JPH10253758A - Method and apparatus for estimation of frequency, doppler sonar and tidal current meter - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a frequency estimation method in which a reliable frequency estimation value can be obtained even when an S/N ratio is small in the frequency estimation method in which a DFT(discrete Fourier transform) is used. SOLUTION: A signal which estimates a frequency is A/D-converted and then discrete-Fourier-transformed, and the frequency is estimated. Then, the estimated frequency and a spectrum number kmax at a time when an amplitude spectrum takes a maximum value are stored. This operation is executed repeatedly regarding a plurality of sampling series of the signal which estimates the signal. Data by this repetitive processing operation are stored, the histogram of the kmax is formed, and its most frequent value (mode) kmax-mode is calculated. The most frequent value refers to a kmax value in which the number of times of appearances is largest. Out of a plurality of found frequency estimated values, values which are contained in a definite frequency range which includes frequencies corresponding to the kmax-mode are sorted, and their mean value is found.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a frequency estimating method for estimating a signal frequency based on a spectrum distribution obtained by a discrete Fourier transform (DFT), and more particularly, to an accurate frequency estimating method when the S / N ratio of a signal is small. The present invention relates to a frequency estimation method and a frequency estimation device capable of estimating a frequency. The present invention also relates to a Doppler sonar and a tide meter capable of accurately measuring a ship speed and a tidal current speed even when the S / N ratio of a reflected wave is small.

[0002]

2. Description of the Related Art In a device using the Doppler effect, such as a Doppler sonar or a tidal current meter, the frequency of a received signal is set to D.
A method of estimating using FT (actual calculation is performed by fast Fourier transform (FFT)) is applied. For example, "High-precision frequency determination method using FFT" by Makoto Tabei and Mitsuhiro Ueda (IEICE Information Magazine, 1987
May 1989, pp. 798 to 805), a highly accurate frequency estimation method using DFT (hereinafter referred to as the Tabei-Ueda method) has been proposed. The Tabei / Ueda method is based on the discrete spectrum obtained by multiplying the input time-series data by a Hanning window and then performing DFT, the spectrum having the maximum value, and the amplitude spectrum at the frequency adjacent thereto. Interpolation based on the ratio of the true peak frequency fpea
k is to be estimated (see FIG. 6), and if the S / N ratio of the signal is larger than a certain level, accurate frequency estimation can be performed by this method.

[0003]

However, in the above frequency estimation method, when the S / N ratio of the signal is small, an estimated value close to the true value cannot always be obtained. For example, a sequence obtained by sampling a sine wave signal including white noise (or a product of this sequence and a sequence of an appropriate window function) is represented by g (n) (n = 0, 1, 2,..., N− Assuming 1), the DFT of g (n) is defined by the following equation.

[0004]

(Equation 1)

Here, g (n) is an amplitude spectrum | G
According to the feature of (k) |, it is classified into the following (A) and (B).

(A) is a sequence in which signal components mainly contribute to the maximum value of | G (k) | (hereinafter referred to as A sequence). An example of | G (k) | of this series is shown in FIG.
The frequency number k when | G (k) | takes the maximum value is km
Assuming that a is ax, kmax of the A series takes a value corresponding to the frequency f of the signal, and the relationship between them is expressed by the following equation. f ≒ kmax (fs / N) where fs is the sampling frequency and N is the number of sampling points.

[0007] On the other hand, (B) is a sequence in which the signal component does not contribute to the maximum value of | G (k) | FIG. 5B shows an example of | G (k) | of this series.
The value kmax of the B sequence takes a value irrelevant to the frequency f of the signal. That is, the spectrum of the signal is buried in the spectrum of the noise, and the noise component has a peak.

Since the probability of occurrence of the B sequence increases as the S / N ratio decreases, the conventional frequency estimation method using DFT
When the S / N ratio is small, there is a problem that reliability is lacking.

An object of the present invention is to provide a frequency estimating method and a frequency estimating apparatus which can obtain a reliable frequency estimated value even when the S / N ratio is small in a frequency estimating method using DFT. SUMMARY OF THE INVENTION It is an object of the present invention to provide a Doppler sonar and a tidal current meter capable of performing accurate measurement.

[0010]

According to a first aspect of the present invention, a signal that has been subjected to analog-to-digital conversion (A / D conversion) is decomposed into a discrete spectrum by DFT, and the signal is divided based on the spectrum distribution. In the method for obtaining a frequency estimation value, a plurality of sampling sequences of the signal are input, the frequency estimation value is stored for each sampling sequence, and a frequency number at which the amplitude spectrum of each sampling sequence has a maximum value is obtained. The frequency is counted, and only the frequency estimation values included in a certain frequency range including the frequency corresponding to the frequency number at which the frequency has the maximum value are selected, and the average value of the selected frequency estimation values is determined as the definite frequency estimation. It is characterized by a value.

The invention according to claim 2 of the present application decomposes an A / D converted signal into a discrete spectrum by DFT,
In a device for obtaining a frequency estimation value of the signal based on the spectrum distribution, a storage means for inputting a plurality of sampling sequences of the signal and storing the frequency estimation value for each sampling sequence; Count the frequency of the frequency number when the spectrum has the maximum value, and select only the frequency estimation value included in a certain frequency range including the frequency corresponding to the frequency number at which the frequency has the maximum value. Frequency determination means for setting an average value of the frequency estimation values to the determined frequency estimation value.

According to a third aspect of the present invention, the ultrasonic wave signal output from the transmitter is received by a receiver from a reflected wave of the target object, and the target object is detected based on the Doppler shift frequency of the reflected wave. A Doppler sonar for measuring a relative speed of a ship with respect to a digital signal; an analog-digital conversion means for converting the reflected wave into a digital signal; a spectrum decomposition means for decomposing the digital signal into a discrete spectrum by performing a discrete Fourier transform; Frequency estimation means for estimating the Doppler shift frequency of the reflected wave based on the spectrum, maximum amplitude number detection means for detecting the maximum amplitude number which is the frequency number when the absolute value of the discrete spectrum takes the maximum value, For a plurality of reflected waves obtained by transmitting a plurality of times, the analog / digital conversion means Vector decomposition means, frequency estimation means,
And executing maximum amplitude number detection means, and selects only frequency estimation values included in a certain frequency range including a frequency corresponding to a frequency number at which the frequency of the maximum amplitude number has a maximum value. Frequency determining means for setting an average value of the estimated values as the determined frequency estimated value.

According to a fourth aspect of the present invention, there is provided an ultrasonic signal output from a transmitter, wherein a reflected wave from the bottom of the water and a body of water is received by a receiver, and a ground based on the Doppler shift frequency of each reflected wave. In a tide meter for determining a velocity and a velocity with respect to water and subtracting these, a tide analog-digital conversion means for converting a reflected wave from the bottom to a digital signal, and a discrete Fourier Underwater reflected wave spectrum decomposing means for decomposing into a discrete spectrum by converting, underwater reflected wave frequency estimating means for estimating the frequency of the underwater reflected wave based on the discrete spectrum, and the absolute value of the discrete spectrum is the maximum value A bottom-bottom reflected wave maximum amplitude number detecting means for detecting a bottom-bottom reflected wave maximum amplitude number which is a frequency number when taking For the plurality of bottom reflection waves obtained by the transmission of the above, the bottom reflection wave analog-to-digital conversion means, bottom reflection wave spectrum decomposition means, bottom reflection frequency estimation means, and bottom reflection maximum amplitude number detection means are executed. Then, the frequency of the maximum amplitude number is selected only underwater reflected wave frequency estimation value included in a certain frequency range including the frequency corresponding to the frequency number having the maximum value, of the selected underwater reflected wave frequency estimation value Underwater reflected wave frequency determining means for determining the average value as the determined underwater reflected wave frequency estimated value, water body reflected wave analog-to-digital conversion means for converting the reflected wave from the water body into a digital signal, and a discrete Fourier A water mass reflection wave spectrum decomposing means for decomposing the water mass into a discrete spectrum by converting the water mass based on the discrete spectrum; Water mass reflection wave frequency estimating means for estimating the frequency of the radiation wave, and water mass reflection wave maximum amplitude for detecting a water mass reflection wave maximum amplitude number which is a frequency number when the absolute value of the discrete spectrum takes the maximum value Number detection means, and for a plurality of water body reflected waves obtained by a plurality of transmissions, the water body reflected wave analog / digital conversion means, water body reflected wave spectrum decomposing means, water body reflected wave frequency estimating means, and Executing the water mass reflection wave maximum amplitude number detecting means and selecting only the water mass reflection wave frequency estimation value included in a certain frequency range including the frequency corresponding to the frequency number at which the frequency of the maximum amplitude number has the maximum value. And a water body reflected wave frequency determining means for setting an average value of the selected water body reflected wave frequency estimated values as a confirmed water body reflected wave frequency estimated value.

According to the present invention, the signal for estimating the frequency is represented by A
/ D conversion → DFT conversion to estimate the frequency. Then, the estimated frequency and the amplitude spectrum | G (k)
The frequency number kmax when | takes the maximum value is stored. This is repeated for a plurality of sampling sequences of the signal for estimating the frequency. Here, each sampling sequence is time-series data including a plurality of sampling values. After accumulating data by this repetitive processing, a histogram of kmax is obtained and its mode (mode) kmax-mode is determined. The mode is the value of kmax that has the highest number of appearances. A frequency corresponding to this frequency number kmax-mode, fo≡kmax-mode (fs / N), fs: sampling frequency N: number of sampling points, and a certain value including the frequency fo among the plurality of obtained frequency estimation values Those included in the frequency range are selected and the average value is obtained. This is adopted as the final frequency estimation value.

As described above, the frequency estimation value having a value close to the frequency fo is selected and averaged to obtain the final frequency estimation value. Thus, even if the S / N ratio of the signal is small, the signal estimation value is small. Since only the estimated value when the spectrum level is superior to the noise is selected and the estimated value at other times is ignored, an accurate frequency estimated value can be obtained.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described with reference to the drawings. FIG. 1 is a view showing a ship equipped with a tide meter to which the present invention is applied, and FIG. 2 is a block diagram of the tide meter.
At the bottom of the ship 1, a transducer 1 as three ultrasonic transducers
1 is provided, emits an ultrasonic beam 2 in three directions, and receives reflected waves of the ultrasonic beam 2 from the sea floor and a body of water. Three transducers 11a, 11b, 11
The transmission drive unit 19 is connected to c through traps 12a, 12b, and 12c, respectively. The transmission driver repeatedly supplies an electric pulse signal to the transducer 11. Each of the traps 12a to 12c has a frequency converter 13a to 13c and an amplifier 14a to 14c as a receiving system.
And A / D converters 15a to 15c are connected.
The trap 12 prevents a high-level signal input from the transmission drive unit 19 from sneaking into a reception system, and
This is a circuit for transmitting a low-level reflection signal received by the transducer 11 to a receiving system. A frequency converter 13 connected to the trap 12 converts the frequency of the received signal to a lower value. By lowering the frequency, the subsequent A / D converter 1
5 can be reduced. In addition,
The Doppler shift is preserved even if the frequency is changed. The frequency-converted signal is amplified by the amplifier 14 and then A / D
The data is converted by the converter 15 into digital data (quantized discrete-time signal). This digital data is stored in arithmetic unit 1
6 is input.

The signal output from the frequency converter 13 is input to a seafloor detector 20. Undersea detector 20
Calculates the seabed depth based on the time difference between the reflected waveforms received by the transducers 11a to 11c.

The arithmetic unit 16 comprises a waveform corresponding to the seafloor depth of a signal (frequency converted and converted into digital data) received by the transducer 11 and a portion corresponding to one or a plurality of tidal current measurement depths. Is extracted, and the frequency is estimated by performing DFT on the extracted waveform. The frequency estimation of the reflected wave at each measurement depth is performed separately for each reception signal of each of the transducers 11a to 11c. Then, based on the estimated frequencies at the respective measured depths, a ground speed, which is a speed vector of the ship with respect to the sea floor, and a water speed, which is a speed vector of a ship with a predetermined depth of water, are obtained. Then, by subtracting the water speed from the ground speed, a tidal current speed, which is a speed vector of the water mass with respect to the seabed, is calculated. The calculated ground speed and tidal current speed are input to the display unit 17. The display unit 17 converts the ground speed and the tidal current speed into a predetermined display format (for example, an arrow vector display or a numerical value / azimuth display), develops the pattern on the screen, and outputs the pattern to the monitor 18. By displaying this pattern on the monitor 18, the user can recognize the ground speed and the tidal current speed.

With reference to FIGS. 3 to 5, the operation of the calculating section 16 for estimating the frequency of the received reflected wave signal will be described. FIG. 3 is a flowchart showing the frequency estimation processing operation, FIG. 4 is a view showing a processing result accumulation table set in the arithmetic unit 16, and FIG. 5 is a view showing a histogram of an amplitude spectrum and kmax obtained by DFT.

The processing operation of FIG. 3 is an operation of repeatedly transmitting and receiving ultrasonic pulses to process a sampling sequence of each received signal. For each of the transducers 11a to 11c, the depth of the seabed and 1 or The processing operation is repeatedly performed for each waveform of a portion corresponding to a plurality of measurement depths. First,
The counter m indicating the number of times of data collection, that is, the number of times of transmission / reception of transmitted / received ultrasonic pulses is set to 1 (s10).
The sampling data of the waveform section corresponding to the measurement depth or the seabed depth that is the tidal current measurement target is N samples (x
(0) to x (N-1)) read (s11), and multiply this sampling data sequence by an appropriate window function w (n) such as a Hanning window as shown in the following equation (s12).

G (n) = w (n) .x (n) (n =
0, 1,..., N−1) This data string {g (n)} is expressed by D in accordance with [Equation 1].
FT conversion is performed (s13). The FFT algorithm may be used for this DFT transformation.

A frequency estimation value f (m) is calculated using the value of G (k) subjected to the DFT transformation (s14). There are several methods for this frequency estimation. For example, the above-mentioned Tabei-Ueda method may be used. This frequency estimation value is recorded in a processing result accumulation table (see FIG. 4) set in the calculation unit 16. And the amplitude spectrum | G
(K) Find the frequency number k when | takes the maximum value, and
max (m) (s15). This kmax (m) is also recorded in the processing result accumulation table. At this time, if the S / N ratio of the received signal is large, the spectrum is as shown in FIG.
And kmax takes a value corresponding to the signal frequency, but if the S / N ratio is small, the spectrum becomes as shown in FIG.
, And kmax may take a value independent of the signal frequency. The above data collection processing is repeated for M received signals (s16 → s11), and then s
The process proceeds to the counting operation of No. 18.

At s18, histograms of M kmax (m) (m = 1 to M) stored in the processing result accumulation table are obtained (see FIG. 5C), and the mode value kmax-m is calculated.
Find the mode. Then, f corresponding to this mode value
o = kmax−mode · (fs / N) is calculated, and the M frequency estimated values f (m) calculated in the operation of s14 are calculated.
From (m = 1 to M), only values included in a certain frequency range including the frequency fo are selected (s19). Then, the average value of the selected frequency estimation values is calculated, and the average value is output as the frequency of the received signal (s2
0).

Since the ultrasonic beam actually transmitted from the transmitter / receiver 11 has a certain spread angle, for example, the incident angle with respect to the sea floor is not constant, and the Doppler shift received by the reflected wave at the sea floor has a certain width. . By selecting the estimated values within the frequency range corresponding to the beam divergence angle, effective estimated values can be used without omission.

As described above, by using the method of the present invention, even when the S / N ratio is small enough to make the spectrum of the received signal invisible and obscured by the spectrum of the noise, effective ones are selected from a plurality of estimated values. By doing so, an accurate frequency can be estimated. By applying this to a tidal current meter or a Doppler sonar, an accurate Doppler shift frequency can be determined, and accurate speed detection can be performed.

In the above embodiment, one ultrasonic transducer is used as the transmitter and the receiver, but the transmitter and the receiver may be provided separately.

Although the above-described embodiment has been described with respect to an example in which the present invention is applied to a tidal current meter or a Doppler sonar mounted on a ship, the present invention is not limited to this field, and various types of processing including frequency estimation processing may be performed. It can be applied to the measurement field.

[0028]

As described above, according to the present invention, since an estimated value that is significantly different from a true value, which can be caused by the influence of noise, can be ignored, an accurate frequency estimated value can be obtained even if the S / N ratio of the signal is small. Can be obtained.

[Brief description of the drawings]

FIG. 1 is a view showing a ship equipped with a tidal current meter to which the present invention is applied;

FIG. 2 is a block diagram of the tidal current meter.

FIG. 3 is a flowchart showing an operation of a calculation unit of the tidal current meter;

FIG. 4 is a diagram showing a processing result accumulation table set in the calculation unit;

FIG. 5 is a diagram showing an example of a spectrum and a histogram calculated by the calculation unit.

FIG. 6 is a diagram illustrating a general frequency estimation method based on DFT.

[Explanation of symbols]

 11a to 11c: transducer, 16: arithmetic unit

Claims (4)

[Claims] 1. A method for decomposing an analog-to-digital converted signal into discrete spectrums by a discrete Fourier transform and obtaining a frequency estimation value of the signal based on the spectrum distribution, wherein a plurality of sampling sequences of the signal are input. In addition to storing the frequency estimation value for each sampling sequence, counting the frequency of the frequency number when the amplitude spectrum of each sampling sequence has the maximum value, the frequency corresponding to the frequency number at which the frequency has the maximum value A frequency estimation method characterized by selecting only frequency estimation values included in a certain frequency range including the above, and using an average value of the selected frequency estimation values as a definite frequency estimation value. 2. An apparatus for decomposing an analog-digital converted signal into a discrete spectrum by a discrete Fourier transform and obtaining a frequency estimation value of the signal based on the spectrum distribution, wherein a plurality of sampling sequences of the signal are input. Storing means for storing the frequency estimation value for each sampling sequence, counting the frequency of the frequency number when the amplitude spectrum of each sampling sequence has the maximum value, and corresponding to the frequency number at which the frequency has the maximum value Frequency selection means for selecting only frequency estimation values included in a certain frequency range including the frequency to be set, and using an average value of the selected frequency estimation values as a definite frequency estimation value. Estimation device. 3. A reflected wave from an object of an ultrasonic signal output from a transmitter is received by a receiver, and a relative speed of a ship with respect to the object is measured based on a Doppler shift frequency of the reflected wave. In Doppler sonar, analog-to-digital conversion means for converting the reflected wave into a digital signal; spectrum decomposition means for decomposing the digital signal into a discrete spectrum by performing a discrete Fourier transform; and Frequency estimation means for estimating the Doppler shift frequency, maximum amplitude number detection means for detecting the maximum amplitude number that is the frequency number when the absolute value of the discrete spectrum takes the maximum value, obtained by multiple transmissions For multiple reflected waves,
The analog-to-digital conversion means, the spectrum decomposition means, the frequency estimating means, and the maximum amplitude number detecting means are executed, and the frequency of the maximum amplitude number becomes a certain frequency range including a frequency corresponding to a frequency number having a maximum value. Doppler sonar comprising: frequency determination means for selecting only included frequency estimation values and using an average value of the selected frequency estimation values as a determined frequency estimation value. 4. A receiver receives reflected waves from the water bottom and water mass of an ultrasonic signal output from a transmitter, and calculates a ground speed and a water speed based on a Doppler shift frequency of each reflected wave, A tide meter for calculating a tidal current velocity by subtracting these; a bottom-of-water reflected-wave analog-to-digital conversion means for converting a reflected wave from the bottom to a digital signal; and a discrete Fourier transform of the digital signal to separate a discrete spectrum. Underwater reflected wave spectrum decomposing means, underwater reflected wave frequency estimating means for estimating the frequency of the underwater reflected wave based on the discrete spectrum, and a frequency number when the absolute value of the discrete spectrum takes the maximum value. Underwater reflected wave maximum amplitude number detecting means for detecting underwater reflected wave maximum amplitude number, obtained by multiple transmissions Several of underwater reflected waves, the water bottom reflection wave analog-to-digital conversion means,
Underwater reflected wave spectrum decomposing means, underwater reflected wave frequency estimating means, and executing underwater reflected wave maximum amplitude number detecting means, a constant frequency including a frequency corresponding to a frequency number at which the frequency of the maximum amplitude number takes a maximum value Bottom reflected wave frequency estimating means for selecting only the bottom water reflected wave frequency estimation value included in the range, and determining the average value of the selected bottom water reflected wave frequency estimation values as the fixed bottom water reflected wave frequency estimation value, Water mass reflection wave analog-to-digital conversion means for converting the reflected wave of the digital signal into a digital signal; water mass reflection wave spectrum decomposing means for decomposing the digital signal into a discrete spectrum by performing a discrete Fourier transform; A water body reflected wave frequency estimating means for estimating the frequency of the water body reflected wave; A water body reflection wave maximum amplitude number detecting means for detecting a water body reflection wave maximum amplitude number that is a frequency number when taking a value, and a plurality of water body reflection waves obtained by transmitting a plurality of times, the water body Reflected wave analog / digital conversion means,
The water mass reflected wave spectrum decomposing means, the water mass reflected wave frequency estimating means, and the water mass reflected wave maximum amplitude number detecting means are executed, and the frequency of the maximum amplitude number includes a frequency corresponding to a frequency number having a maximum value. Only the estimated value of the reflected wave frequency of the water included in the certain frequency range is selected, and the average value of the estimated values of the reflected wave frequencies of the selected water mass is determined. Means, and a tide meter.