquisition module, a data analysis storage module, and a gait data analysis and processing module; the calculation method uses a non-negative matrix decomposition algorithm to extract gait muscle coordination from the gait cycle surface electromyography signal, and the The adult gait muscle synergy pattern is used as a standard template, and the muscle synergy pattern of the subjects is scored using the Synergy Comprehensive Calculation (SCC) model based on the similarity of the gait muscle synergy pattern to obtain the characteristic parameters reflecting their gait . This method reveals the difference in walking ability of different individuals from the perspective of the control of the central nervous system of the brain, and an objective parameter of walking ability can be obtained by this method. This parameter reflects the effect of the central nervous system on gait control.

Description

Translated fromChinese

一种基于肌肉协同的步行能力分析方法和装置A method and device for analyzing walking ability based on muscle coordination

技术领域technical field

本发明是基于传感器技术和生物信号处理技术的步态分析技术，特指一种基于肌肉协同的步行能力分析方法和装置，属于传感器技术和生物信号处理技术领域。The invention is a gait analysis technology based on sensor technology and biological signal processing technology, in particular to a walking ability analysis method and device based on muscle coordination, and belongs to the technical field of sensor technology and biological signal processing.

背景技术Background technique

步态分析已成为分析人体运动系统不可缺少的手段之一。目前三维步态分析技术存在两大问题：首先步态分析系统采用专用视频设备和计算机设备，价格昂贵，操作专业性强，难以普及。Gait analysis has become one of the indispensable means to analyze human motion system. At present, there are two major problems in the 3D gait analysis technology: firstly, the gait analysis system uses special video equipment and computer equipment, which is expensive and highly professional in operation, making it difficult to popularize.

表面肌电信号(Surface Electromyography,SEMG)是通过表面电极从体表获得的肌电信号，基于表面肌电信号的步态数据获取不需要专用的实验场地，可以克服步态分析视频采集设备容易受到环境干扰的缺点；并且表面肌电信号不仅能提供步态的基本信息，还可以从表面肌电信号中提取肌肉协同，为步态计算提供客观有效的参数。因此，提供一种基于肌肉协同的步行能力分析方法和装置具有重要的意义。Surface Electromyography (Surface Electromyography, SEMG) is the electromyography signal obtained from the body surface through surface electrodes. The gait data acquisition based on the surface electromyography signal does not require a dedicated experimental site, which can overcome the vulnerability of gait analysis video acquisition equipment. The disadvantages of environmental interference; and the surface EMG signal can not only provide the basic information of gait, but also extract muscle synergy from the surface EMG signal, providing objective and effective parameters for gait calculation. Therefore, it is of great significance to provide a walking ability analysis method and device based on muscle coordination.

发明内容Contents of the invention

针对目前步态分析技术存在的不足，本发明旨在提出一种基于肌肉协同的步行能力分析方法和装置。Aiming at the shortcomings of the current gait analysis technology, the present invention aims to propose a walking ability analysis method and device based on muscle coordination.

一种基于肌肉协同的步行能力分析方法包括：A walking ability analysis method based on muscle coordination includes:

(1)利用非负矩阵分解(Non-negative Matrix Factorization,NMF)算法从步态周期表面肌电信号(Surface Electromyography,SEMG)中提步双腿的步态肌肉协同模式。具体指用NMF对受试者双腿的步态周期包络矩阵(步态周期包络矩阵是预处理和步态周期分割之后的步态周期表面肌电信号)进行非负矩阵分解以提取左右腿肌肉协同矩阵L,R。将L,R分别表示为：(1) Use the Non-negative Matrix Factorization (NMF) algorithm to extract the gait muscle coordination pattern of the legs from the Surface Electromyography (SEMG) of the gait cycle. Specifically, NMF is used to perform non-negative matrix decomposition on the gait cycle envelope matrix of the subject's legs (the gait cycle envelope matrix is the gait cycle surface EMG signal after preprocessing and gait cycle segmentation) to extract the left and right Leg muscle synergy matrix L, R. Express L and R as:

L＝L^m×l＝{l₁,l₂…l_l}L＝L^m×l ＝{l₁ ,l₂ …l_l }

R＝R^m×r＝{r₁,r₂…r_r}R＝R^m×r ＝{r₁ ,r₂ …r_r }

其中列向量l_i,r_j(i＝1,2…l；j＝1,2…r)分别表示左右腿的肌肉协同模式，l,r分别表示左右腿的步态肌肉协同数量，m表示步态分析中所选取的一组肌肉的数量；用成年人左右腿共有的肌肉协同模式建立标准模板集T：Among them, the column vectors l_i , r_j (i=1,2...l; j=1,2...r) represent the muscle coordination patterns of the left and right legs respectively, l and r respectively represent the gait muscle coordination numbers of the left and right legs, and m represents The number of a group of muscles selected in gait analysis; the standard template set T is established using the muscle coordination patterns common to the left and right legs of adults:

T＝T^m×s＝{t₁,t₂…t_s}T＝T^m×s ＝{t₁ ,t₂ …t_s }

其中列向量t_k(k＝1,2…s)表示标准步态肌肉协同模式，即成年人双腿共有的肌肉协同模式；s表示标准步态肌肉协同模式的数量，即成年人左右腿共有的肌肉协同模式数量；Among them, the column vector t_k (k=1,2…s) represents the standard gait muscle coordination pattern, that is, the muscle coordination pattern shared by both legs of an adult; s represents the number of standard gait muscle coordination patterns, that is, the adult’s left and right legs share the number of muscle synergy patterns;

(2)利用基于步态肌肉协同模式相似性的协同综合计算(Synergy ComprehensiveCalculation，SCC)模型实现受试者步行能力指标的计算；SCC模型综合考虑受试者步态肌肉协同模式的数量，结构以及左右腿在步行中表现出的对称性三个因素；并且基于肌肉协同的步行能力分析方法，从神经系统控制的角度揭示了大脑对步态运动的协同控制策略，该模型的计算结果可以从本质上反映大脑对步态运动的协调控制能力。(2) Using the Synergy Comprehensive Calculation (SCC) model based on the similarity of gait muscle synergy patterns to realize the calculation of the subject's walking ability index; the SCC model comprehensively considers the number, structure and structure of the subject's gait muscle synergy patterns The three factors of symmetry between the left and right legs in walking; and the analysis method of walking ability based on muscle coordination reveals the brain's coordinated control strategy for gait movement from the perspective of nervous system control. The calculation results of this model can be essentially It reflects the brain's ability to coordinate and control gait movements.

所述SCC模型计算步行能力指标的方法，包括以下计算步骤：The method for calculating the walking ability index of the SCC model comprises the following calculation steps:

(1)用皮尔森相关系数计算公式R(x,y)计算受试者左右腿各个步态肌肉协同模式l_i,r_j(i＝1,2…l；j＝1,2…r)与各个标准步态肌肉协同模式t_k(k＝1,2…s)之间相关系数R(t_k,l_i),R(t_k,r_j)。定义左右腿肌肉协同矩阵L,R与标准模板集T的相似程度Sim(T,L),Sim(T,R)，其计算公式如下：(1) Use the Pearson correlation coefficient calculation formula R(x, y) to calculate the synergy mode l_i , r_j of each gait muscle of the subject's left and right legs (i=1,2...l; j=1,2...r) Correlation coefficients R(t_k , l_i ), R(t_k , r_j ) with each standard gait muscle synergy mode t_k (k=1,2...s). Define the similarity Sim(T,L),Sim(T,R) between the left and right leg muscle synergy matrix L,R and the standard template set T, the calculation formula is as follows:

$S S i i m m ((T T,, L L)) = = {Σ Σ}_{i i = = 11}^{l l} F f ((T T,, {l l}_{i i}))$

$S S i i m m ((T T,, R R)) = = {Σ Σ}_{j j = = 11}^{r r} F f ((T T,, {r r}_{j j}))$

F(T,l_i)＝max(R(t_k,l_i)) k＝1,2…sF(T,l_i )=max(R(t_k ,l_i )) k=1,2...s

F(T,r_j)＝max(R(t_k,r_j)) k＝1,2…sF(T,r_j )=max(R(t_k ,r_j )) k=1,2…s

F(T,l_i)，F(T,r_j)分别表示左右腿肌肉协同模式l_i,r_j与标准模板集T的相似程度。对每个F(T,l_i)，F(T,r_j)乘上得分系数B＝100/s，则左右腿相似性得分S_l,S_r分别为：F(T,l_i ), F(T,r_j ) represent the similarity between the left and right leg muscle synergy patterns l_i , r_j and the standard template set T, respectively. For each F(T,l_i ), multiply F(T,r_j ) by the scoring coefficient B=100/s, then the similarity scores S_l and S_r of the left and right legs are respectively:

S_l＝B*Sim(T,L)S_l ＝B*Sim(T,L)

S_r＝B*Sim(T,R)S_r =B*Sim(T,R)

S_l,S_r的取值在0-100之间。The values of S_l and S_r are between 0-100.

(2)由于受试者左右腿步态肌肉协同模式之间的相关系数可以反映左右腿在步态运动中表现出的对称性，因此在SCC模型中定义了对称性得分Sym。在该模型中用皮尔森相关系数计算公式R(x,y)计算受试者左右腿各个步态肌肉协同模式l_i,r_j(i＝1,2…l；j＝1,2…r)之间的相关系数R(r_j,l_i)并定义左右腿肌肉协同矩阵L,R之间的相似程度Sim(R,L)，其计算公式为：(2) Since the correlation coefficient between the gait muscle synergy patterns of the subjects' left and right legs can reflect the symmetry of the left and right legs in gait movement, the symmetry score Sym is defined in the SCC model. In this model, the Pearson correlation coefficient calculation formula R(x, y) is used to calculate the synergy mode l_i , r_j of each gait muscle of the subject's left and right legs (i=1,2...l; j=1,2...r ) between the correlation coefficient R(r_j ,l_i ) and define the similarity Sim(R,L) between the left and right leg muscle synergy matrix L, R, the calculation formula is:

$S S i i m m ((R R,, L L)) = = \{\begin{matrix} {Σ Σ}_{i i = = 11}^{l l} F f ((R R,, {l l}_{i i})) & r r &GreaterEqual; &Greater Equal; l l \\ {Σ Σ}_{j j = = 11}^{r r} F f ((L L,, {r r}_{j j})) & r r < < l l \end{matrix}$

F(R,l_i)＝max(R(r_j,l_i)) j＝1,2…rF(R,l_i )=max(R(r_j ,l_i )) j=1,2…r

F(L,r_j)＝max(R(r_j,l_i)) i＝1,2…lF(L,r_j )=max(R(r_j ,l_i )) i=1,2…l

F(R,l_i)表示该SCC模型中左腿肌肉协同模式l_i与右腿肌肉协同矩阵R的相似程度；F(L,r_j)则表示右腿肌肉协同模式r_j与左腿肌肉协同矩阵L的相似程度。对称性得分Sym的计算公式如下：F(R,l_i ) represents the similarity between the left leg muscle synergy pattern l_i and the right leg muscle synergy matrix R in the SCC model; F(L,r_j ) represents the right leg muscle synergy pattern r_j and the left leg muscle The degree of similarity of the synergy matrix L. Symmetry score Sym is calculated as follows:

Sym＝A*Sim(R,L)Sym=A*Sim(R,L)

其中对称性得分系数A＝100/min(l,r)，Sym的取值在0-100之间。The symmetry score coefficient A=100/min(l,r), and the value of Sym is between 0-100.

最后，求S_l,S_r,Sym的平均值，得到总体得分S：Finally, find the average of S_l , S_r , Sym to get the overall score S:

S＝(S_l+S_r+Sym)/3S＝(S_l +S_r +Sym)/3

S即为SCC模型的计算结果，取值在0-100之间。S is the calculation result of the SCC model, and the value is between 0-100.

一种实现上述基于肌肉协同的步行能力分析方法的装置，采用下面的技术方案：该装置由多通道表面肌电采集模块、数据解析存储模块、步态数据分析处理模块组成。其中多通道表面肌电采集模块用于采集步态运动中双腿的SEMG，并通过有线或无线的方式将信号发送给数据解析存储模块；数据解析存储模块实时地解析步态数据，并进行存储；步态数据分析处理模块从数据解析存储模块中加载存储的步态数据进行分析处理，计算步行能力指标。A device for realizing the above-mentioned walking ability analysis method based on muscle coordination adopts the following technical scheme: the device is composed of a multi-channel surface electromyography acquisition module, a data analysis and storage module, and a gait data analysis and processing module. Among them, the multi-channel surface electromyography acquisition module is used to collect the SEMG of the legs in gait movement, and send the signal to the data analysis storage module through wired or wireless means; the data analysis storage module analyzes the gait data in real time and stores them The gait data analysis and processing module loads and stores the gait data from the data analysis storage module for analysis and processing, and calculates the walking ability index.

所述步态分析处理模块包括步态数据预处理单元①、步态周期分割单元②、步态肌肉协同提取单元③、步行能力指标计算单元④、显示控制单元⑤；①—④作为步行能力指标计算的核心单元；显示控制单元⑤用于设置系统的工作参数，工作状态，以及处理结果的显示等。The gait analysis processing module includes a gait data preprocessing unit ①, a gait cycle segmentation unit ②, a gait muscle coordination extraction unit ③, a walking ability index calculation unit ④, and a display control unit ⑤; ①-④ are used as walking ability indicators The core unit of calculation; the display control unit ⑤ is used to set the working parameters of the system, working status, and display of processing results.

所述步行能力指标计算①—④是四个串行计算单元，其中步态数据预处理单元①用于对每通道的SEMG进行滤波、去均值和整流处理，以消除步态数据采集中基线噪声等的影响，最后输出SEMG的包络信号。步态周期分割单元②采用基于加速度信号的步态周期分割算法对预处理后SEMG的包络进行步态周期的分割；分割后对所有步态周期每通道SEMG的幅值进行归一化处理，以消除由电极放置位置等因素引起的不同受试者之间的信号幅值差异，最后将所有已经分割出的步态周期SEMG信号的每个通道降采样为统一长度，以得到步态周期包络矩阵。步态肌肉协同提取单元③对分割出的步态周期包络矩阵进行非负矩阵分解，计算左右腿步态肌肉协同模式及其肌肉协同模式的数量。步行能力指标计算单元④则采用SCC模型实现步行能力指标的计算。The calculation of the walking ability index ①-④ is four serial calculation units, wherein the gait data preprocessing unit ① is used to filter, remove the mean value and rectify the SEMG of each channel, so as to eliminate the baseline noise in the gait data acquisition etc., and finally output the envelope signal of SEMG. The gait cycle segmentation unit ② adopts the gait cycle segmentation algorithm based on the acceleration signal to segment the gait cycle of the preprocessed SEMG envelope; after the segmentation, the amplitude of each channel SEMG of all gait cycles is normalized, In order to eliminate the difference in signal amplitude between different subjects caused by factors such as electrode placement positions, and finally down-sample each channel of all the segmented gait cycle SEMG signals to a uniform length to obtain the gait cycle package network matrix. The gait muscle synergy extraction unit ③ performs non-negative matrix decomposition on the segmented gait cycle envelope matrix, and calculates the left and right leg gait muscle synergy patterns and the number of muscle synergy patterns. The walking ability index calculation unit ④ uses the SCC model to realize the calculation of the walking ability index.

所述数据解析存储模块和步态数据分析处理模块可以在普通电脑或者带WIFI的智能手机，平板电脑等移动终端上实现，因此该步行能力分析装置不受使用场合的限制，可在医院，康复机构，家庭等环境中使用。The data analysis and storage module and the gait data analysis and processing module can be implemented on mobile terminals such as ordinary computers or smart phones with WIFI, tablet computers, etc. Therefore, the walking ability analysis device is not limited by the use occasion and can be used in hospitals, rehabilitation Institutions, households and other environments.

附图说明Description of drawings

图1是本发明装置的结构示意图；Fig. 1 is the structural representation of device of the present invention;

图2是本发明实施例中表面肌电电极安放方法示意图；Fig. 2 is a schematic diagram of a surface electromyography electrode placement method in an embodiment of the present invention;

图3是本发明中计算步行能力指标的总体框图；Fig. 3 is the overall block diagram of calculating walking ability index among the present invention;

图4是本发明实施例中步态周期信号分割的示意图；Fig. 4 is a schematic diagram of gait cycle signal segmentation in an embodiment of the present invention;

图5是本发明实施例中SCC模型计算流程图。Fig. 5 is a flow chart of SCC model calculation in the embodiment of the present invention.

具体实施方式detailed description

下面结合附图说明，给出本发明的一个实施例，具体的实施过程如下：Below in conjunction with accompanying drawing description, provide an embodiment of the present invention, concrete implementation process is as follows:

1.根据图3所示步骤1，在该实施案例中我们以选取左右腿胫骨前肌(TA)、比目鱼肌(SO)、腓肠肌外侧(LG)、股外侧肌(VL)、股直肌(RF)、半腱肌(SE)、股二头肌长头(BF)和阔筋膜张肌(TF)共16块肌肉作为步态SEMG信号的获取部位为例。因此图1所示的多通道表面肌电采集模块(1)应至少包含16通道表面肌电电极(左右腿各8通道)。如图2所示，该实施例所用电极与皮肤接触面采用两个条形电极棒，放置时两电极连线与肌纤维走行方向一致，电极放置位置处事先剃去汗毛并用酒精棉擦拭。同时在该实施例中采用基于加速度信号ACC的步态周期分割，因此多通道表面肌电采集模块还集成了2通道三轴加速度传感器，实施过程中将其分别放置在受试者左右腿膝盖下方胫骨处。受试者按照自己最舒适的速度执行步行任务，实施过程中受试者需在水平地面前行N个步态周期，采集模块同步采集步行时的SEMG和ACC信号并通过有线的方式发送至图1所示的数据解析存储模块(2)。1. according to step 1 shown in Fig. 3, in this implementation case we select left and right leg tibialis anterior muscle (TA), soleus muscle (SO), gastrocnemius lateral (LG), vastus lateralis muscle (VL), rectus femoris ( RF), semitendinosus (SE), long head of biceps femoris (BF) and tensor fascia lata (TF), a total of 16 muscles are used as the acquisition site of gait SEMG signal as an example. Therefore, the multi-channel surface electromyography acquisition module (1) shown in Fig. 1 should include at least 16 channel surface electromyography electrodes (eight channels for the left and right legs). As shown in Figure 2, the electrode used in this embodiment uses two bar-shaped electrode rods on the contact surface with the skin. When placed, the connecting lines of the two electrodes are in the same direction as the muscle fibers. At the same time, in this embodiment, the gait cycle segmentation based on the acceleration signal ACC is adopted, so the multi-channel surface electromyography acquisition module also integrates a 2-channel three-axis acceleration sensor, which are respectively placed under the knees of the left and right legs of the subject during the implementation process at the tibia. The subject performs the walking task at his most comfortable speed. During the implementation, the subject needs to walk N gait cycles forward on the level ground. The acquisition module synchronously collects the SEMG and ACC signals during walking and sends them to the graph through a wired method. The data analysis storage module (2) shown in 1.

2.装置的数据解析存储模块(2)在普通计算机上实现，首先根据数据的具体的帧格式进行实时的解析，然后将解析过的数据存入文件。2. The data analysis and storage module (2) of the device is implemented on a common computer, firstly performing real-time analysis according to the specific frame format of the data, and then storing the analyzed data into a file.

3.步态数据分析处理模块(3)在普通计算机上实现。其显示控制单元首先启动文件载入功能，加载待处理的步态数据，当文件加载完成界面提示加载完成，显示控制单元启动数据分析，具体过程如下：3. The gait data analysis and processing module (3) is implemented on a common computer. Its display control unit first starts the file loading function and loads the gait data to be processed. When the file loading is completed, the interface prompts that the loading is complete, and the display control unit starts data analysis. The specific process is as follows:

(1)原始步态数据首先进入预处理单元，按照如图3所示步骤2进行步态信号预处理：在该实施案例中，先使用截止频率为40Hz、50阶的FIR高通滤波器对受试者的步态数据进行滤波，然后去均值、整流，最后使用截止频率为10Hz、50阶的FIR低通滤波器再次滤波以得到如图4中B所示的步态SEMG包络。(1) The original gait data first enters the preprocessing unit, and the gait signal preprocessing is carried out according to step 2 as shown in Figure 3: in this implementation case, the FIR high-pass filter with the cut-off frequency of 40Hz and 50th order is first used to filter the subject The gait data of the subjects were filtered, then averaged, rectified, and finally filtered again with a 50-order FIR low-pass filter with a cutoff frequency of 10 Hz to obtain the gait SEMG envelope shown in B in Figure 4.

(2)步态SEMG包络数据进入步态周期分割单元，利用ACC信号对其分割：如图4中C所示，在步行中加速度信号ACC是准周期的时变信号，因此在该实施例中利用基于窗函数的峰值检测方法检测左右腿加速度传感器Z轴信号的峰值点，将峰值点分别映射到左右腿的步态SEMG包络，分割出N个步态周期包络信号。图4中的B、C、D展示了利用步态ACC信号分割步态周期的详细过程。最后将N个步态周期包络信号的每通道统一降采样为长度n＝500，并利用步态周期每通道的最大值对该通道进行归一化处理，得到用于提取肌肉协同的步态周期包络矩阵，如图4中的E所示，矩阵的每一行对应一个肌肉通道。将预处理后单腿的N个步态周期包络矩阵记为EMG_o＝{O_i,i＝1,2…N}，O_i是一个m×n的矩阵，m表示步态分析中所选取的一组肌肉的数量，该实施例中m＝8。(2) The gait SEMG envelope data enters the gait cycle segmentation unit, and utilizes the ACC signal to segment it: as shown in Figure 4 C, the acceleration signal ACC is a time-varying signal of a quasi-period in walking, so in this embodiment In this paper, the peak detection method based on the window function is used to detect the peak points of the Z-axis signals of the left and right leg acceleration sensors, and the peak points are mapped to the gait SEMG envelopes of the left and right legs respectively, and N gait cycle envelope signals are segmented. B, C, and D in Fig. 4 show the detailed process of segmenting gait cycles using gait ACC signals. Finally, each channel of the N gait cycle envelope signal is uniformly down-sampled to a length of n=500, and the maximum value of each channel of the gait cycle is used to normalize the channel to obtain the gait for extracting muscle coordination The periodic envelope matrix, shown as E in Fig. 4, each row of the matrix corresponds to a muscle channel. The N gait cycle envelope matrix of a single leg after preprocessing is denoted as EMG_o ＝{O_i ,i=1,2…N}, where O_i is an m×n matrix, and m represents the The quantity of a group of muscles selected, m=8 in this embodiment.

(3)步态肌肉协同提取单元对步态周期包络矩阵EMG_o进行处理，由图3的步骤3知，步态肌肉协同提取单元会分别提取左右腿的肌肉协同模式，并计算协同模式的数量，其详细过程如下：(3) The gait muscle synergy extraction unit processes the gait cycle envelope matrix EMG_o . According to step 3 in Figure 3, the gait muscle synergy extraction unit will extract the muscle synergy patterns of the left and right legs respectively, and calculate the synergy pattern Quantity, the detailed process is as follows:

1)对每个步态周期矩阵O_i，按照非负矩阵分解NMF算法，计算出肌肉协同数量为j(j＝1,2…m)对应的肌肉协同矩阵W_i^m×j，1) For each gait cycle matrix O_i , according to the non-negative matrix factorization NMF algorithm, calculate the muscle synergy matrix W_i^m×j corresponding to the number of muscle synergy j (j=1,2...m),

2)求步态肌肉协同模式数量s：利用W_i^m×j，求得肌肉协同模式数量为j时的重建步态模式矩阵集其中重建步态模式矩阵然后计算所有EMG_j与分解前EMG_o的平方误差矩阵集V＝{V_j,j＝1,2…m}，其中V_j计算公式为：2) Calculate the number s of gait muscle coordination modes: use W_i^m×j , Obtain the reconstructed gait pattern matrix set when the number of muscle synergy patterns is j where the reconstructed gait pattern matrix Then calculate the square error matrix set V={V_j ,j=1,2...m} of all EMG_j and EMG_o before decomposition, where the calculation formula of V_j is:

$\begin{matrix} {V V}_{j j} = = \frac{11}{N N} {Σ Σ}_{i i = = 11}^{N N} ((11 - - \frac{{(({O o}_{i i} - - {R R}_{i i}^{j j}))}^{22}}{{O o}_{i i}^{22}})) & 00 < < {V V}_{j j} < < 11 \end{matrix}$

V_j表示肌肉协同数量为j时重建步态模式矩阵集EMG_j与原始步态周期矩阵集EMG_o的平均平方误差。根据先验知识，步态肌肉协同模式数量s确定为在t检验(p<0.05)下V_j的值大于0.95所对应的最小肌肉协同数量j，受试者的肌肉协同矩阵即为：V_j represents the average square error between the reconstructed gait pattern matrix set EMG_j and the original gait cycle matrix set EMG_o when the number of muscle coordination is j. According to prior knowledge, the number s of gait muscle synergy patterns is determined as the minimum muscle synergy number j corresponding to the value of V_j greater than 0.95 under the t test (p<0.05), and the muscle synergy matrix of the subject is:

${W W}^{m m \times \times s the s} = = {{{w w}_{11},, {w w}_{22} ... ... {w w}_{s the s}}} = = \frac{11}{N N} {Σ Σ}_{i i = = 11}^{N N} {W W}_{i i}^{m m \times \times s the s}$

3)将受试者左右腿肌肉协同矩阵分别表示为：3) The muscle synergy matrix of the left and right legs of the subject is expressed as:

L＝L^m×l＝{l₁,l₂…l_l}L＝L^m×l ＝{l₁ ,l₂ …l_l }

R＝R^m×r＝{r₁,r₂…r_r}R＝R^m×r ＝{r₁ ,r₂ …r_r }

其中列向量l_i,r_j(i＝1,2…l；j＝1,2…r)分别表示左右腿的肌肉协同模式，l,r分别表示左右腿的步态肌肉协同数量。按照该实施例所述方法1-3预先提取数名成年人左右腿共有的肌肉协同模式建立标准模板集T：The column vectors l_i , r_j (i=1,2...l; j=1,2...r) respectively represent the muscle coordination modes of the left and right legs, and l, r respectively represent the gait muscle coordination numbers of the left and right legs. According to the method 1-3 described in this embodiment, the muscle synergy patterns shared by several adults' left and right legs are pre-extracted to establish a standard template set T:

T＝T^m×s＝{t₁,t₂…t_s}T＝T^m×s ＝{t₁ ,t₂ …t_s }

其中列向量t_k(k＝1,2…s)表示标准步态肌肉协同模式，s表示标准步态肌肉协同模式的数量，并将标准模板存储在数据分析处理模块中。The column vector t_k (k=1,2...s) represents the standard gait muscle coordination pattern, s represents the number of the standard gait muscle coordination pattern, and the standard template is stored in the data analysis and processing module.

(4)步行能力指标计算单元根据SCC模型，按照如图5所示的流程图计算步行能力：首先将步态肌肉协同提取单元的计算结果L,R,l,r和标准模板集T及标准步态肌肉协同模式数量s输入该单元；然后分别计算左右腿各自的相似性得分S_l,S_r；根据l,r的比较结果，计算左右腿在步行中的对称性得分Sym；最后综合S_l,S_r,Sym得到步行能力指标。其具体的计算过程如下：(4) The walking ability index calculation unit calculates the walking ability according to the SCC model according to the flow chart shown in Figure 5: firstly, the calculation results L, R, l, r of the gait muscle synergy extraction unit and the standard template set T and the standard The number of gait muscle synergy patterns s is input into this unit; then the similarity scores S_l and S_r of the left and right legs are calculated respectively; according to the comparison results of l and r, the symmetry score Sym of the left and right legs in walking is calculated; finally, S_l , S_r , Sym get the walking ability index. The specific calculation process is as follows:

1)计算左右腿相似性得分S_l,S_r时先用皮尔森相关系数计算公式R(x,y)计算受试者左右腿各个步态肌肉协同模式l_i,r_j(i＝1,2…l；j＝1,2…r)与各个标准步态肌肉协同模式t_k(k＝1,2…s)之间相关系数R(t_k,l_i),R(t_k,r_j)，通过如下公式求得l_i,r_j与标准模板集T的相似程度F(T,l_i)，F(T,r_j)：1) When calculating the similarity scores S_l , S_r of the left and right legs, first use the Pearson correlation coefficient calculation formula R(x, y) to calculate the synergy modes of each gait muscle of the left and right legs l_i , r_j (i=1, 2…l; j=1,2…r) and each standard gait muscle synergy pattern t_k (k=1,2…s) correlation coefficient R(t_k ,l_i ),R(t_k ,r_j ), the degree of similarity F(T,l_i ), F(T,r_j ) between l_i ,r_j and the standard template set T is obtained by the following formula:

F(T,l_i)＝max(R(t_k,l_i)) k＝1,2…sF(T,l_i )=max(R(t_k ,l_i )) k=1,2...s

F(T,r_j)＝max(R(t_k,r_j)) k＝1,2…sF(T,r_j )=max(R(t_k ,r_j )) k=1,2…s

$R R ((x x,, y the y)) = = \frac{{Σ Σ}_{i i = = 11}^{s the s i i z z e e ((x x))} (({x x}_{i i} - - \overset{&OverBar; &OverBar;}{x x})) (({y the y}_{i i} - - \overset{&OverBar; &OverBar;}{y the y}))}{\sqrt{{Σ Σ}_{i i = = 11}^{s the s i i z z e e ((x x))} {(({x x}_{i i} - - \overset{&OverBar; &OverBar;}{x x}))}^{22}} \sqrt{{Σ Σ}_{i i = = 11}^{s the s i i z z e e ((x x))} {(({y the y}_{i i} - - \overset{&OverBar; &OverBar;}{y the y}))}^{22}}}$

然后对每个F(T,l_i)，F(T,r_j)乘上得分系数B＝100/s，则左右腿与标准模板的相似性得分S_l,S_r分别为：Then for each F(T,l_i ), F(T,r_j ) is multiplied by the scoring coefficient B=100/s, then the similarity scores S_l and S_r of the left and right legs and the standard template are respectively:

${S S}_{l l} = = B B {Σ Σ}_{i i = = 11}^{l l} F f ((T T,, {l l}_{i i}))$

${S S}_{r r} = = B B {Σ Σ}_{j j = = 11}^{r r} F f ((T T,, {r r}_{j j}))$

其中分别表示SCC模型中L,R与T的相似程度Sim(T,L),Sim(T,R)；S_l,S_r的取值在0-100之间。in Respectively represent the similarity Sim(T,L) and Sim(T,R) of L, R and T in the SCC model; the values of S_l and S_r are between 0 and 100.

2)计算左右腿对称性得分时，先用皮尔森相关系数计算公式R(x,y)计算受试者左右腿各个步态肌肉协同模式l_i,r_j(i＝1,2…l；j＝1,2…r)之间的相关系数R(r_j,l_i)，当r≥l，左右腿肌肉协同矩阵L,R之间的相似程度Sim(R,L)为：2) When calculating the symmetry scores of the left and right legs, first use the Pearson correlation coefficient calculation formula R(x, y) to calculate the synergy modes of each gait muscle of the subject's left and right legs l_i , r_j (i=1,2...l; The correlation coefficient R(r_j ,l_i ) between j=1,2...r), when r≥l, the similarity Sim(R,L) between the left and right leg muscle synergy matrix L, R is:

$S S i i m m ((R R,, L L)) = = {Σ Σ}_{i i = = 11}^{l l} F f ((R R,, {l l}_{i i}))$

F(R,l_i)＝max(R(r_j,l_i)) j＝1,2…rF(R,l_i )=max(R(r_j ,l_i )) j=1,2…r

当r<l时，左右腿肌肉协同矩阵L,R之间的相似程度Sim(R,L)为：When r<l, the similarity Sim(R,L) between the left and right leg muscle synergy matrices L and R is:

$S S i i m m ((R R,, L L)) = = {Σ Σ}_{j j = = 11}^{l l} F f ((L L,, {r r}_{j j}))$

F(L,r_j)＝max(R(r_j,l_i)) i＝1,2…jF(L,r_j )=max(R(r_j ,l_i )) i=1,2…j

Sim(R,L)与对称性得分系数A＝100/min(l,r)相乘，则对称性得分Sym为：Multiplying Sim(R,L) with the symmetry score coefficient A=100/min(l,r), then the symmetry score Sym is:

Sym＝A*Sim(R,L)Sym=A*Sim(R,L)

最后步行能力指标计算单元计算S_l,S_r,Sym的平均值，得到总体得分S：Finally, the walking ability index calculation unit calculates the average value of S_l , S_r , Sym to obtain the overall score S:

S＝(S_l+S_r+Sym)/3S＝(S_l +S_r +Sym)/3

通过该方法计算的步行能力指标S可以反映不同个体之间步行能力的差异。The walking ability index S calculated by this method can reflect the difference in walking ability among different individuals.