VIBRATION DETECTION DEVICE AND VIBRATION DETECTION METHOD

Abstract

A vibration detection device comprises: a radio wave sensor (11) for detecting a change in the electromagnetic near field; an inertial sensor (12); and a signal processing device (2) for inputting waveform data from the radio wave sensor and waveform data from the inertial sensor into a learning model and for acquiring a vibration inference result from the learning model inferred by the learning model.

Description

TECHNICAL AREA

[0001] The present disclosure relates to a technique for detecting vibrations. STATE OF THE ART

[0002] Patent literature 1 discloses a technique relating to a pulse wave information processing device that processes pulse wave information indicating a pulse wave of a passenger of a moving body, and comprises an information acquisition unit that can acquire first pulse wave information acquired from a first pulse wave information acquisition device, and second pulse wave information acquired from a second pulse wave information acquisition device using a method different from that of the first pulse wave information acquisition device, together with reliability grade information indicating the reliability of the second pulse wave information, and an information selection unit that selects based on the reliability grade information.which of the first pulse wave information and which of the second pulse wave information should be output as the passenger's pulse wave information. REFERENCE LIST PATENT LITERATURE

[0003] Patent Literature 1: JP 2020-103461 A SUMMARY OF THE INVENTIONAL TECHNICAL TASK

[0004] According to the technique described in patent literature 1, one of the first pulse wave information and the second pulse wave information is selected alternatively as the pulse wave information of the passenger based on the reliability grade information, which indicates the reliability of the second pulse wave information, and therefore the problem is that the measurement accuracy decreases if data is missing in the selected first pulse wave information or second pulse wave information.

[0005] The present disclosure was made in response to the recognition of such a problem, and one objective of an aspect of the present disclosure is to provide a vibration detection technique that can prevent a reduction in measurement accuracy due to a lack of data. TECHNICAL SOLUTION

[0006] One aspect of a vibration detection device according to the embodiment of the present disclosure comprises: a radio wave sensor for detecting a change in the electromagnetic near field; an inertial sensor; and a signal processing device for inputting waveform data from the radio wave sensor and waveform data from the inertial sensor into a learning model and for acquiring a vibration result derived by the learning model from the learning model. ADVANTAGEOUS EFFECTS OF THE INVENTION

[0007] A vibration detection device according to the embodiment of the present disclosure can prevent a reduction in measurement accuracy due to missing data. BRIEF DESCRIPTION OF THE DRAWINGS Fig. Figure 1A is a configuration representation of a vibration detection device. Fig. Figure 1B is a flowchart showing an operating example of the vibration detection device. Fig. Figure 2 is a configuration representation of a vibration detection device. Fig. 3A is a view showing an example of a frame antenna. Fig. 3B is a view showing an example of a meander dipole antenna. Fig. Figure 4 is a configuration representation of a signal processing device. Fig. Figure 5 is a configuration representation of a vibration detection device. Fig. Figure 6 is a configuration representation of a vibration detection device. Fig. Figure 7 is a configuration representation of a signal processing device. Fig. Figure 8 is a flowchart of the adaptive control of an adjustment circuit performed by the vibration detection device. Fig. Figure 9 is a configuration representation of a vibration detection device. Fig. Figure 10 is a view showing a waveform of the vibration detection device and an electrocardiograph. Fig. Figure 11 is a view illustrating a comparison result obtained by comparing a root mean square error (RMSE) value between a peak interval of a heartbeat waveform derived using a conditional variational auto-encoder (CVAE) and a peak interval of an electrocardiogram waveform that is a correct response value per person. DESCRIPTION OF THE EXAMPLES OF EXECUTION

[0008] Exemplary embodiments of the present disclosure are described in detail below with reference to the accompanying drawings. It should be noted that identical or similar parts in the drawings are identified by identical or similar reference numerals, thus eliminating the need for redundant descriptions of these parts. Furthermore, the term "or" in this disclosure means a comprehensive logical OR unless otherwise specified. Example 1. <konfiguration>

[0009] A configuration of a vibration detection device D1 according to embodiment 1 is described with reference to Fig. 1A described. Fig. Figure 1A is a configuration diagram showing the vibration detection device D1 according to embodiment 1. As shown in Fig. As shown in Figure 1A, the vibration detection device D1 comprises a sensor unit 1 and a signal processing device 2.

[0010] The sensor unit 1 comprises a radio wave sensor 11 and an inertial sensor 12. The radio wave sensor 11 is a near-field electromagnetic radio wave sensor that uses a small antenna with a wavelength smaller than that used by the radio wave sensor 11 and that detects a change in the near-field electromagnetic field of the radio wave sensor 11. A small antenna with low radiation efficiency is used to detect only a change in the near-field electromagnetic field of the radio wave sensor 11. In the context of this disclosure, "near-field" means the inner area of ​​a region of a sphere whose distance from the radio wave sensor 11 is λ / 2π (λ: wavelength) or less.

[0011] By positioning the radio wave sensor 11 near a detection target, the radio wave sensor 11 can observe any movement of the detection target. The "detection target" is any object located near the radio wave sensor 11. While typical examples of a detection target are a human lung, a human heart, or a part of a device, an object such as the lung of a non-human animal can also be a detection target. For example, let's assume the detection target is a human heart. When the heart moves, the human body surface area changes due to the movement of the heart. This fluctuation of the human body surface area alters the electromagnetic field. Therefore, by positioning the radio wave sensor 11 near the human body surface, it is possible to observe the movement of the human heart.

[0012] The inertial sensor 12 is a sensor that detects inertial motion. Examples of inertial sensors 12 include triaxial accelerometers and gyroscopes. By positioning the inertial sensor 12 near a detection target, it can detect local vibrations caused by the target's movement. Local vibration refers to vibration at the location where the inertial sensor 12 is positioned.

[0013] The signal processing device 2 comprises an inference unit 2A and a learning model 2B. A configuration example for the hardware of the signal processing device 2 is described in embodiment 2. The learning model 2B is a learning model that learns in advance a correlation between the information obtained from the radio wave sensor 11 and the inertial sensor 12 and a peak of a waveform, which is a desired signal. A detailed example of the learning model 2B is described later in embodiment 5.

[0014] The sensor unit 1 is positioned near a detection target, such as a human lung or heart. The arrangement of the sensor unit 1 described above makes it possible to detect changes in the electromagnetic field or local oscillations caused by the movement of the detection target more accurately. <funktionsweise>

[0015] Next, the functionality of the vibration detection device D1 will be explained using the following examples: Fig. 1B described. Fig. Figure 1B is a flowchart illustrating the operation and functionality of the vibration detection device D1. In step ST1, the radio wave sensor 11 detects changes in its electromagnetic near field. By positioning the radio wave sensor 11 close to the detection target, it detects changes in the electromagnetic field caused by the target's movement. Based on these changes in the electromagnetic near field, a volume change of the detection target, which the radio wave sensor 11 approaches without physical contact, can be determined.

[0016] In step ST2, which is performed in parallel to step ST1, the inertial sensor 12 acquires non-contact local vibrations caused by the movement of the detection target.

[0017] It should be noted that the ‘non-contact type’ described here means that the radio wave sensor 11 and the inertial sensor 12 are not in contact with human skin or any device.

[0018] By combining the sensors of the radio wave sensor 11 and the inertial sensor 12, whose detection principles are different, and by obtaining a signal that a single sensor cannot detect, it is possible to compensate for the deterioration in accuracy due to missing data.

[0019] In step ST3, the inference unit 2A of the signal processing device 2 inputs the information received from the radio wave sensor 11 and the inertial sensor 12 into the learning model 2B and derives the oscillation (waveform) of the desired signal as an inference result from the learning model 2B. The inference result includes a peak of the waveform. As described above, the data from both sensors are input into the learning model 2B to derive the oscillation, so that even if data is missing in one sensor output, it is possible to compensate for the degradation in accuracy caused by the missing data and to improve the signal-to-noise ratio of the desired signal by deriving oscillations from the data of the other sensor.

[0020] The 2B learning model can be pre-trained with a correct answer value, or it can be trained sequentially with the obtained data. The correct answer value here refers, for example, to information with guaranteed accuracy, such as an electrocardiogram, e.g., when training on the peak of a heartbeat waveform.

[0021] Using the peak of the acquired waveform, which is the desired signal, as the feature value, and the number of channels from the acquired data, a variety of elements of chronological data can be read and inferred. Inference is not limited to algorithms and can be a supervised learning, semi-supervised learning, or reinforcement learning algorithm, as long as the algorithms allow the above operation.

[0022] As described above, the vibration detection device D1 comprises the sensor unit 1 and the signal processing device 2, and the learning model 2B of the inference unit 2A contained in the signal processing device 2 derives a peak of a waveform using all the information obtained by the radio wave sensor 11 and the inertial sensor 12 contained in the sensor unit 1, so that it is possible to achieve an effect such that it is possible to provide a vibration detection device that can compensate for a deterioration in accuracy due to missing data and extract a peak of the vibrations as a feature amount and has a high signal-to-noise ratio. Design 2. <konfiguration>

[0023] The construction of the vibration detection device D1 was described in embodiment 1. A vibration detection device D2, which is a more specific aspect of the vibration detection device D1, is described as embodiment 2 with reference to the Fig. 2, Fig. 3 to Fig. 4 described. Fig. Figure 2 is a configuration representation showing the vibration detection device D2 according to embodiment 2. Fig. 3A and Fig. Figure 3B shows examples of antennas used for the radio wave sensor, and Fig. Figure 4 is a configuration representation of a signal processing device. It should be noted that the same reference symbols are used in the Fig. 2, Fig. 3A, Fig. 3B and Fig. 4 as in Fig. 1. Designate identical or corresponding parts.

[0024] As in Fig. As shown in Figure 2, the radio wave sensor 11 comprises an antenna 111, a signal generation unit 3, a divider circuit 41 and a detection circuit 5 and obtains vibrations of a target to be detected without contact using a radio wave.

[0025] Antenna 111 is a small antenna whose wavelength is shorter than a usable wavelength, and it can be either a loop antenna or a dipole antenna. As in Fig. As shown in Figure 2, antenna 111 includes a connection 1111.

[0026] Fig. Figure 3A shows the example of a double loop antenna 1112, and Fig. Figure 3B shows an example of a meandering dipole antenna 1113. Fig. 3A and Fig. 3B is the antenna 1112 or 1113 connected to terminal 1111. In a case where a wavelength of a useful frequency λ is, the length of one side of a rectangular shape of an outermost shape surrounding a conducting medium that forms the antenna is in any case λ / 10 or less.

[0027] The inertial sensor 12 is a sensor that detects inertial motion and includes an output terminal 1201, as shown in Fig. 2 shown.

[0028] The signal generation unit 3 is a circuit that generates a high-frequency signal and includes an output terminal 3001. The signal generation unit 3 generates, for example, the high-frequency signal, which is a continuous wave signal.

[0029] The divider circuit 41 is a high-frequency circuit comprising three terminals 4101, 4102, and 4103, which divides an input signal from terminal 4103 to terminals 4101 and 4102. Since terminals 4101 and 4102 are isolated from each other, the signal input to terminal 4101 is output only to terminal 4103 and not to terminal 4102. Because only the output terminals of the divider circuit 41 need to be isolated, the circuit can be configured with a Wilkinson divider or a directional coupler whose termination resistor is connected to an isolation terminal.

[0030] The detection circuit 5 comprises two terminals, 5001 and 5002, and detects a signal input to terminal 5001, generates a detection signal, and outputs the detection signal to terminal 5002. A quadrature detection circuit or similar can be used for the detection circuit. In this case, two input terminals and two output terminals must be provided, and an I-signal (in phase) and a Q-signal (quadrature) are generated and output as detection signals. The Q-signal is a signal whose phase differs from that of the I-signal by 90 degrees.

[0031] As in Fig. As shown in Figure 4, the signal processing device 2 comprises an analog-to-digital converter 22, which converts an analog signal output by the detection circuit 5 and an analog signal output by the inertial sensor 12 into digital signals, a storage device 23, which stores values ​​of the digital signals converted by the analog-to-digital converter 22, and a processor 21, which controls a circuit or derives data.

[0032] The analog-to-digital converter 22 converts the analog signal (sensing signal) output by the sensing circuit 5 and the analog signal (sensing signal) output by the inertial sensor 12 into digital signals that the processor 21 can control.

[0033] The storage medium 23 is a generic term for memory such as read-only memory (ROM) or random-access memory (RAM) and external storage media such as a hard drive. A program or data used by the processor 21 is stored in the storage device 23, and the processor 21 reads the program stored in the storage device 23 to perform various operations, including inputting and outputting data to and from the learning model 2B. The storage medium 23 is also used as temporary data storage. Furthermore, the output signal of the sensing circuit 5 and the output signal of the inertial sensor 12 are also read from and written to the storage device 23.

[0034] The connection of the individual terminals is then described. Terminal 1111 of antenna 111 is connected to terminal 4103 of divider circuit 41, terminal 3001 of signal generation unit 3 and terminal 4101 of divider circuit 41 are connected, and terminal 5001 of detection circuit 5 and terminal 4102 of divider circuit 41 are connected. <funktionsweise>

[0035] The following describes the operation of the vibration detection device D2. First, a signal output from the output terminal 3001 of the signal generation unit 3 is fed into terminal 4101 of the divider circuit 41 and output to terminal 4103. The signal output from terminal 4103 is fed into terminal 1111 of the antenna 111 and radiated as a radio wave by the antenna 111. The radio wave radiated by the antenna 111 is reflected by the detection target or an object adjacent to the detection target and fed into the antenna 111 as a signal of a reflected wave. The reflected input signal of the antenna 111 is output from terminal 1111 and fed into terminal 4103 of the divider circuit 41.The reflected signal, which is input to terminal 4103 of the divider circuit 41, is output by terminal 4102 of the divider circuit 41, input to terminal 5001 of the detection circuit 5 and output as a detection signal from terminal 5002 to the signal processing device 2.

[0036] When a vibration signal of a part caused by the movement of the detection target is detected, the inertial sensor 12 installed near the detection target outputs the detected detection signal from terminal 1201 of the inertial sensor 12 to the signal processing device 2.

[0037] The signal processing device 2 converts the signal data acquired by the antenna 111 and the inertial sensor 12 from analog signals into digital signals in the analog-to-digital converter 22, which are accepted by the processor 22, and stores the signals converted into digital signals and the data calculated by the processor 21 in the storage device 23, which has the function of temporarily or semi-permanently storing the signals and the data.

[0038] Furthermore, components that can be mounted on a substrate and perform each function described in the present embodiment constitute the vibration sensing device D2, so that all functions of the vibration sensing device D2 can be implemented on the substrate. For example, the antenna 111 can consist of a copper foil pattern or of components that can be mounted on the substrate and perform the functions of the inertial sensor 12, the signal generation unit 3, the divider circuit 41, the sensing circuit 5, and the signal processing device 2. At this stage, the dielectric constant and the layer configuration of the substrate are design considerations, and any material and layer configuration can be used as long as the material or layer configuration fulfills the functions of the vibration sensing device D2.

[0039] As described above, the vibration detection device D2 comprises the sensor unit 1 with the radio wave sensor 11 and the inertial sensor 12, as well as the signal processing device 2. The radio wave sensor 11 comprises the antenna 111, the signal generation unit 3, the divider circuit 41, and the detection circuit 5. The signal processing device 2 comprises the processor 21, the analog-to-digital converter 22, and the storage device 23. Similar to embodiment 1, the learning model 2B of the signal processing device 2 derives a peak of a waveform using all the information obtained from the radio wave sensor 11 and the inertial sensor 12 contained in the sensor unit 1, thus making it possible to compensate for the degradation in accuracy due to missing data and to implement the vibration detection device D2, which can operate with a high signal-to-noise ratio in a compact and cost-effective manner. Example 3.

[0040] A configuration in which a quadrature detection circuit is used as the detection circuit 5 in embodiment 2 and a three-axis accelerometer is used as the inertial sensor 12 is described in embodiment 3. Fig. Figure 5 is a configuration diagram showing a vibration detection device D3 according to the present embodiment. It should be noted that the same reference numerals are used in Fig. 5 as in Fig. 1. To denote identical or corresponding parts.

[0041] As in Fig. As shown in Figure 5, the radio wave sensor 11 of the vibration detection device D3 comprises the antenna 111, the signal generation unit 3, the divider circuits 41 and 42 and a quadrature detection circuit 5A and detects the vibrations caused by the movement of a detection target without contact via a radio wave.

[0042] A triaxial accelerometer 12A detects inertial motion along three axes: X, Y, and Z. The triaxial accelerometer 12A has three output terminals: 1201, 1202, and 1203. The triaxial accelerometer 12A outputs an X-signal (the X-axis detection signal) to output terminal 1201, a Y-signal (the Y-axis detection signal) to output terminal 1202, and a Z-signal (the Z-axis detection signal) to output terminal 1203.

[0043] The divider circuit 42 is a high-frequency circuit comprising three terminals 4201, 4202, and 4203, which divides an input signal from terminal 4201 to terminals 4202 and 4203. Since terminals 4202 and 4203 are isolated from each other, the signal input to terminal 4202 is output only by terminal 4201 and not by terminal 4203.

[0044] Since only the output terminals of the divider circuit 42 need to be isolated, the divider circuit 42 can be configured with a Wilkinson divider or a directional coupler whose termination resistor is connected to an isolation terminal.

[0045] The quadrature detection circuit 5A comprises two input terminals (the input terminal 5001 and one input terminal 5003) and two output terminals (the output terminal 5002 and one output terminal 5004) and performs quadrature detection on a signal that is input to the input terminal 5001, using a local signal that is input to the input terminal 5003, outputs an I signal (In Phase) from the output terminal 5002 and outputs a Q signal (Quadrature), whose phase differs by 90 degrees from that of the I signal, to the output terminal 5004.

[0046] The signal processing device 2 converts the X signal, the Y signal, the Z signal, the I signal and the Q signal, which are signals generated by the three-axis accelerometer 12A and the quadrature detection circuit 5A, from analog signals into digital signals in the analog-to-digital converter 22 and temporarily or semi-permanently stores the signals converted into digital signals in the storage device 23.

[0047] The connection of the individual terminals is then described. Terminal 3001 of signal generation unit 3 and terminal 4201 of divider circuit 42 are connected, terminal 4202 of divider circuit 42 and terminal 4101 of divider circuit 41 are connected, terminal 4103 of divider circuit 41 and terminal 1111 of antenna 111 are connected, terminal 4102 of divider circuit 41 and terminal 5001 of quadrature detection circuit 5A are connected, and terminal 4203 of divider circuit 42 and terminal 5003 of quadrature detection circuit 5A are connected.

[0048] The following describes the operation of the vibration detection device D2. A high-frequency signal is output from terminal 3001 of the signal generation unit 3 and input to terminal 4201 of the divider circuit 42. The input signal is split into two parts by the divider circuit 42, and the split signals are output from terminals 4202 and 4103. A signal output from terminal 4202 of the divider circuit 42 is input to terminal 4101 of the divider circuit 41 and output from terminal 4103 of the divider circuit 41. The output signal is input to terminal 1111 of the antenna 111 and radiated as a radio wave by the antenna 111. The radio wave radiated by the antenna 111 is reflected by the detection target or an object near the detection target and fed back into the antenna 111 as a signal of a reflected wave.The reflected signal input to antenna 111 is output from terminal 1111 and input to terminal 4103 of divider circuit 41. The reflected signal input to terminal 4103 of divider circuit 41 is output to terminal 4102. The signal output from terminal 4203 of divider circuit 42 is input to terminal 5003 of quadrature detection circuit 5A, and the signal output from terminal 4102 of divider circuit 41 is input to terminal 5001 of quadrature detection circuit 5A. These signals undergo quadrature detection by quadrature detection circuit 5A, and the I signal and the Q signal are output from output terminal 5002 and output terminal 5004, respectively, to signal processing device 2.

[0049] Furthermore, the signal detected by the triaxial accelerometer 12A is output from the output terminals 1201, 1202 and 1203 of the triaxial accelerometer 12A to the signal processing device 2.

[0050] Furthermore, components that can be mounted on a substrate and perform each function described in the present embodiment constitute the vibration sensing device D3, so that all functions of the vibration sensing device D3 can be implemented on the substrate. For example, the antenna 111 can be formed from a copper foil pattern or consist of components that can be mounted on the substrate and perform the functions of the triaxial accelerometer 12A, the signal generation unit 3, the divider circuits 41 and 42, the quadrature sensing circuit 5A, and the signal processing device 2. At this stage, the dielectric constant and the layer configuration of the substrate are design considerations, and any material and layer configuration can be used as long as the material or layer configuration fulfills the functions of the vibration sensing device D3.

[0051] As described above, in the present embodiment, the vibration detection device D3 is configured to use the quadrature detection circuit 5A as the detection circuit 5 in embodiment 2 and the triaxial accelerometer 12A as the inertial sensor 12. The learning model 2B of the signal processing device 2 derives a peak of a waveform by using all the information from the I, Q, X, Y, and Z signals obtained from the radio wave sensor 11 and the triaxial accelerometer 12A contained in the sensor unit 1. This makes it possible to compensate for the degradation in accuracy due to missing data and to implement the vibration detection device D3, which can operate with a high signal-to-noise ratio in a compact and cost-effective manner. Example 4.

[0052] A configuration in which an adaptive matching circuit is added between the divider circuit and the antenna in embodiment 2 or 3 is described in embodiment 4. Fig. Figure 6 is a configuration representation showing a vibration detection device D4 according to the present embodiment. Although Fig. Figure 6 shows a configuration in which the adaptation circuit is added to embodiment 2; the adaptation circuit can also be used in a case of embodiment 3 according to Fig. 6 are added between the divider circuit and the antenna. Fig. Figure 7 is a functional diagram showing a signal processing device according to the present embodiment. Fig. Figure 8 is a flowchart of an adaptive control program for the adaptation circuit. It should be noted that the same reference symbols are used in the Fig. 6, Fig. 7 to Fig. 8 as in Fig. 1. Designate identical or corresponding parts.

[0053] As in Fig. As shown in Figure 6, the matching circuit 6 is arranged between the divider circuit 41 and the antenna 111. The matching circuit 6 is a circuit with three terminals 6001, 6002, and 6003, which matches a load impedance connected to terminal 6002 to the impedance of a circuit connected to terminal 6001 and can vary its transmit and reflect characteristics. The third terminal 6003 is also connected to ground (GND).

[0054] The matching circuit 6 contains a variable element that can be adaptively controlled, and the variable element can be a diode with variable capacitance, a variable air capacitor or a variable polycapacitor.

[0055] Each terminal of the matching circuit 6 is connected as follows. Terminal 6001 is connected to terminal 4103 of the divider circuit 41, terminal 6002 is connected to terminal 1111 of the antenna 111, and terminal 6003 is grounded.

[0056] Furthermore, the matching circuit 6 can adjust a circuit constant depending on a change in the load impedance connected to the matching circuit 6 by adaptively changing a value of the variable element according to a control signal sent by the signal processing device 2.

[0057] Fig. Figure 7 is a configuration representation of the signal processing device 2, and a digital-to-analog converter 24, which converts digital signals into analog signals, is shown in addition to those in Fig. The 4 components shown are provided.

[0058] The digital-to-analog converter 24 converts a digital signal, which is a calculation result of the processor 21, into an analog signal that the matching circuit 6 can process. It should be noted that the digital-to-analog converter 24 can be implemented using either hardware that performs specific processing or a program that performs the same processing.

[0059] Next, a setup procedure for the matching circuit will be described using the flowchart in Fig. As described in section 8, when switched on according to a control program, the processor 21 of the vibration detection device D4 stores in the storage device 23 a detection signal (referred to for simplicity as the "Id signal") from the detection circuit 5 or the quadrature detection circuit 5A, which has been converted into a digital signal, as well as the value of the variable element of the matching circuit 6 (step ST101). Subsequently, in each step, an operation is performed by the processor 21 in accordance with the control program.

[0060] In a case where a specific measurement time T seconds does not elapse (NO: step ST102), signal recording continues. In a case where the T seconds elapse (YES: step ST103), an absolute time-averaged value of Id signals for the T seconds stored in storage device 23 is calculated (step ST103). It should be noted that signal recording continues if the shutdown processing is not performed, even if a time-averaged calculation or other processing is carried out.

[0061] If the absolute value of the time-averaged values ​​of the Id signals falls below a certain threshold Δ (NO: ST1041), processing returns to ST101. If this absolute value is equal to or greater than the certain threshold Δ (YES: ST1042), a matching value (the value of the variable element) is calculated to match a load impedance connected to terminal 6002 of matching circuit 6 with the impedance of a circuit connected to terminal 6001 (step ST105). The calculated matching value is then output to matching circuit 6 (step ST106).

[0062] Next, it is determined whether the shutdown processing is executed with respect to processor 21 or not. In a case where the shutdown processing is not executed (NO: step ST1071), the processing returns to ST101. In a case where the shutdown processing is executed (YES: ST1072), processor 21 executes the shutdown processing and terminates the adaptive control processing. Fig. 8.

[0063] As described above, by arranging the matching circuit 6, which can be adaptively controlled, between the antenna 111 and the divider circuit 41, it is possible to implement the vibration detection device D4, in which a signal detected by the antenna 111 has a high S / N value. Example 5.

[0064] A measurement method in a case where a Conditional Variational Auto-Encoder (CVAE) is implemented in machine learning as described in embodiments 1 to 4 is presented in embodiment 5. As an example of the measurement method, a procedure for capturing a heartbeat waveform is described. Fig. Figure 9 is a configuration representation of a vibration detection device D5 according to embodiment 5.

[0065] The vibration detection device D5 includes the frame antenna 1112 in Fig. 3A in the sensor unit in Fig. 5. As in Fig. As shown in 9, the vibration detection device D5 uses a configuration that includes the matching circuit 6 in Fig. 6 and the signal processing device 2 in Fig. 7 contains an adaptive control processing of the adaptation circuit in Fig. 8 carries out.

[0066] The loop antenna 1112 described in the present embodiment has two loops, and the length of one side of the rectangular shape of the outermost shape surrounding the loop antenna 1112 is λ / 100 of the operating frequency. It should be noted that the length described here is only one aspect, and that in a case where the antenna is small and the length of the outermost shape is λ / 10 or less of the operating frequency, the details of the types, shapes, sizes, and the like are not limited, and various aspects can be selected according to a target type or the regulations of the radio law.

[0067] In the vibration detection device D5 with the in Fig. In the configuration shown in Figure 9, it is possible to form the vibration detection device D5 on a substrate by forming the antenna 1112 with the copper foil pattern and using surface mounting parts such as the signal generation unit 3 and other circuits.

[0068] The copper foil-structured antenna 1112 in the vibration detection device D5 is positioned near a body surface close to the heart, without touching the body surface. The vibration detection device D5 is, for example, attached to clothing.

[0069] Fig. Figure 10 shows an example of a waveform (heartbeat / respiration waveforms) detected by the vibration detection device D5 arranged on the clothing in a case where the radio wave sensor 11 including the matching circuit 6 and an algorithm (matching circuit setting algorithm) in connection with the adaptive control processing of the matching circuit in Fig. 8 can be used. In Fig. Figure 10 shows a waveform labeled "Module" that was detected by the vibration detection device D5. The term "Development Sensor" is used in... Fig. 10 refers to the vibration detection device D5. At this point, an electrocardiogram waveform is simultaneously recorded by synchronizing the waveform recorded by the vibration detection device D5 with a time. In Fig. Figure 10 shows a waveform labeled "ECG" (electrocardiogram), which is not illustrated. By adaptively controlling the matching circuit using the matching circuit setting algorithm, it is possible to improve the signal-to-noise ratio of a signal detected by the radio wave sensor 11 and to capture a waveform with a peak synchronized with a peak time of the electrocardiogram waveform, thus providing a correct response value. Although a VHF band (30 to 300 MHz) was used as the frequency of a radio wave emitted by the radio wave sensor 11, this frequency is only an example, and other frequencies can also be used.

[0070] Next, an example of a method for accurately predicting the peak value of a heartbeat using machine learning is described. The radio wave sensor 11 and the triaxial accelerometer 12A, which together form the vibration detection device D5, are sensors with different signal acquisition principles, yet they can still detect vibrations caused by the movement of a detection target. Five time-series data points of the I-signal, Q-signal, X-signal, Y-signal, and Z-signal can be acquired by these two sensors.

[0071] The CVAE is a model that can detect a difference in the distribution per class with a conditional probability by learning a dataset with a class label based on a Variational Auto-Encoder (VAE), which is a deep generative model of an autoencoder network structure configured as a probabilistic graphical model with a latent variable. In this embodiment, the class label refers to the information of a correct response value. It should be noted that the CVAE is merely an example of the Learning Model 2B, and the Learning Model 2B is not limited to the CVAE.

[0072] The I, Q, X, Y, and Z signals are input values, and an electrocardiogram waveform is provided as the correct response signal to generate class label data (training data) in which a captured signal and the correct response signal match. The class label data (training data) obtained here are stored in storage device 23 of signal processing device 2. At the time of inference, the captured I, Q, X, Y, and Z signal data are fed into training model 2B to derive the electrocardiogram waveform. When a peak of a heartbeat waveform is captured, a window function of the triangular type or similar is prepared, and peaks and other components are classified and learned to further improve the accuracy of the inferences.The difference in the distribution per class is captured in the conditional probability in order to perform the inference, thus reducing noise.

[0073] Fig. Figure 11 is a view illustrating a comparative result obtained by comparing a mean squared error (RMSE) value between a peak interval of a heartbeat waveform derived using CVAE per learning data and a peak interval of an electrocardiogram waveform representing a correct response value for each individual. There are four subjects: A, D, F, and H.

[0074] Fig. Figure 11 illustrates 1) a case in which training is performed using the I-signal and the Q-signal obtained from a radio wave sensor (legend: only an electromagnetic near-field sensor) and an electrocardiogram waveform, 2) a case in which training is performed using the X-signal, the Y-signal and the Z-signal obtained from a triaxial accelerometer (legend: an accelerometer alone) and an electrocardiogram waveform, and 3) a case in which training is performed using the I-signal, the Q-signal, the X-signal, the Y-signal and the Z-signal obtained by a sensor configuration (legend: the electromagnetic near-field sensor and the accelerometer) in the present embodiment and an electrocardiogram waveform.

[0075] As from Fig. As can be seen in Figure 11, an error is reduced by performing the training using all data obtained from the sensor configuration described in the present embodiment, instead of performing the training using data obtained from a single sensor. Accordingly, it is possible to achieve an improvement in inference accuracy according to the present embodiment.

[0076] As described above, by machine learning using all signals obtained from the configuration with the radio wave sensor and the three-axis accelerometer, and by deriving a peak of a waveform, it is possible to obtain a vibration detection device that can compensate for a deterioration in accuracy due to missing data and operates with a high S / N ratio. <Ergänzende Hinweise>

[0077] Some aspects of the various embodiments described above can be summarized as follows. (Supplementary note 1)

[0078] A vibration detection device according to Supplementary Note 1 comprises: a radio wave sensor (11) to detect a change in the electromagnetic near field; an inertial sensor (12); and a signal processing device (2) to input waveform data from the radio wave sensor and waveform data from the inertial sensor into a learning model and to obtain from the learning model a result of the vibration derived by the learning model. (Supplementary note 2)

[0079] A vibration detection device according to Supplementary Note 2 is the vibration detection device described in Supplementary Note 1, and the radio wave sensor, the inertial sensor and the signal processing device are arranged on an identical substrate. (Supplementary note 3)

[0080] A vibration detection device according to Supplementary Note 3 is the vibration detection device described in Supplementary Note 1 or 2, the radio wave sensor detects the change in the electromagnetic field caused by a movement of the lungs of a person who is a detection target, the inertial sensor detects a local vibration caused by the movement of the lungs, and the signal processing device infers a peak in the person's breathing as the vibration. (Supplementary note 4)

[0081] A vibration detection device according to Supplementary Note 4 is the vibration detection device described in Supplementary Note 1 or 2, wherein the radio wave sensor detects the change in the electromagnetic field caused by a movement of the heart of a person who is a detection target, the inertial sensor detects a local vibration caused by the movement of the heart, and the signal processing device derives a peak in the person's breathing as the vibration. (Supplementary note 5)

[0082] A vibration detection device according to Supplementary Note 4 is the vibration detection device described in one of Supplementary Notes 1 to 4, and as described in Fig. As shown in Figure 2, the radio wave sensor includes a signal generation unit (3) for generating a high-frequency signal, an antenna (111) for transmitting the generated high-frequency signal, a detection circuit (5) for detecting an input signal, generating a detection signal and outputting the generated detection signal, a divider circuit (41) for outputting an input signal to a first terminal (4101) from a third terminal (4103), and for dividing and outputting a signal input to the third terminal (4103) to the first terminal (4101) and a second terminal (4102), wherein the divider circuit (41) includes the first terminal (4101), the second terminal (4102) and the third terminal (4103), and the first terminal is connected to the signal generation unit, the second terminal to the detection circuit and the third terminal to the antenna. (Supplementary note 6)

[0083] A vibration detection device according to Supplementary Note 6 is the vibration detection device described in one of Supplementary Notes 1 to 4, as shown in Fig. As shown in Figure 6, the radio wave sensor includes a signal generation unit (3) for generating a high-frequency signal, an antenna (111) for transmitting the generated high-frequency signal, a detection circuit (5) for detecting an input signal, generating a detection signal and outputting the generated detection signal, a divider circuit (41) for outputting a signal input to a first terminal (4101) from a third terminal (4103) and for dividing and outputting a signal input to the third terminal (4103) to the first terminal (4101) and a second terminal (4102), wherein the divider circuit (41) includes the first terminal (4101), the second terminal (4102) and the third terminal (4103), and the first terminal is connected to the signal generation unit and the second terminal is connected to the detection circuit, and an matching circuit (6) which includes a fourth terminal (6001),which is connected to the third terminal of the divider circuit, and includes a fifth terminal (6002) which is connected to the antenna and contains a variable element, and the signal processing device outputs a control signal to control a setting value of the variable element to the matching circuit in order to match an impedance of the fourth terminal of the matching circuit and an impedance of the fifth terminal based on a value of the detection signal output by the detection circuit. (Supplementary note 7)

[0084] A vibration detection device according to Supplementary Note 7 is the vibration detection device described in one of Supplementary Notes 1 to 4, as shown in Fig. As shown in Figure 9, the radio wave sensor includes a signal generation unit (3) for generating a high-frequency signal, an antenna (111) for transmitting the generated high-frequency signal, a detection circuit (5) for detecting an input signal, generating a detection signal, and outputting the generated detection signal, a first divider circuit (41) for outputting a signal input from a third terminal (4103) to a first terminal (4101), and for dividing and outputting a signal input to the third terminal (4103) to the first terminal (4101) and a second terminal (4102), the first divider circuit (41) comprising the first terminal (4101), the second terminal (4102), and the third terminal (4103), with the second terminal connected to the detection circuit, and a second divider circuit (42) to divide a signal input from a fourth terminal (4201) to a fifth terminal (4202). to output a signaland to divide a signal input to the fourth terminal (4201) and output it to the fifth terminal (4202) and a sixth terminal (4203), wherein the second divider circuit (42) comprises the fourth terminal (4201), the fifth terminal (4202) and the sixth terminal (4203), and the fourth terminal is connected to the signal generation unit, the fifth terminal to the first terminal of the first detection circuit and the sixth terminal to the detection circuit, and an matching circuit (6) comprising a seventh terminal (6001) connected to the third terminal of the first divider circuit and an eighth terminal (6002) connected to the antenna and comprising a variable element, the detection circuit (quadrature detection circuit 5A) performs quadrature detection on a signal input from the first divider circuit and a signal input from the second divider circuit as local signals,to generate an IQ signal, and outputs the generated IQ signal as the detection signal, and the signal processing device outputs a control signal to control a setting value of the variable element to the matching circuit in order to match an impedance of the seventh terminal of the matching circuit and an impedance of the eighth terminal based on a value of the detection signal output by the detection circuit. (Supplementary note 8)

[0085] A vibration detection device according to Supplementary Note 8 is the vibration detection device described in one of Supplementary Notes 5 to 7, and in a case where the antenna has a rectangular shape and a wavelength of the radio frequency signal is λ, a length of one side of the rectangular shape of an outermost shape of the antenna is λ / 10 or less. (Supplementary note 9)

[0086] A vibration detection method according to Supplementary Note 9 is a vibration detection method of a vibration detection device comprising a radio wave sensor (11), an inertial sensor (12) and a signal processing device (2), and comprising: a step of detecting a change in the electromagnetic near field (ST1) by the radio wave sensor; a step of detecting a local vibration of a detection target (ST2) by the inertial sensor; and a step of inputting waveform data from the radio wave sensor and waveform data from the inertial sensor into a learning model by the signal processing device and detecting a vibration result derived by the learning model (ST3) from the learning model.

[0087] It should be noted that the embodiments can be combined, and that each embodiment can be modified or omitted as required. INDUSTRIAL APPLICABILITY

[0088] The vibration detection device according to the present disclosure can be used as a device for obtaining, for example, vital data such as a person's heart rate. REFERENCE MARK LIST

[0089] 1: Sensor unit, 2: Signal processing device, 2A: Inference unit, 2B: Learning model, 3: Signal generation unit, 5: Detection circuit, 5A: Quadrature detection circuit, 6: Matching circuit, 11: Radio wave sensor, 12: Inertial sensor, 12A: Triaxial accelerometer, 21: Processor, 22: Analog-to-digital converter, 23: Storage device, 24: Digital-to-analog converter, 41: Divider circuit, 42: Divider circuit, 111: Antenna, 1112: Loop antenna, 1113: Meander dipole antenna, D1 to D5: Vibration detection device QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] JP 2020-103461 A

[0003] < / funktionsweise> < / konfiguration> < / funktionsweise> < / konfiguration>

Claims

Vibration detection device comprising: a radio wave sensor for detecting changes in the electromagnetic field in the near field; an inertial sensor; and a signal processing device for inputting waveform data from the radio wave sensor and waveform data from the inertial sensor into a learning model and for acquiring a vibration result derived from the learning model. Vibration detection device according to claim 1, wherein the radio wave sensor, the inertial sensor and the signal processing device are arranged on an identical substrate. Vibration detection device according to claim 1 or 2, wherein the radio wave sensor detects the change in the electromagnetic field caused by the movement of the lungs of a person who is a detection target, the inertial sensor detects local vibrations caused by the movement of the lungs, and the signal processing device infers a peak value of the person's respiration from the vibrations. Vibration detection device according to claim 1 or 2, wherein the radio wave sensor detects the change in the electromagnetic field caused by the movement of the heart of a person who is a detection target, the inertial sensor detects local vibrations caused by the movement of the heart, and the signal processing device infers a peak value of the person's respiration from the vibrations. Vibration detection device according to one of claims 1 to 4, wherein the radio wave sensor comprises a signal generation unit for generating a high-frequency signal, an antenna for transmitting the generated high-frequency signal, a detection circuit for detecting an input signal, for generating a detection signal and for outputting the generated detection signal, and a divider circuit for outputting a signal input to a first terminal from a third terminal and for dividing a signal input to the third terminal and outputting it to the first terminal and a second terminal, wherein the divider circuit comprises the first terminal, the second terminal and the third terminal and the first terminal is connected to the signal generation unit, the second terminal is connected to the detection circuit and the third terminal is connected to the antenna. Vibration detection device according to any one of claims 1 to 4, wherein the radio wave sensor comprises a signal generation unit for generating a high-frequency signal, an antenna for transmitting the generated high-frequency signal, a detection circuit for detecting an input signal, for generating a detection signal and for outputting the generated detection signal, a divider circuit for outputting a signal input to a first terminal from a third terminal and for dividing a signal input to the third terminal and outputting it to the first terminal and a second terminal, wherein the divider circuit comprises the first terminal, the second terminal and the third terminal and the first terminal is connected to the signal generation unit and the second terminal is connected to the detection circuit, and an matching circuit with a fourth terminal which is connected to the third terminal of the divider circuit.and a fifth terminal connected to the antenna, and to a variable element, and the signal processing device outputs a control signal to control a setting value of the variable element to the matching circuit in order to match an impedance of the fourth terminal of the matching circuit and an impedance of the fifth terminal based on a value of the detection signal output by the detection circuit. Vibration detection device according to one of claims 1 to 4, wherein the radio wave sensor comprises a signal generation unit for generating a high-frequency signal, an antenna for transmitting the generated high-frequency signal, a detection circuit (5) for detecting an input signal, for generating a detection signal and for outputting the generated detection signal, a first divider circuit for outputting a signal input to a first terminal from a third terminal and for splitting and outputting a signal input to the third terminal, to the first terminal and a second terminal, wherein the first divider circuit comprises the first terminal, the second terminal and the third terminal and the second terminal is connected to the detection circuit, a second divider circuit for outputting a signal input to a fifth terminal from a fourth terminal and for splitting and outputting a signal input to the fourth terminal,to the fifth terminal and a sixth terminal, wherein the second divider circuit comprises the fourth terminal, the fifth terminal and the sixth terminal, and the fourth terminal is connected to the signal generation unit, the fifth terminal is connected to the first terminal of the first detection circuit and the sixth terminal is connected to the detection circuit, and a matching circuit with a seventh terminal connected to the third terminal of the first divider circuit, and an eighth terminal connected to the antenna, and with a variable element, the detection circuit performs quadrature detection on a signal input from the first divider circuit and a signal input from the second divider circuit as local signals to generate an IQ signal, and outputs the generated IQ signal as the detection signal,and the signal processing device outputs a control signal to the matching circuit to control a setting value of the variable element in order to match an impedance of the seventh terminal of the matching circuit and an impedance of the eighth terminal based on a value of the sensing signal output by the sensing circuit. Vibration detection device according to one of claims 5 to 7, wherein in a case where the antenna has a rectangular shape and a wavelength of the high-frequency signal is λ, a length of one side of the rectangular shape of an outermost shape of the antenna is λ / 10 or less. Vibration detection method of a vibration detection device comprising a radio wave sensor, an inertial sensor and a signal processing device, wherein the vibration detection method comprises: a step of detecting a change in the electromagnetic field in the near range by the radio wave sensor; a step of detecting local vibrations of a detection target by the inertial sensor; and a step of inputting waveform data from the radio wave sensor and waveform data from the inertial sensor into a learning model by the signal processing device and acquiring a vibration inference result inferred by the learning model.

Images