WO2023074598A1

WO2023074598A1 - Filtering device and filtering method

Info

Publication number: WO2023074598A1
Application number: PCT/JP2022/039434
Authority: WO
Inventors: 裕幸三田村; 竜太綿貫
Original assignee: 裕幸三田村
Priority date: 2021-10-28
Filing date: 2022-10-24
Publication date: 2023-05-04
Also published as: JP7024983B1; JP2023066157A

Abstract

The purpose of the present invention is to accurately realize demodulation in numerical phase detection. A sampling unit 1 samples an alternating current signal with a sampling period corresponding to each frequency to be extracted. Here, the sampling period is set such that one or half period of each frequency to be extracted is an integral multiple of the sampling period. A filtering unit 3 performs numerical phase detection on the sampled alternating current signal to filter the alternating current signal and thus acquire a frequency component to be extracted.

Description

Filtering device and filtering method

The present invention relates to a technique for filtering AC signals.

Analog phase detection and numerical phase detection (also called digital phase detection) are known as techniques for performing phase detection (also called synchronous detection) for AC signals.

In analog phase detection, if each frequency of the signal to be demodulated is ω ₀ , the original signal is multiplied by 2 cos _{ω 0} t or 2 sin ω ₀ t to remove the double vibration components (i.e., cos 2 ω ₀ t and sin 2 ω ₀ t). By doing so, a constant component that does not depend on time is extracted. Here, in the analog phase detection, since the component of double vibration is removed by the low-pass filter, there is a problem that integration for a long time is required.

On the other hand, numerical phase detection has the advantage of being able to extract signal components in a short period of time because it is possible to remove components of double vibrations by using integral multiples or half-integer multiples of the reference period.

By the way, in numerical phase detection, if the number of integrated vibrations (in other words, the number of cycles used for interval integration) is small, there is a problem that the noise removal performance is significantly degraded. On the other hand, if the number of integrated vibrations is increased, there arises a problem that it takes time to extract the signal component. In other words, in numerical phase detection, there is a trade-off relationship between reduction in processing time and noise removal performance.

Therefore, the inventor of the present invention proposed a technique in Patent Document 1 below that can perform low-noise signal extraction processing in a short period of time while using numerical phase detection. That is, in the technique of Patent Document 1 below, by taking advantage of the fact that the characteristics of the pass-through gain differ depending on the difference in the number of integrated vibrations in numerical phase detection, a single integrated vibration is obtained by linearly combining the detection results from a plurality of the number of integrated vibrations. It was shown that it is possible to give a pass gain characteristic that is not found in the detection by the number of times.

In addition, as a development of the technique of Patent Document 1 below, the present inventors have found in Patent Document 2 below that the position of the frequency component is set to a specific position, thereby shortening the time required to separate the multiplexed frequency components. showed how it can be done.

International Publication 2018/163364 Japanese Patent No. 6905265

By the way, in these techniques, the AC signal to be filtered is generally a continuous function of time. On the other hand, actual signal acquisition (that is, sampling) is performed discretely. Then, in the separation and demodulation of a plurality of frequencies, it becomes necessary to adjust the sampling method so that an integral value (or a half-integer value) of integrated vibration counts can be realized. The aforementioned Patent Document 1 (paragraph 0090) proposes a solution of setting the frequency so that the reciprocal of the frequency is an integral multiple of the reciprocal of the sampling rate. However, in this case, if it is attempted to match the frequency allocation conditions for efficient filtering as in Patent Document 2, the restrictions on possible frequencies become severe.

The present invention has been made in view of the situation described above. A main object of the present invention is to provide a technique for accurately realizing demodulation in numerical phase detection. Another object of the present invention is to provide a technique capable of increasing the degree of freedom in setting frequencies to be extracted in numerical phase detection.

The means for solving the above problems can be described as the following items.

(Item 1)
A filtering device for filtering an AC signal,
a sampling unit, a detection unit, and a linear combination unit,
The sampling unit is configured to sample the AC signal at a sampling period corresponding to each frequency to be extracted,
Here, the sampling period is set so that one period or half period of each frequency to be extracted is an integral multiple of the sampling period,
The detection unit is configured to obtain a plurality of detection results corresponding to the number of integrated vibrations by performing numerical phase detection on the sampled AC signal,
The filtering device, wherein the linear combination unit linearly combines the plurality of detection results using a linear coefficient corresponding to a filter characteristic.

(Item 2)
The filtering device according to item 1, wherein the sampling section is configured to perform sampling of the AC signal in parallel at a sampling period corresponding to each frequency to be extracted.

(Item 3)
The filtering device according to item 1, wherein the sampling unit changes the sampling period according to fluctuations in the frequency of the AC signal so as to follow the fluctuations.

(Item 4)
4. The filtering device according to any one of items 1 to 3, wherein the frequency components to be extracted in the AC signal are arranged at regular intervals.

(Item 5)
In addition, it has an output section,
5. The filtering device according to any one of items 1 to 4, wherein the output unit is configured to output a result obtained by linear combination in the linear combination unit.

(Item 6)
Furthermore, it has a linear coefficient determination part,
The linear coefficient determination unit is configured to calculate the linear coefficient using the relationship between the linear coefficients, the overall pass gain in the numerical phase detection, and the target filter characteristics. 6. The filtering device according to any one of 5.

(Item 7)
The filtering device according to any one of items 1 to 6, wherein at least two of the sampling periods corresponding to the different frequencies to be extracted are common.

(Item 8)
A filtering method for filtering an AC signal, comprising:
sampling the AC signal with a sampling period corresponding to each frequency to be extracted, wherein the sampling period is such that one period or half period of each frequency to be extracted is an integral multiple of the sampling period; is set to
obtaining a plurality of detection results corresponding to the number of integrated vibrations by performing numerical phase detection on the sampled AC signal;
and linearly combining the plurality of detection results using linear coefficients according to filter characteristics.

(Item 9)
A filtering method for filtering an AC signal, comprising:
sampling the AC signal with a sampling period corresponding to each frequency to be extracted, wherein the sampling period is such that one period or half period of each frequency to be extracted is an integral multiple of the sampling period; is set to
Filtering the sampled AC signal by performing numerical phase detection on the AC signal to obtain frequency components to be extracted.

(Item 10)
A computer program for causing a computer to execute each step according to item 8 or 9.

This computer program is recorded on an appropriate recording medium (for example, an optical recording medium such as a CD-ROM or DVD disk, a magnetic recording medium such as a hard disk or flexible disk, or a magneto-optical recording medium such as an MO disk). can be stored. This computer program can be transmitted via a communication line such as the Internet.

(Item 11)
A filtering device for filtering an AC signal,
It comprises a sampling section and a filter section,
The sampling unit is configured to sample the AC signal at a sampling period corresponding to each frequency to be extracted, wherein the sampling period is one period or half period of each frequency to be extracted. It is set to be an integral multiple of the period,
The filter unit filters the AC signal by performing numerical phase detection on the sampled AC signal, and acquires a frequency component to be extracted.

According to the present invention, it is possible to accurately realize demodulation in numerical phase detection. Moreover, according to the present invention, it is possible to increase the degree of freedom in setting the frequencies to be extracted in the numerical phase detection.

It is a block diagram showing an outline of a filtering device concerning one embodiment of the present invention. 2 is a block diagram showing an outline of a filter unit in the filtering device of FIG. 1; FIG. 2 is a flowchart for explaining a filtering method performed using the filtering device of FIG. 1; FIG. 4 is an explanatory diagram for explaining a procedure for sampling and demodulating a mixed signal in which AC signals are mixed; FIG. 4 is an explanatory diagram for explaining a procedure for sampling a plurality of AC signals whose frequencies are evenly spaced, in comparison with the OFDM method; The numbers on the vertical axis in these figures represent frequency (Hz), and the vertical axis represents amplitude for each frequency. FIG. 10 is an explanatory diagram for explaining an example in which the filtering method of the present embodiment is applied to frequency sweep; The vertical axis in these figures indicates normalized signal intensity.

A filtering device according to an embodiment of the present invention will be described below with reference to FIG.

(Configuration of filtering device of this embodiment)
The filtering device of this embodiment includes a sampling unit 1 and a filter unit 3 as main components (see FIG. 1). Furthermore, the filtering device of this embodiment includes a sampling rate storage unit 2, an output unit 4, and a linear coefficient determination unit 5 as additional elements.

(Sampling section)
The sampling unit 1 is configured to sample an AC signal at a sampling period corresponding to each frequency to be extracted. Here, the sampling period is set so that one period of each frequency to be extracted is an integral multiple of the sampling period. Here, suitable sampling periods corresponding to different frequencies are generally considered to be different from each other. However, as long as the condition that "one period of each frequency to be extracted is an integral multiple of the sampling period" is satisfied, any of the sampling periods may be common. Therefore, the number M of sampling periods corresponding to N frequencies can be any of N=M, N>M, and M=1. A detailed sampling operation will be described later.

(Sampling rate storage unit)
The sampling rate storage unit 2 stores the sampling period corresponding to the frequency to be extracted (which is usually known). The sampling rate storage unit 2 provides the sampling unit 1 with a predetermined sampling period.

(filter part)
The filter section 3 of this embodiment is composed of a detection section 31 and a linear combination section 32 (see FIG. 2).

The detection unit 31 acquires a plurality of detection results corresponding to the cumulative number of vibrations by performing numerical phase detection on the sampled AC signal.

The linear combining unit 32 linearly combines a plurality of detection results using linear coefficients according to filter characteristics.

The detection section 31 and the linear coupling section 32 can be configured basically in the same manner as in

Patent Documents

1 and 2 described above. Detailed operations of the filter unit 3 will also be described later.

(output section)
The output unit 4 is configured to output the result obtained by the linear combination in the linear combination unit 32 . The output destination of the output unit 4 is, for example, a display or a printer, but may also be some storage means or a remote device. In this embodiment, the concept of output also includes sending data to some device.

(Linear coefficient determination unit)
The linear coefficient determination unit 5 is configured to calculate the linear coefficients used in the linear combining unit 32 using the relationship between the linear coefficients, the overall pass gain in the numerical phase detection, and the target filter characteristics. there is A method for determining the linear coefficients may be the same as that in Patent Document 1 described above.

(Procedure of the filtering method of the present embodiment)
Next, a filtering method using the filtering device described above will be described with further reference to FIG.

(Steps SA-1 and SA-2 in FIG. 3)
Upon receiving an input of an AC signal to be detected, the sampling unit 1 receives the sampling period corresponding to the frequency to be extracted from the sampling rate storage unit 2, and samples the AC signal at this sampling period. Here, the sampling period is set so that one period of each frequency to be extracted is an integral multiple of the sampling period.

(Step SA-3 in FIG. 3)
Then, by performing numerical phase detection on the sampled AC signal, a plurality of detection results corresponding to the number of integrated vibrations are acquired. After that, a plurality of detection results are linearly combined using linear coefficients according to filter characteristics. For these numerical phase detection and linear combination, the methods described in Patent Document 1 or Patent Document 2 can be used, so detailed description thereof will be omitted. The linear coefficients to be used are determined by the linear coefficient determination unit 5 in the same manner as in Patent Document 1. After that, the output unit 4 outputs the obtained filtering result.

(Principle of filtering device)
Here, the principle of the filtering device according to this embodiment will be described in detail, and then specific examples will be described. Hereinafter, the citation of the formula in Patent Document 1 is expressed as A-(**), and the citation of the formula in Patent Document 2 is expressed in the form of B-(**).

(Note: A similar argument can be made for the case of n=half integer, but here only the case of natural numbers is considered for the sake of simplification.). For the sake of simplification, hereafter the angular frequency is assumed to be infinite

In general, the real part and the imaginary part of the signal component to be detected must have a finite and equal passing gain. for simplicity

If the condition is imposed, A-(15) automatically holds.

If this holds, all these noises can be removed.

holds, B-(1) can be said. Therefore, the coalition of A-(14) and A-(21)

A filter with desired characteristics is completed.

must be established. here

because

A filter is created.

in numbers. Therefore, if the frequency to be removed is selected well, it is possible to further shorten the integration time (see Patent Document 2 mentioned above). This is effective when a group of frequencies to be used can be selected in advance such as in broadband communication. The outline of the method is shown below.

From A-(8) and A-(12),

Therefore

becomes. Than this,

This means that the components are also removed at the same time (this holds true even in the case where the integration period is a set of half-integers).

and

becomes. Therefore,

means that In order to make good use of this relational expression, it is desirable that each frequency be arranged at regular intervals and be biased to the high frequency side in the possible frequency band. By doing so, the overall integration time can be further shortened compared to the method of Patent Document 1.

The theoretical discussion above treats the signal as a continuous function of time. On the other hand, since actual signal acquisition is performed discretely on both vertical and horizontal axes, it is necessary to consider the influence of deviations from the ideal situation. In particular, when time-series data is acquired with a finite sampling period Δt, it is very difficult to calculate Equations A-(7 to 10) unless the integration interval is just an integer multiple of the sampling period Δt . If it is not an integer multiple, and if the phase and amplitude of the signal are known and there is no noise, various interpolation methods can be used. Therefore, for any natural number n by the method of Patent Document 1,

Must be an integer multiple. In the case of n=half-integer sequences, "the half-cycle of the signal to be demodulated is an integral multiple of the sampling period". In the description of this embodiment, n is assumed to be an integer to avoid complication, but when n is a half-integer, one cycle can be read as a half-cycle as described above. Incidentally, in orthogonal frequency-division multiplexing (OFDM), which is based on the discrete Fourier transform (DFT), only frequencies where the first and last phases are the same within the interval of integer multiples of Δt are , this kind of problem does not occur in the first place (see FIG. 5(b), which will be described later).

When considering separation and demodulation of a plurality of frequencies, it is sufficient if each frequency can be selected so that the period of each frequency is an integral multiple of Δt. This means that the reciprocals of the frequencies are integer ratios. However, as suggested by the above-mentioned Patent Document 2, equal frequency intervals are a requirement for efficient transmission and reception of data. In this case, there arises a problem that it becomes difficult to simultaneously satisfy the period of each frequency being an integral multiple of Δt. The following example shows how to solve this problem.

(Example 1)
First, in the method of Patent Document 1, one period of the frequency to be extracted needs to be an integer multiple of Δt, but one period of the target frequency to be removed does not necessarily have to be an integer multiple of Δt. Focused on As a result, by independently sampling and demodulating the same input signal before detection with different sampling periods for each frequency to be extracted, one period is always an integral multiple of the sampling period for each frequency to be extracted. can be made to be That is, for frequencies to be extracted, both ends of the integration interval can be reserved as data points.

This point will be explained in more detail with reference to FIG. Consider the detection of two AC signals from a mixed signal (FIG. 4(c)) of two AC signals (see FIGS. 4(a) and (b)). In this case, the mixed signal is sampled (FIG. 4(d)) at a sampling period corresponding to the first AC signal (FIG. 4(a)). Before, after, or in parallel with this, sampling (FIG. 4(e)) is performed at a sampling period corresponding to the second AC signal (FIG. 4(b)). White circles in FIG. 4 indicate sampling points. Here, each sampling period is set so that one period of the frequency to be extracted is an integral multiple of the sampling period. By performing numerical phase detection using the sampled data in the filter section 3, each AC signal can be extracted as shown in FIGS. 4(f) and 4(g). In other words, demodulation is performed based on sampling data, and the original waveform can be reproduced.

(Example 2)
Next, an example in which frequencies are arranged at regular intervals will be described with reference to FIG. For this example, assume that frequencies between 11 Hz and 15 Hz are used. The light gray indicates the number of oscillation integration times required in the conventional OFDM method (Fig. 5(a)). In the OFDM method, even if the same sampling period is used, the integration interval required for detection is an integral multiple of the sampling period for all frequency components under consideration, so interpolation is not necessary in the first place (Fig. 5 (b )). That is, there is no need to adjust the sampling period.

On the other hand, the number of vibration integration times required in the signal separation method shown in

Patent Document

1 or 2 is shown in dark gray in FIG. According to this method, demodulation can be performed with data in a shorter time than the OFDM method (FIG. 5(a)). However, when data acquisition is performed at a single sampling rate, it is difficult to set the integration interval required for detection of all the frequency components under consideration to an integral multiple of the sampling period. In particular, it is difficult when the frequency components are arranged at regular intervals as shown in FIG. 5 (see FIG. 5(c)).

Therefore, by using a different sampling rate for each frequency component under consideration, the integration interval can always be an integral multiple of the sampling period (see FIG. 5(d)). Thereafter, as in the case of the first embodiment, the sampled data points can be used to extract the AC signal of the target frequency. Therefore, according to the method of this embodiment, it is possible to increase the degree of freedom in setting the frequencies to be extracted in the numerical phase detection.

(Example 3)
The concept of interlocking the sampling period with the frequency to be extracted is also effective when the frequency is swept continuously/discontinuously. An example for frequency sweeping will now be described with reference to FIG.

In the case of sweeping, the sampling rate is interlocked according to frequency changes, and the sampling unit 1 samples data so that one period of the frequency is always an integral multiple of the sampling period Δt (see FIG. 6A). ). In FIG. 6(a), the horizontal axis is the actual time. For reference, FIG. 6B shows a diagram in which the intervals between samplings are rearranged. The horizontal axis of FIG. 6(b) is the converted time. By interlocking the sampling rate according to the frequency change, data can be sampled so that one period of the frequency is exactly an integral multiple of the sampling period. Thereafter, as in the case of the first embodiment, the sampled data points can be used to extract the AC signal of the target frequency.

As described above, in the method of the present embodiment, since a separate sampling period is given for each frequency, the period of the frequency to be extracted is always an integral multiple of the sampling period. It can be an integer multiple of the oscillation period. However, as described above, it is possible to use a common sampling period for a plurality of frequencies.

In general, the usual way to increase the time resolution of signal detection is to increase the frequency used for modulation. If the frequency cannot be increased easily, the signal processing methods of

Patent Documents

1 and 2 are particularly effective. For example, according to the techniques of these documents, various frequency components are obtained using information of a shorter time width than the orthogonal frequency-division multiplexing (OFDM) currently used in the field of broadcasting and communication. can demodulate a so-called broadband signal composed of Therefore, in the field of broadcasting and communication, including Beyond 5G, it is expected to improve communication speed, fault tolerance, and interference resistance by improving signal density. Also, in the medical field, it is expected that imaging devices such as ultrasonic diagnostic equipment and MRI will increase in speed and resolution. In the sensor field, it is thought that it will contribute to improving the sensitivity, time response, and crosstalk prevention performance of devices necessary for autonomous driving technology such as electronic compasses, acceleration sensors, millimeter wave radars, and ultrasonic sonars.

In the days when the numerical processing part of broadband signal demodulation was mainly single-core, OFDM based on discrete Fourier transform (DFT) by fast Fourier transform (FFT) was an overwhelmingly efficient method of signal demodulation and separation. rice field. On the other hand, at present, when multi-core (parallel processing) is the mainstream, as in this embodiment, when the same signal is acquired and processed in parallel at different sampling periods for each frequency to be extracted, the processing between each core is can be performed independently. For example, in the second embodiment described above, by assigning each sampling rate to each core (that is, the sampling unit 1) and performing parallel processing, it is possible to shorten the time required to extract a plurality of frequencies.

In addition, there was no method for efficiently handling numerical phase detection in a short time in the frequency sweep technique often used in millimeter-wave radar, ultrasonic sonar, etc. Practical use can be achieved also in these cases by applying the method of the third embodiment.

The contents of the present invention are not limited to the above embodiments. The present invention allows various changes to be made to the specific configuration within the scope of the claims.

For example, in the above-described embodiment, the filter section 3 is composed of the detection section 31 and the linear combination section 32, but both functions can be realized by one functional block. The filter unit 3 of this embodiment includes a unit that performs processing that is mathematically equivalent to the detection and linear combination described above, and for example, it is not necessary to perform detection and linear combination separately as separate processing. In short, the filter unit 3 may be configured to filter the AC signal by performing numerical phase detection on the sampled AC signal and acquire the frequency component to be extracted.

Also, each of the components described above may exist as a functional block, and does not have to exist as independent hardware. Moreover, as an implementation method, either hardware or computer software may be used. Furthermore, one functional element in the present invention may be implemented by a set of multiple functional elements, and multiple functional elements in the present invention may be implemented by one functional element.

Furthermore, the functional elements may be arranged at physically separated positions. In this case, functional elements may be connected by a network. It is also possible to implement functions or configure functional elements by means of grid computing or cloud computing.

1 Sampling Section 2 Sampling Rate Storage Section 3 Filter Section 31 Detection Section 32 Linear Combining Section 4 Output Section 5 Linear Coefficient Determining Section

Claims

A filtering device for filtering an AC signal,
a sampling unit, a detection unit, and a linear combination unit,
The sampling unit is configured to sample the AC signal at a sampling period corresponding to each frequency to be extracted,
Here, the sampling period is set so that one period or half period of each frequency to be extracted is an integral multiple of the sampling period,
The detection unit is configured to obtain a plurality of detection results corresponding to the number of integrated vibrations by performing numerical phase detection on the sampled AC signal,
The filtering device, wherein the linear combination unit linearly combines the plurality of detection results using a linear coefficient corresponding to a filter characteristic.
The filtering device according to claim 1, wherein the sampling section is configured to perform sampling of the AC signal in parallel at sampling periods corresponding to the frequencies to be extracted.
The filtering device according to claim 1, wherein the sampling section changes the sampling period according to fluctuations in the frequency of the AC signal so as to follow the fluctuations.
The filtering device according to any one of claims 1 to 3, wherein the frequency components to be extracted from the AC signal are arranged at regular intervals.
In addition, it has an output section,
The filtering device according to any one of claims 1 to 4, wherein the output section is configured to output a result obtained by linear combination in the linear combination section.
Furthermore, it has a linear coefficient determination part,
2. The linear coefficient determination unit is configured to calculate the linear coefficient using the relationship between the linear coefficients, the overall pass gain in the numerical phase detection, and the target filter characteristics. 6. The filtering device according to any one of 1 to 5.
The filtering device according to any one of claims 1 to 6, wherein at least two of said sampling periods corresponding to said different frequencies to be extracted are common.
A filtering method for filtering an AC signal, comprising:
sampling the AC signal with a sampling period corresponding to each frequency to be extracted, wherein the sampling period is such that one period or half period of each frequency to be extracted is an integral multiple of the sampling period; is set to
obtaining a plurality of detection results corresponding to the number of integrated vibrations by performing numerical phase detection on the sampled AC signal;
and linearly combining the plurality of detection results using linear coefficients according to filter characteristics.
A filtering method for filtering an AC signal, comprising:
sampling the AC signal with a sampling period corresponding to each frequency to be extracted, wherein the sampling period is such that one period or half period of each frequency to be extracted is an integral multiple of the sampling period; is set to
Filtering the sampled AC signal by performing numerical phase detection on the AC signal to obtain frequency components to be extracted.
A computer program for causing a computer to execute each step according to claim 8 or 9.
A filtering device for filtering an AC signal,
It comprises a sampling section and a filter section,
The sampling unit is configured to sample the AC signal at a sampling period corresponding to each frequency to be extracted, wherein the sampling period is one period or half period of each frequency to be extracted. It is set to be an integral multiple of the period,
The filter unit filters the AC signal by performing numerical phase detection on the sampled AC signal, and acquires a frequency component to be extracted.