JP6876324B2 - Signal measuring device and signal measuring method - Google Patents

Description

The present invention relates to a signal measurement technique, and more particularly to simultaneous measurement of myoelectric potential and external force (or impact) and signal separation.

Myoelectric prosthetic limbs are a useful tool for assisting and supporting the elderly and disabled. The movement of the myoelectric prosthetic limb is controlled by using a myoelectric potential (EMG) signal, which is a kind of biological signal. Myoelectric potential is an electric potential generated in a muscle by a command from the brain, and represents a muscle contraction level. When measuring muscle activity from the surface of the skin, it is particularly called surface electromyography (SEMG). By using the EMG signal, the user can control the movement of the artificial limb with a more natural feeling.

In the measurement and application of myoelectric potential (EMG) signals, the biggest problem is the deterioration of the signal-to-noise (S / N: Signal-to-Noise) ratio due to the mixing of an external impact force (external force). Since the S / N ratio deteriorates even with a small external force, it is necessary to remove or correct the influence of the small external force. In order to remove or correct the influence of external force, generally, a high-sensitivity impact sensor or acceleration sensor is installed in addition to the myoelectric sensor, and the information of the myoelectric sensor is corrected by using the information of the impact sensor.

A method has been proposed in which an accelerometer mounted on the arm of a user operating a prosthetic hand device measures a signal according to the movement of the arm and separates the measured signal into an acceleration signal and a mechanomyogram signal (for example). See Patent Document 1). In this method, the movement method of the artificial hand is estimated from the acceleration signal, and the movement of the artificial hand is estimated from the muscle sound signal.

Japanese Unexamined Patent Publication No. 2008-192004

When two types of sensor information are processed by using the myoelectric sensor and the impact sensor to control the movement of the artificial limb worn by the user, the miniaturization of the myoelectric prosthetic limb device is hindered. In the method of separating the acceleration signal and the muscle sound signal from the measurement signal of the acceleration sensor, since the myoelectric potential is not directly measured, it is difficult to make the artificial hand device perform the operation as intended by the user. In the above-mentioned document, the feature pattern vector is extracted by using one or more acceleration sensors, and the operation intended by the user is identified by using the neural network.

An object of the present invention is to provide a technique for effectively separating a myoelectric signal and an external force (impact) information from a measured signal by using a single myoelectric sensor.

In order to solve the above problems, an electrode made of a soft conductive polymer material is used as an electrode of the myoelectric sensor. By using a flexible conductive material having a hardness or elastic modulus lower than that of metal, it is used as a band-passing filter that absorbs a shock wave due to an external force and passes a low frequency component.

As one aspect of the present invention, the signal measuring device is
An electromyographic sensor having at least a pair of electrodes made of a conductive polymer material,
A signal processing unit that separates myoelectric signals and information on external forces from the signals detected by the electrodes of the myoelectric sensor based on frequency components.
Have.

With the above configuration, the myoelectric signal and the external force information can be effectively separated from the signal measured by using a single myoelectric sensor.

It is the schematic of the signal measuring apparatus of an embodiment. It is a figure explaining the principle of signal separation. It is a figure of the setup of the preliminary experiment. It is a figure which shows the EMS waveform and FSR waveform when only myoelectric is input without applying an external force. It is a figure which shows the EMS output waveform of FIG. 4 and the FFT analysis result. It is a figure which shows the measurement result when only the external force of high frequency is input, and the FFT analysis result of EMG output. It is a figure which shows the measurement result when only the external force of a medium frequency is input, and the FFT analysis result of EMG output. It is a figure which shows the EMG waveform and FSR waveform when only a low frequency external force is input. It is a figure which shows the EMG output waveform of FIG. 8 and the FFT analysis result thereof. It is a figure of the EMG waveform and the FSR waveform when both myoelectric and external force are input. It is a figure which shows the EMG output waveform of FIG. 10 and the FFT analysis result. It is a flowchart of the signal measurement method of embodiment.

FIG. 1 is a schematic view of the signal measuring device 1 of the embodiment. The signal measuring device 1 includes an EMG sensor 10 that detects a surface myoelectric signal, and a signal processing unit 20 that separates an external force information such as an impact from the myoelectric signal from the output of the EMG sensor 10. The EMG sensor 10 has at least a pair of detection signals 11a and 11b made of a conductive polymer material. The EMG sensor 10 also has a sensor element 13 that is electrically connected to the detection electrodes 11a and 11b, and a ground electrode 12. In FIG. 1, the sensor element 13 and the signal processing unit 20 are drawn as separate blocks, but the signal processing unit 20 may be realized by a microprocessor and incorporated in the sensor element 13. Alternatively, all the functions of the sensor element 13 may be arranged in the signal processing unit 20, and the outputs of the detection electrodes 11a and 11b may be connected to an amplifier (AMP) provided in the signal processing unit 20. From the viewpoint of avoiding noise mixing due to the movement of the lead wire, it is desirable to arrange the sensor element 13 having an amplification function in the immediate vicinity of the detection electrodes 11a and 11b.

The detection electrodes 11a and 11b are made of a conductive material having a hardness or elastic modulus lower than that of metal, and for example, a conductive polymer material is used. The ground electrode 12 may be made of a conductive polymer material or may be a metal electrode. The pair of detection electrodes 11a and 11b are arranged along the direction of the muscle fiber of the portion to be measured, and the potential difference between the two electrodes is detected as a surface myoelectric potential signal. If necessary, a conductive adhesive layer may be formed on the contact surfaces of the detection electrodes 11a and 11b with the skin. The ground electrode 12 is an electrode that determines a potential reference, and is attached on the skin such as a joint having few muscles.

The sensor element 13 amplifies the surface myoelectric potential signal detected as a potential difference between the detection electrode 11a and the detection electrode 11b and outputs the signal to the signal processing unit 20. Generally, the surface myoelectric potential signal is mixed with background noise and noise due to an impact applied to the skin. The signal separation unit 21 of the signal processing unit 20 separates the myoelectric signal from the signal detected by the EMG sensor 10 and the signal component due to the impact force or the external force. As a separation method, a bandpass filter that passes a predetermined band may be used, or a predetermined range of the power spectrum may be extracted by performing an FFT (Fast Fourier Transform) analysis. For example, the signal separation unit 21 digitally samples an analog input signal and performs signal processing to obtain a signal component having a signal separation threshold value (here, 100 Hz) or more and a myoelectric signal detection upper limit value (here, 1 kHz) or less. The signal component below the signal separation threshold (here, 100 Hz) may be extracted as an external force. Alternatively, a band-pass filter (BPF) that splits the analog input signal into two and passes the signal component from the signal separation threshold (here, 100 Hz) to the detection upper limit of the myoelectric signal (here, 1 kHz). The signal may be separated by passing it through a low-pass filter (LPF) that passes a signal component below the signal separation threshold (here, 100 Hz). The myoelectric signal extracted by the signal separation unit 21 is supplied to the myoelectric signal acquisition unit 22, and the external force information is supplied to the external force information acquisition unit 23.

When the signal measuring device 1 is used for measuring, for example, the muscle contraction level or the degree of tension, the myoelectric signal obtained by the myoelectric signal acquisition unit 22 may be output as it is, or after rectification and smoothing. It may be output. When the signal measuring device 1 is used for measuring muscle fatigue, the frequency information analyzed by FFT may be output to analyze the shift state of the frequency spectrum corresponding to the myoelectric signal. When the signal measuring device 1 is applied to the myoelectric prosthesis, the acquired myoelectric signal may be converted into a motor control signal and output.

The external force information obtained by the external force information acquisition unit 23 can be used for environmental estimation at the time of measurement, analysis of the influence on muscle contraction, and the like. When the signal measuring device 1 is applied to the myoelectric prosthetic limb, the external force information can be used for detecting a load on the user or for correcting a motor control signal.

FIG. 2 is a diagram illustrating the principle of signal measurement and signal separation of the embodiment. The upper row shows the output waveform of the EMG sensor when both myoelectricity (expression of muscular strength) and external force are input, and the horizontal axis shows time and the vertical axis shows electric potential. The lower part of FIG. 2 shows the FFT analysis result of the output waveform of the upper EMG sensor, the horizontal axis shows the frequency, and the vertical axis shows the amplitude. The spectrum that draws the gentle envelope that appears in the region of 100 Hz to 1 kHz, more specifically 100 Hz to 500 Hz, is the power due to muscle contraction. The spectrum appearing in the lower frequency domain is the power caused by the external force. The sharp peak seen around 50 Hz is hum noise caused by a commercial power source.

As will be described later, in order to effectively extract the myoelectric signal, the inventors detect the biological signal using an electrode having a soft property, and telegraph a frequency component of 100 Hz to 1 kHz, more preferably 100 Hz to 500 Hz. I found that it is better to extract it as a number. It was also found that the nature of the external force can be distinguished by the peak position of the envelope of the spectrum in the region lower than 100 Hz. In FIG. 2, 20 Hz is set as the lower limit of the frequency, but a frequency of 20 Hz or less (for example, 10 Hz, 5 Hz, etc.) may be set as the lower limit.

By using the detection electrodes 11a and 11b of the conductive polymer having a hardness or elastic modulus smaller than that of the metal, the detection electrodes 11a and 11b can be used as an interference filter or a band passing filter for an external shock wave. The mechanical impedance characteristics of the detection electrodes 11a and 11b made of the soft polymer material are lower than those of the metal electrodes. The impact generated by the movement of the biological part to which the EMG sensor 10 is attached or the application of an external force is absorbed by the signal electrodes 11a and 11b and superimposed on the surface myoelectric potential signal as a slowly changing shock wave.

As the electrode material having such properties, a conductive silicone resin, polyisopyrene, polybutadiene, other elastomer to which an appropriate amount and type of electric dopant is added, and the like can be used. As the dopant, conductive carbon powder, germanium powder or the like can be used.

FIG. 3 is a preliminary experimental setup that supports the findings of the invention described with reference to FIGS. 1 and 2. The conditions for the preliminary experiment are as follows.
(1) The measurement site is the body surface near the flexor digitorum superficialis forearm.
(2) The measurement system uses an FSR (Force Sensing Resistor) sensor and an EMG sensor.
(3) The measurement method is to attach the FSR sensor directly above the EMG sensor and install it on the body surface.

According to these conditions, as shown in FIG. 3A, the conductive silicone rubber electrodes 11a and 11b connected to the EMG sensor element 23 are attached to the body surface directly above the flexor digitorum superficialis muscle of the forearm, and the EMG sensor element is attached. The FSR sensor 30 is placed on the 13 and measured. The electrodes 11a and 11b of the conductive silicone rubber and the EMG sensor element 23 are collectively referred to as an "EMG sensor". The ground electrode 12 is attached to an appropriate place such as an indirect vicinity on the back side of the conductive polymer electrodes 11a and 11b. The impact force is measured using the FSR sensor 30, and the surface myoelectric potential signal is measured using the EMG sensor.

FIG. 3B is an image of the setup actually used. The FSR sensor 30 is a commercially available pressure sensor. For the preliminary experiment, the EMG sensor element 13 was electrically connected to the detection electrodes 11a and 11b of the conductive polymer to prepare the EMG sensor 10. The ground electrode 12 is also connected to the EMG sensor element 13, but is hidden behind the sensor mounting supporter in FIG. 3 (B).

The setup of FIG. 3 is for confirming what kind of signal information is included in the output of the EMG sensor, and the FSR sensor 30 is arranged for the purpose of confirming the presence or absence of an external force. Using the setup shown in Fig. 3, acquire measurement data for (i) inputting only myoelectric force, (ii) inputting only external force, and (iii) inputting both myoelectric force and external force. .. The measurement results of the preliminary experiment are shown in FIGS. 4 to 11.
<Preliminary experiment 1>
4 and 5 show the results of preliminary experiment 1. In preliminary experiment 1, only myoelectricity is input. Record the EMG sensor output and FSR sensor output when the flexor digitorum superficialis muscle is contracted so that only myoelectricity is output while suppressing the movement of the forearm as much as possible. The upper side of FIG. 4 is the output waveform of the EMG sensor, and the lower side is the output waveform of the FSR sensor 30. The output of the EMG sensor shows an EMG waveform with an amplitude exceeding 0.5 V, but the output of the FSR sensor 30 keeps a constant value at around 0.2 V. From this measurement result, it can be seen that no external force such as impact force is applied to the forearm.

FIG. 5A shows the first EMG measurement and the FFT analysis result in the absence of an external force. FIG. 5B shows the second EMG measurement in the absence of external force and the FFT analysis result thereof. In both FIGS. 5 (A) and 4 (B), a power of sufficient intensity is detected between 100 Hz and 1 kHz. On the other hand, in the frequency region below 100 Hz, no power waveform for estimating the external force is observed. The amplitudes seen in the vicinity of 50 Hz, 100 Hz, and 200 Hz in the FFT analysis are the 50 Hz hum noise of the commercial power supply and its aliasing. From the preliminary experiment 1, the existence of the frequency domain in which the myoelectric signal is observed is confirmed.
<Preliminary experiment 2>
6 to 9 show the results of the preliminary experiment 2. In the preliminary experiment 2, the EMG sensor output and the FSR sensor output when only an external force is input are recorded. The flexor digitorum superficialis muscle is relaxed so as not to generate a myoelectric potential as much as possible, and the surface of the FSR sensor 30 is pressed. The output of the EMG sensor and the output of the FSR sensor 30 are acquired by changing the pushing speed with respect to the FSR sensor 30.

FIG. 6A shows the output waveforms of the EMG sensor and the FSR sensor 30 when only a high frequency external force is input, and FIG. 6B shows the EMG waveform and its FFT analysis result. The high frequency here means a relatively high frequency in the frequency domain lower than the frequency domain of the myoelectric signal, and the shock wave generated when the surface of the FSR sensor 30 is quickly pressed is used as the external force of the high frequency. ..

As shown in FIG. 6A, when an external force acts, the measured value of the EMG sensor fluctuates in synchronization with the fluctuation of the SFR value. Even when the muscular force is not expressed, an external force such as acceleration is applied to the EMG sensor, so that the EMG sensor is output as myoelectric information, which causes a malfunction of the device to which the EMG sensor is applied. High-frequency external force is generated, for example, when a user wearing a myoelectric prosthetic hand quickly shakes his arm.

From the FFT analysis result of FIG. 6B, it can be seen that there is almost no amplitude in the region above 100 Hz, but there is power over the entire region below 100 Hz. The power spectrum observed in the region below 100 Hz indicates that an external force is applied in the absence of myoelectric motion.

FIG. 7A shows the output waveforms of the EMG sensor and the FSR sensor 30 when only a medium frequency external force is input, and FIG. 7B shows the EMG waveform and its FFT analysis result. The medium frequency means a medium frequency in a frequency region lower than the frequency region of the myoelectric signal, and a shock wave generated when the surface of the FSR sensor 30 is pressed at a medium speed is subjected to a medium frequency external force. And.

In the FFT analysis result of FIG. 7B, almost no amplitude is observed in the region above 100 Hz, but the power spectrum is observed in the region below 100 Hz. This power spectrum is shifted to the slightly lower frequency side as compared with FIG. 6 (B).

FIG. 8 shows the output waveforms of the EMG sensor and the FSR sensor 30 when only a low frequency external force is input. FIG. 9A shows the EMG waveform of the first measurement and the FFT analysis result thereof, and FIG. 9B shows the EMG waveform of the second measurement and the FFT analysis result thereof. Here, the low frequency means a relatively low frequency in the frequency domain lower than the frequency domain of the myoelectric signal, and the shock wave generated when the surface of the FSR sensor 30 is slowly pressed is referred to as a low frequency external force. To do.

In the FFT analysis results of FIGS. 9 (A) and 9 (B), almost no power is observed in the region of 100 Hz or higher except for a slight measurement noise, but a strong power spectrum is observed in the region of several tens of hertz. ing. The power spectrum is further shifted to the lower frequency side as compared with FIG. 7 (B).

From the results of FIGS. 6 to 9, it can be seen that by frequency analysis of the EMG sensor output, not only the existence of an external force or an impact force but also the nature of the external force can be determined.
<Preliminary experiment 3>
10 and 11 show the results of the preliminary experiment 3. In the preliminary experiment 3, the EMG sensor output and the FSR sensor output when both myoelectric force and external force are input are recorded. As shown in FIG. 10, when external force and muscular activity act at the same time, it is not possible to determine whether it is myoelectric or external force only by the output of the EMG sensor.

FIG. 11 (A) shows the first EMG measurement result and the FFT analysis result when the external force and the muscular force are applied at the same time, and FIG. 11 (B) shows the second EMG measurement result when the external force and the muscular force are applied at the same time. And its FFT analysis result are shown. By FFT analysis of the EMG sensor output, it is possible to clearly distinguish between the envelope-shaped power spectrum appearing at 100 Hz to 1 kHz and the power spectrum in the frequency region below 100 Hz. In the region of less than 100 Hz, the amplitude is observed exclusively in the region of less than 30 Hz, so that it can be determined that a low frequency external force is applied.

From the results of preliminary experiments 1 to 3, even when EMG measurement is performed in a situation where external force is applied due to movements such as swinging the arm or raising the leg, a single EMG sensor is used to obtain myoelectric signals and external force information (impact, etc.). Can be taken out. At that time, a soft conductive polymer electrode is used to slowly respond to the input of an external force. In other words, the flexibility or low mechanical impedance characteristics of the conductive polymer are utilized as an external force bandpass filter. In the case of a hard metal electrode, the responsiveness (change in impedance) due to a pressure change is large, and it is difficult to detect an external force in a frequency region of less than 100 Hz.

FIG. 12 is a flowchart of the signal measurement method of the embodiment. First, a signal is acquired by an EMG sensor having a conductive polymer electrode (S11). By using a soft conductive polymer material as the detection electrode, the pass band of the shock wave due to the external force is lowered, and the distinction from the frequency band of the myoelectric signal is promoted.

The myoelectric signal and the external force information are separated from the signal detected by the conductive polymer electrode pair (S12). The signal separation may be performed by frequency-analyzing the output of the EMG sensor to extract components in a predetermined frequency band, or by using a bandpass filter that passes a signal in a predetermined band. Since the applied external force or shock wave is converted into a low frequency band by the conductive polymer electrode, the external force and the myoelectric signal can be effectively separated.

Based on the extracted external force information, it is possible to judge the nature of the external force, for example, whether it is a high-frequency external force, a low-frequency external force, a force applied by a fast movement, or a force applied by a slow movement. Good (S13). On the other hand, the extracted myoelectric signal is output after undergoing appropriate processing (S14). Although steps S13 and S14 are not essential elements of the signal measurement method of the present invention, they can be appropriately executed depending on the application situation.

As described above, according to the signal measuring method and the signal measuring device of the embodiment, the impact force and the myoelectric signal from the outside are detected from the same electrode pair, and the impact force and the myoelectric signal are separated and taken out. Can be done. By utilizing the softness of the conductive polymer, it is possible to relax the pass band of the impact force to the lower side and facilitate the separation from the pass band of the myoelectric signal. Since the noise caused by the external force is effectively removed from the myoelectric signal, the S / N ratio of the myoelectric signal is improved. In addition, the nature of the external force can be determined from the frequency information of the external force.

The present invention is not limited to the above-described embodiment, and includes some modifications. The electrode of the conductive polymer used in the EMG sensor is not limited to a pair of electrodes, and a plurality of electrode pairs can be used as needed. By forming each of the plurality of electrode pairs with a conductive polymer, the influence of an external force can be effectively removed from the myoelectric signal of each portion.

1 Signal measuring device 10 EMG sensor (myographic sensor)
11a, 11b Detection electrode 12 Ground electrode 13 Sensor element 20 Signal processing unit 22 Myoelectric signal acquisition unit 23 External force information acquisition unit

Claims (9)

An electromyographic sensor having at least a pair of electrodes made of a conductive polymer material,
A signal processing unit that separates myoelectric signals and information on external forces from the signals detected by the electrodes of the myoelectric sensor based on frequency components.
Have a,
The signal processing unit is a signal measuring device characterized by extracting a frequency component of less than 100 Hz as external force information from the output signal of the myoelectric sensor. An electromyographic sensor having at least a pair of electrodes made of a conductive polymer material,
A signal processing unit that separates myoelectric signals and information on external forces from the signals detected by the electrodes of the myoelectric sensor based on frequency components.
Have,
A signal measuring device characterized in that the electrode formed of the conductive polymer material functions as a band-passing filter that absorbs a shock wave due to an external force and passes a frequency component lower than the myoelectric signal. The signal measuring apparatus according to claim 1 or 2 , wherein the conductive polymer material is selected from an elastomer to which a conductive silicone rubber, polyisopyrene, polybutadiene, or an electric dopant is added. The signal measuring device according to any one of claims 1 to 3, wherein the signal processing unit extracts a frequency component of 100 Hz to 1 kHz as a myoelectric signal from the output signal of the myoelectric sensor. Before SL signal processing unit, the signal measurement apparatus according to claim 1, characterized in that to determine the nature of the external force in accordance with the peak positions of the frequency spectrum to be extracted as the external force information. A step of detecting a surface myoelectric potential signal with at least a pair of electrodes made of a conductive polymer material,
A step of separating the myoelectric signal and the information about the external force from the surface myoelectric potential signal based on the frequency component,
Have a,
A signal measurement method characterized by frequency-analyzing the surface myoelectric potential signal and extracting a frequency component of less than 100 Hz as external force information. A step of detecting a surface myoelectric potential signal with at least a pair of electrodes made of a conductive polymer material,
A step of separating the myoelectric signal and the information about the external force from the surface myoelectric potential signal based on the frequency component,
Have,
Signal measuring how to characterized in that the functioning of the electrode formed of the conductive polymer material, as a band pass filter which passes frequency components lower than the myoelectric signal by absorbing shock waves by an external force. The signal measuring method according to claim 6 or 7, wherein the surface myoelectric potential signal is frequency-analyzed and a frequency component of 100 Hz to 1 kHz is extracted as a myoelectric signal. The step of determining the nature of the external force according to the peak position of the frequency spectrum extracted as the external force information,
The signal measuring method according to claim 6, further comprising.

Images