WO2017073770A1

WO2017073770A1 - Method of identifying myoelectric pattern employing time series information, and myoelectric prosthetic hand

Info

Publication number: WO2017073770A1
Application number: PCT/JP2016/082171
Authority: WO
Inventors: 昌宏粕谷; 龍加藤; 浩史横井
Original assignee: 国立大学法人電気通信大学; 株式会社メルティンＭｍｉ
Priority date: 2015-10-29
Filing date: 2016-10-28
Publication date: 2017-05-04
Also published as: JP2019004915A

Abstract

Provided is an algorithm which dramatically improves the identification rate of myoelectric patterns, while minimizing the number of myoelectric electrodes and the learning time, in order to control a myoelectric prosthetic hand having multiple degrees of freedom. This myoelectric prosthetic hand is provided with: a measuring unit for measuring a myoelectric pattern at constant time intervals; a myoelectric identification unit which identifies the myoelectric pattern from the measured myoelectricity; and a buffer which stores, in chronological order, a prescribed number of the myoelectric patterns identified by the myoelectric identification unit. With reference to the prescribed number of myoelectric patterns stored in the buffer, if an occupancy rate at least equal to a certain proportion is found, control conforming to a motion command corresponding to the myoelectric pattern found to have an occupancy rate at least equal to the certain proportion is carried out. According to the present invention, rapid learning and a high identification rate can be achieved while employing a low number of electrode channels, and the invention is thus effective for practical control of a myoelectric prosthetic hand having multiple degrees of freedom.

Description

Myoelectric pattern identification method and myoelectric prosthesis using time series information

The present invention relates to the identification of myoelectric patterns used in devices controlled by myoelectricity.

Devices such as myoelectric prosthetic hands and myoelectric prostheses have been developed that control motion based on the amount of surface myoelectric potential generated in the muscle.

[Background of prosthetic hand]
Prosthetic hands are roughly classified into decorative prostheses, active prostheses, and electric prostheses. The main purpose of the decorative prosthesis is to reproduce the appearance, and the active prosthesis is to reproduce the functionality. Therefore, the decorative prosthetic hand cannot be moved actively, and the active prosthetic hand is driven by transmitting the movement of the remaining part of the user as power using a wire or the like. On the other hand, as the name suggests, an electric prosthesis is driven by electric power and a motor. Electric prosthetic hands have fewer restrictions on wire handling than active prosthetic hands, and the direction and amount of movement can be freely designed to some extent, and various prosthetic hands such as those with multiple degrees of freedom have been produced. As an electric prosthetic hand, a myoelectric prosthetic hand controlled using a surface myoelectric signal for measuring myoelectricity on the skin surface is generally used.

[Required specifications of myoelectric prosthetic hand and target elements of this study]
Elements that are said to be the most important in using a myoelectric hand include weight, certainty of operation, ease of wearing, gripping force, and degree of freedom. In particular, it is difficult to achieve both freedom, reliability of operation, and ease of mounting.

Since the myoelectric prosthetic hand is used as a substitute for the hand and has a function of grasping an object, the certainty of the operation is regarded as very important. In particular, in terms of operational certainty, it is desirable that control be performed with a certainty of at least 80% in order to be used practically.

As for the ease of wearing, there are those related to physical bonding to the living body of the prosthetic hand and those related to calibration. Since prosthetic hands are used on a daily basis, wearing is required to be simple and immediate. For physical joints, many myoelectric prosthetic hands can be installed simply by placing an electromyograph inside the socket and inserting the stump into the socket. As for calibration, it is mainstream that calibration can be controlled only by setting a threshold value of myoelectric potential amplitude because it is necessary to set each time it is worn and the number of setting parameters is extremely small.

[Control method for multi-degree-of-freedom myoelectric hand]
As the number of degrees of freedom increases, it is necessary to improve the degree of freedom of control in order to move the fingers of the degree of freedom freely. In the command method, two electromyographs are placed in the palm flexor muscle group and the dorsiflexor muscle group, and the myoelectric pattern is divided into four categories: weakness, palm flexion, dorsiflexion, and antagonism. Enter and perform a specific action. For example, the extensor and flexor muscles are momentarily antagonized twice, and the forearm prolapses when force is applied to the flexor muscles. The reason why such an operation system is adopted is that, when myoelectric waves show various waveforms depending on the individual, and control is performed based on how to apply force and how much force is applied, the number of amputees that can be used is limited. Therefore, it is common to use On / Off of myoelectricity that can be clearly distinguished and used reliably for each channel of the electromyograph. However, since the surface myoelectric potential measures the electric potential of the muscle on the skin surface, crosstalk occurs in which signals of nearby muscles are mixed. For this reason, it is difficult to calculate the potential of each muscle from the surface electromyogram, and even if many electromyographs are arranged on the skin surface, similar signals are measured in nearby electromyographs. The myoelectricity does not simply increase the degree of freedom in controlling the myoelectric potential even if the number of channels of the electromyograph is increased. From this background, many myoelectric prostheses often place electromyographs at two locations in the flexor and extensor groups that can be contracted separately. Therefore, it is possible to determine 2-bit, that is, 4 states by contracting the flexor and extensor groups separately, but due to the above-mentioned crosstalk problem, further increases in electromyographs are effective. This is the limit for the number of states. These four states are often assigned to rest, forward rotation, reverse rotation, and mode change. As the number of operations performed with the prosthetic hand increases, the number of mode changes increases and the operation becomes complicated, so training for about 4 weeks is required to use it freely. Therefore, a new control method suitable for increasing the degree of freedom of the prosthetic hand is required.

[Previous research on control of multi-degree-of-freedom myoelectric hand]
At the research level, attempts have been made to estimate the user's finger posture directly from surface EMG. In many of them, the myoelectric pattern is classified into a plurality of classes for each kind of finger movement and learned, and the class of the measured myoelectric pattern belongs to which class to realize the action of the prosthetic hand.

Ajiboye et al. Used fuzzy theory to measure and learn EMG 8 times for each EMG pattern, and measured about 97 EMGs in the palm of the wrist, dorsiflexion of the wrist, ulnar displacement, grip, and rest. Successful identification with% accuracy (Non-Patent Document 1). Even the cutting person has an accuracy of about 96%, and the identification accuracy can withstand practical use. However, since it is recognized that ease of wearing is important for practical use of the prosthetic hand, it may be problematic to measure and learn each of the five patterns of myoelectricity eight times. Therefore, it is necessary to prepare a method before use and to maintain a high identification rate. In some cases, we estimated 3 degrees of freedom with an accuracy of about 93% and succeeded in achieving an accuracy of about 89% even in a combined motion that moves multiple degrees of freedom simultaneously. Pattern learning is as fast as 3 seconds, and myoelectric measurement is half as much as Ajiboye et al. In addition, the number of electrodes used is six, which is larger than that of two common myoelectric prosthetic hands, which may make it difficult to align the electrodes when worn. Therefore, a method has been used in which the classifier is composed of two stages, the first stage classifier performs rough classification with a low classification rate, and a filter algorithm is developed to improve the classification rate in the second stage (non-patented). Reference 2). This method showed an improvement in recognition rate of 5-15%. The number of myoelectric patterns to be identified is 15, which is higher than that of existing myoelectric prostheses, and the classification rate is over 90%. However, the number of electrodes is still 12, which is larger than the existing myoelectric prosthesis. A myoelectric identification algorithm having a high number of electrodes and a small calibration time and high identification accuracy is desired.

JP 2010-268404 A JP 2011-03362 A

The currently available multi-degree-of-freedom myoelectric prosthesis does not require calibration, but there are only a few types of myoelectric patterns that can be identified. In the research stage, many myoelectric patterns have been identified, but the number of electrodes required for calibration, the length of time, and the accuracy of identification have become problems, and they have not been put into practical use.

In the myoelectric pattern identification, if the number of classes is increased, more operations can be realized, but if the number of identification classes is increased, the identification rate decreases. In order to improve the identification rate, it is necessary to increase the number of myoelectric electrodes and learning time, but there is a possibility that practicality may be lost.

In order to solve the problem that the discriminator is not sufficiently learned and the discrimination is not stable when the number of myoelectric electrodes and the learning time are limited, in the present invention, in order to control the multi-degree-of-freedom myoelectric prosthetic hand, An object of the present invention is to provide an algorithm that dramatically improves the recognition rate of myoelectric patterns while minimizing time.

The first invention according to the present application is:
A measurement unit for measuring myoelectricity at regular time intervals;
A myoelectric identification unit for identifying which electromyographic pattern of the electromyogram pattern learned in advance should be classified,
A buffer for storing a predetermined number of myoelectric patterns identified by the myoelectric identification unit in time series,
With reference to the predetermined number of myoelectric patterns stored in the buffer and when an occupation ratio exceeding a certain ratio is recognized in a specific myoelectric pattern, control according to the operation command associated with the myoelectric pattern is performed. It is a myoelectric prosthesis characterized by performing.

According to a second invention of the present application, the buffer is a ring buffer, and when the number of myoelectric patterns to be stored reaches the predetermined number, the old myoelectric pattern is discarded. It is a myoelectric prosthesis described in the invention.

A third invention according to the present application is the myoelectric prosthetic hand according to the second invention, characterized in that when the occupation ratio is 50% or more, control according to the specific myoelectric pattern is performed. .

According to a fourth aspect of the present application, in the case where an action that can be manifested simultaneously in the myoelectric prosthetic hand is set as an n pattern, the occupation ratio is 100 / (n + 1)% or more when the occupation ratio is 100 / (n + 1)% or more. The myoelectric prosthetic hand according to the second or third aspect of the invention is characterized in that control is performed simultaneously in accordance with an operation command corresponding to each of a plurality of myoelectric patterns occupying the same.

5th invention which concerns on this application is a myoelectric pattern identification method, and in the apparatus to which myoelectric control is applied, myoelectricity is measured at a fixed time interval, and the myoelectric pattern learned in advance by the apparatus Among the electromyographic patterns to be classified, and the device stores the identified electromyographic patterns in a predetermined number of times in time series, and the device stores the electromyographic patterns. With reference to the predetermined number of myoelectric patterns, when an occupation ratio higher than a certain ratio is recognized in a specific myoelectric pattern, control is performed according to an operation command corresponding to the myoelectric pattern.

6th invention which concerns on this application is a myoelectric pattern identification program performed with the processor of the apparatus to which myoelectric control is applied, Comprising: The procedure which measures myoelectricity by a fixed time interval, and was learned beforehand Among the myoelectric patterns, a procedure for identifying which myoelectric pattern the measured electromyogram should be classified into, and a procedure for storing a predetermined number of the identified myoelectric patterns in chronological order are stored. And a procedure for referring to the predetermined number of myoelectric patterns and performing control according to an operation command corresponding to the myoelectric pattern when an occupation ratio equal to or higher than a certain ratio is recognized in the specific myoelectric pattern.

According to the present invention, it is possible to achieve high-speed learning and a high identification rate with a small number of electrode channels, which is effective for practical control of a multi-degree-of-freedom myoelectric hand.

(A) is a conventional algorithm, (b) is a conceptual diagram showing the algorithm of the present invention. It is a conceptual diagram which shows the structure of the neural network used by this invention. It is a flowchart which shows the algorithm of this invention. It is a schematic diagram which shows the measurement site | part of an electromyograph. (A) is an electromyograph, (b) is a photograph which shows the state which attached the wet electrode to the electromyograph of (a). It is a conceptual diagram which shows the PC screen in experiment. It is a graph which shows the comparison of the recognition rate at the time of mounting the stabilization algorithm of myoelectric identification, and non-mounting. (A) to (d) are graphs showing comparison of identification when the proposed algorithm is installed and when it is not installed.

Hereinafter, specific embodiments to which the present invention is applied will be described in detail.

The present invention is a filter algorithm that acts as a second-layer discriminator that further improves the discrimination result by using the myoelectric discrimination algorithm (Non-patent Document 3) announced by Kato et al. In 2007 as the first-layer discriminator. . Kato et al.'S myoelectric identification algorithm uses backpropagation (BP) neural networks to identify myoelectric patterns. In order to improve this discrimination rate, we propose an algorithm for filtering BP neural networks by focusing on features that are mutually contradictory to events in the output time series. In the BP neural network, the connection strength between nodes constituting the neural network is learned by back-propagation. In the present invention, a hierarchized filter algorithm has been developed that dramatically improves the identification rate without changing the structure of the BP neural network at all. First, FIG. 1A is a schematic diagram showing a conventional myoelectric pattern identification algorithm. On the other hand, the algorithm proposed in the present invention is shown in a schematic diagram in FIG.

In the present invention, the time series of the identified myoelectric pattern is stored in a buffer, and the identification rate is improved by performing the second stage identification based on the time series data. The first-stage identification uses an algorithm based on a BP neural network that has been used conventionally. In this algorithm, first, an electromyographic waveform is measured and stored in a vector emg. Vector emg stores time-series data measured from 3-channel electromyographs. Next, a feature vector X indicating the frequency distribution of the vector emg is extracted from the vector emg by the feature extraction function GFE.

This feature vector X is input as a feature quantity to the input layer of the BP neural network, and as a result, any node in the output layer corresponding to a different hand posture is fired and identified (see FIG. 2). A mechanism for performing this myoelectric pattern identification process is schematically shown as a myoelectric identifier in FIGS. 1 (a) and 1 (b).

Suppose here that the output value in the output layer of the BP neural network is vector L and the number of EMG patterns to be identified is M.

It is expressed. The output of the final BP neural net y, the ID of the myoelectric patterns m _i, the ID is the set of vectors L when the myoelectric patterns of m _i is identified as the output y and O _i

Is determined. If it cannot be identified, -1 is assigned to y. Or myoelectric patterns m _i is assigned to any orientation of the prosthetic hand is programmed in advance by the experimenter. Generally, a smoothing process such as moving average is performed on the myoelectric waveform. However, since the output identified by the neural network has a discrete posture, a simple smoothing process cannot be used. Further, in the smoothing, if the time width is too large, the responsiveness is lowered, and therefore the time width of the smoothing needs to be kept to a minimum. Therefore, attention was paid to the fact that there is a limited number of actions that may occur simultaneously in the target myoelectric pattern. The electromyographic pattern handled in this example is, for example, pronation / extroversion of the forearm, palm flexion / dorsiflexion of the wrist, gripping / opening of the five fingers, bending of the thumb, gripping of the fourth and fifth fingers, Nine patterns in a resting state. For example, palm dorsiflexion of the wrist can be performed while holding the five fingers, but crouching the wrist while bending the wrist means that the wrist is in an antagonistic state, and the current myoelectric prosthetic hand performs antagonistic control. Since it is not performed, it is not necessary to identify. Based on this, the combined action of the three patterns of forearm pronation or pronation, wrist palm flexion or dorsiflexion, and finger pattern is the largest. The myoelectric pattern that can be handled in the present invention is not limited to nine patterns, and may be less than nine patterns or ten or more patterns.

The algorithm was constructed as shown in Fig. 3 based on the fact that there are 3 patterns of movement that may occur simultaneously when the myoelectric pattern is 9 patterns. In this algorithm, the output is determined based on the ratio of the myoelectric patterns included in the time series by referring to the time series data of the myoelectric patterns for a certain time stored in the ring buffer. If the number of myoelectric patterns buffered at a certain time is N, the ring buffer buff _{t at} a certain time _t is

It is expressed. The electromyographic pattern ID is stored in y _i which is a component of buff _t . Thereafter, performed again electromyographic measurements in the next control cycle, after t _l s identification is performed of the new myoelectric patterns y, the ring buffer buff is

And updated. The number S _i myoelectric patterns m _i contained in the ring buffer buff _{t + tl} is

It is expressed. Than this, the ratio P _i occupied by the myoelectric patterns m _i contained in the ring buffer buff

It becomes. In the proposed algorithm, first, it is determined whether the myoelectric pattern is single or includes a plurality of myoelectric patterns. If more than half of the ring buffer buff is an arbitrary myoelectric pattern M _p , it is determined that the ring buffer buff is single, and the action associated with the myoelectric pattern M _p is output to the prosthetic hand.

That is, first, the proposed algorithm refers to N myoelectric patterns stored in the ring buffer buff. When more than half are the myoelectric patterns M _p , the control unit issues an operation command associated with the myoelectric pattern M _p , and the prosthetic hand operates according to the operation command.

When the ring buffer buff includes a plurality of myoelectric patterns M _p1 , M _p2 ..., The number of operations that occur simultaneously is set to 3, so that the myoelectric pattern occupies at least 1/4 of the ring buffer buff. Output all M _p1 , M _p2 . Here, for generalization, if the operation o that is finally output is described as n, the maximum value of the number of operations that can be expressed simultaneously

It becomes. Here, when there are a plurality of myoelectric patterns o to be output, they are not actually output at the same time, and y _i to y _k are sequentially output.

When the number of operations that are simultaneously expressed is n, the threshold occupancy ratio of the ring buffer buff serving as a reference for simultaneous expression is set to 1 / (n + 1), and the ratio of the ring buffer buff to the ratio is 1 / (n + 1). It is desirable to express all of the above myoelectric patterns simultaneously. However, the reference threshold occupation ratio is not limited to 1 / (n + 1), and can be arbitrarily set.

Suppose that the filter algorithm proposed in this study is defined as a function GTS that outputs the output o to the prosthetic hand as a filter result using the time series buff of the identification result as an argument.

Can be described as follows.

[Smooth EMG pattern identification]
For the algorithm GTS proposed in the present invention, when the buffer length is N = 6 and the maximum number n of simultaneous operations is n = 3, it is assumed that buff is given by Equation (11).

On the system of the embodiment, each element in the buff is identified and updated in about 20 ms. In myoelectric prosthetic hand operation, it is considered unlikely that the user has intentionally output a myoelectric pattern that is discontinuously different in only one element as shown in Equation (11) within 20 ms. Must be removed as a misidentification. To solve this problem, the algorithm GTS first calculates S _i and P _{i according} to equation (7).

And calculate. Next, using equation (9)

It can be seen that the myoelectric pattern mixed in the buff can be removed discontinuously.

[Support for multiple operations]
When multiple myoelectric patterns are output at the same time, such as bending the 4th and 5th fingers while palm flexing the wrist, many of those myoelectric patterns appear, so the buff at this time is

It becomes. From Equation (7), S _i and P _i are

It becomes. Next, using equation (9)

Thus, it can be seen that it is also effective when a plurality of myoelectric patterns are output simultaneously.

[Scope of application]
The algorithm proposed in the present invention is effective when the myoelectricity is constantly exerted. However, it is difficult to apply in the case where myoelectricity is instantaneously exhibited. For example, if spiked myoelectricity is output in a short time such as tapping operation, buff sets the weak myoelectric pattern as m ₀

It becomes. At this time, S _i and P _i are expressed by Equation (7).

Thus, the output o to the prosthesis is m _{0 according} to the equation (9).

[Experimental purpose]
Through experiments, the myoelectric pattern identification algorithm according to the present invention was evaluated. We verified how much the electromyographic pattern identification equipped with the proposed algorithm could improve the accuracy compared to the conventional electromyographic pattern identification.

[Subject and experiment setup]
This experiment was conducted with 2 healthy people in their 20s who had experience with the operation of myoelectric prostheses. The electromyograph was attached at three positions (see FIG. 4) of the forearm extensor group, flexor group, and long thumb flexor, and the reference electrode was the elbow. The subject sits in a chair so that the PC monitor can be seen in a comfortable posture. Each item in the myoelectric measurement is as shown in Table 1, and the wet surface electromyograph shown in FIG. 5 was used for the measurement. The buffer length N of the myoelectric pattern ring buffer buff was set to 15. Note that the time required for the ring buffer to be updated after the electromyogram measurement and the myoelectric pattern identification is about 10 ms.

[BP Neural Network Learning Procedures and Conditions]
In the learning process of the myoelectric pattern of the BP neural network, the test subject presses the key of the number of the action he wants to correspond while outputting the myoelectric pattern. Correspondence between movement and number is 0: Rest, 1: Forearm gyrus, 2: Forearm pronation, 3: Wrist palmar flexion, 4: Wrist dorsiflexion, 5: Five finger grip, 6: Five finger open, 7: Mother Finger flexion, 8: 4th and 5th finger flexion. By this work, a specific myoelectric pattern and a specific prosthetic hand movement are labeled, and a neural network is learned by BP. All the combinations of myoelectric patterns taught in the past and their identifying operations are stored, and consideration was given so that the discrimination rate of the last taught myoelectric pattern would not be significantly increased.

As teaching conditions, after the myoelectric pattern to be assigned to each motion was taught once, additional learning was allowed up to one additional time for the myoelectric pattern considered to be insufficient. The time taken to teach one myoelectric pattern varies depending on the similarity between the newly taught myoelectric pattern and the already learned myoelectric pattern, and the first teaching without any other learning data is 100 to 300 ms. The second and subsequent teachings are about 100 to 1000 ms because BP learning continues until a certain identification rate is secured for all other learning data.

FIG. 6 shows a screen displayed on the PC monitor. The target posture is displayed on the left side, and the posture identified from the myoelectric pattern is displayed on the right side. The subject is given a task to control the electromyographic pattern so that the identified posture matches the target posture.

[Experimental procedure]
In the experimental procedure, first, an electromyograph is attached to the subject, and the BP neural network is made to learn the electromyographic pattern of the subject. The subject is then assigned an experimental task and data is collected. Since the myoelectric pattern identification system uses supervised learning using a BP neural network, the system first teaches the subject's myoelectric pattern. Next, the target myoelectric pattern and the identification result are displayed on the PC monitor (see FIG. 6), and the task of matching the identification result to the target myoelectric pattern is performed. The task repeatedly outputs the target myoelectric pattern displayed on the PC monitor for 4 seconds and then weakens for 6 seconds. Each target EMG pattern is displayed five times during one task. At this time, each of the ones with and without the proposed algorithm is performed for about 7 minutes, and the degree of matching of the identified myoelectric pattern with respect to the target myoelectric pattern is evaluated as the discrimination rate.

[Experimental result]
The proposed algorithm is evaluated by comparing the proposed system with the developed algorithm with the conventional system without it. First, the identification rates of both the conventional system and the proposed system are shown in FIG. The identification rate of the algorithm of the embodiment is 82.5% when all nine patterns (indicated as “all movements” in the right half of FIG. 7) are the identification targets, and seven patterns excluding the forearm pronation and pronation from the identification targets In the case of (left half of Fig. 7), it was 92.9%. When the F test was performed with trials with and without the algorithm of the example, the trial for all patterns had a one-sided probability of 0.095, and the trial excluding pronation and pronation was 0.37, both exceeding the 5% level. The t test was performed as equal variance. As a result, both trials showed a significant difference (p <0.001) in the high recognition rate of the proposed method. The improvement of the recognition rate was 17.7% for all 9 patterns, and 11.5% for 7 patterns excluding forearm pronation and pronation.

Fig. 8 shows the time series of identification, and Table 2 below shows the identification rate under each condition. FIG. 8 is a log of the myoelectric pattern identified by the algorithm in CSV format, and the interval between samples is 10 to 20 ms. In addition, the correspondence of the myoelectric pattern and the number in FIG. 8 is as follows: 0: Rest, 1: Forearm supination, 2: Forearm pronation, 3: Wrist palm flexion, 4: Wrist dorsiflexion, 5: Grip of 5 fingers, 6 : 5 finger opening, 7: thumb flexion, 8: 4th finger and 5th finger flexion.

8 (a) to 8 (d) are comparisons of identification when the algorithm of the present invention is installed and when it is not installed. 8A and 8B are graphs of the subject 1, and FIGS. 8C and 8D are graphs of the subject 2. 8A and 8C show the proposed method of the present application, and FIGS. 8B and 8D show the conventional method. The sampling interval on the horizontal axis is 10 to 20 ms. The top and bottom are differences between subjects, and ○ indicates each identification result. The change in the target posture is indicated by a solid line. The target posture is shown on the display only in the portion where the solid line is horizontal. The test subject showed the identification result and the target myoelectric pattern in a representative section in about 7 minutes to output the target myoelectric pattern displayed on the PC monitor. It can be seen from FIG. 8 that the continuity of the myoelectric pattern is maintained and there are few misidentifications as compared with the conventional method.

[About identification rate]
As a result of the experiment, it was found that the discrimination rate changed slightly depending on whether or not pronation and pronation were included. The classification rate when the pronation and pronation were not included exceeded 90%. The identification rate was 82.5% even when including pronation and pronation. When attention is paid to the subject 1 in FIG. 8 (a) showing the time series of identification, it can be seen that the hand is misidentified as gyrus (myoelectric pattern 1) when grasping (myoelectric pattern 5). There is a possibility that it is difficult to distinguish between gripping and pronation because there is force in the pronation direction when grasping. In addition, paying attention to the subject 2 in FIG. 8C, the pronation (myoelectric pattern 2) is also identified in the pronation (myoelectric pattern 1) and the opening (myoelectric pattern 6), and the bias is easily identified. It is suggested that it depends.
[Control cycle]
The proposed method buffers 15 EMG discriminator outputs with a control period of 10 ms. However, as shown in Fig. 3, the delay time of the proposed method is determined by 1/4 or 1/2 of the buffer. Is estimated to be 40-80ms. Considering that myoelectricity is observed several tens to hundreds of milliseconds before the start of joint movement, it is considered to have sufficient responsiveness to be used as a prosthetic hand.

In order to use the myoelectric prosthesis practically, it is desirable that the control is performed with a certainty of at least 80%. In conventional multi-degree-of-freedom myoelectric pattern identification, the number of electrodes (Non-Patent Document 2) and the length of learning time (Non-Patent Document 1) have been problems in order to realize the identification rate. On the other hand, in the present invention, a discrimination rate of 90% or more is realized when seven patterns excluding pronation and pronation are identified with a small number of electrodes of 3 channels and a short learning time of 1 second or less. . Further, even when including pronation and pronation, an identification rate of 82.5% is realized, so that it can be used effectively as a prosthetic hand. In addition, depending on the subject, the recognition rate of over 90% was obtained for 9 patterns of all patterns, and the possibility of a recognition rate of over 90% for all 9 patterns was suggested by proficiency.

In the present invention, by focusing on the time series of the identification results and performing two-stage identification, practical identification of 7 patterns from 3 electrodes, 90% and 9 patterns, 82.5%, without greatly increasing the calculation cost. We built a system that can be identified at a rate. In addition, the learning time of the classifier is less than a dozen seconds with all seven patterns taught, and a control system for a multi-degree-of-freedom prosthesis that can withstand practical use is nearing realization.

Note that the algorithms shown in FIGS. 1B and 3 can be executed by a microprocessor mounted on a device such as a myoelectric prosthesis. In this case, a myoelectric pattern identification program is stored in advance in a memory in the microprocessor, and myoelectric control can be performed by the processor executing the program.

In recent years, there is an increasing need for multi-degree-of-freedom myoelectric prostheses, and a series of research announcements of high-performance multi-degree-of-freedom myoelectric prostheses. Application of myoelectric control to prosthetic legs is also being studied. However, the difficulty of various problems such as countermeasures against myoelectric noise and changes in the electrical properties of the skin has hindered substantial improvement in control freedom. Therefore, in order to make a multi-degree-of-freedom robot hand function as a myoelectric prosthetic hand, a methodology capable of absorbing a characteristic change of a living body in information processing and stably identifying a plurality of myoelectric patterns is required. This application proposes an algorithm that improves the recognition rate of myoelectric patterns. Focusing on the output time series of the classifier, we developed a filter algorithm that dramatically improves the classification rate by being positioned behind the classifier without changing the structure of the classifier itself. In the evaluation experiment, the discrimination rate of the myoelectric pattern was compared between the discriminating system equipped with the developed algorithm and the discriminating system not equipped. The surface electrode 3 channels, the learning time per class was within 1 second, the number of identification classes was set to 2 and 9 conditions, and the pattern identification output was sampled at 10-20 ms. As a result, the accuracy was improved by 11.5% for 7 classes and 17.7% for 9 classes. The classification rate by the proposed method was 82.5% for 9 classes and 92.9% for 7 classes. Although there are few electrode channels, high-speed learning and a high identification rate have been achieved, and it has been proposed a methodology that is highly effective for practical control of a multi-degree-of-freedom myoelectric hand.

This application is based on Patent Application No. 2015-212840 filed with the Japan Patent Office on October 29, 2015, and includes the entire contents thereof.

Claims

A measurement unit for measuring myoelectricity at regular time intervals;
A myoelectric identification unit for identifying which electromyographic pattern of the electromyogram pattern learned in advance should be classified,
A buffer for storing a predetermined number of myoelectric patterns identified by the myoelectric identification unit in time series,
With reference to the predetermined number of myoelectric patterns stored in the buffer, if an occupation ratio exceeding a certain ratio is recognized in a specific myoelectric pattern, control is performed according to an operation command corresponding to the myoelectric pattern. A myoelectric prosthesis characterized by that.
The myoelectric prosthetic hand according to claim 1, wherein when the buffer is a ring buffer and the number of myoelectric patterns to be stored reaches the predetermined number, the old myoelectric pattern is discarded.
The myoelectric prosthetic hand according to claim 2, wherein when the occupation ratio is 50% or more, control according to an operation command corresponding to the specific myoelectric pattern is performed.
When the myoelectric pattern that can be expressed simultaneously in the myoelectric prosthetic hand is set as n pattern, each of the plurality of myoelectric patterns that occupy the occupation rate when the occupation rate is 100 / (n + 1)% or more. 4. The myoelectric prosthetic hand according to claim 2, wherein control according to a corresponding operation command is performed simultaneously.
In devices to which EMG control is applied, measure EMG at regular time intervals,
With the device, among the electromyographic patterns learned in advance, identify which myoelectric pattern should be classified into the myoelectric pattern,
In the device, the identified myoelectric patterns are stored in a predetermined number of times in chronological order,
In the device, referring to the predetermined number of stored myoelectric patterns, and when an occupation ratio higher than a certain ratio is recognized in a specific myoelectric pattern, control according to an operation command corresponding to the myoelectric pattern is performed. Do,
A myoelectric pattern identification method characterized by the above.
A myoelectric pattern identification program executed by a processor of a device to which myoelectric control is applied,
Procedures for measuring myoelectricity at regular time intervals;
A procedure for identifying which electromyographic pattern should be classified among the pre-trained myoelectric patterns,
A procedure for storing a predetermined number of identified myoelectric patterns in chronological order,
A procedure for referring to the predetermined number of stored myoelectric patterns and performing control according to an operation command corresponding to the myoelectric pattern when an occupation ratio of a certain ratio or more in a specific myoelectric pattern is recognized,
A myoelectric pattern identification program comprising: