JPH08339200A - Frequency analysis method for received arrival signal - Google Patents

Abstract

PURPOSE: To obtain a noise reducing effect larger than the noise reducing effect brought by a conventional phasing processing which is insufficient. CONSTITUTION: Externally arriving audio signals are received by a receiver group 50 consisting of multiple receivers, a cross-frequency spectrum among the individual signals is calculated (63) by using a linear prediction analysis applying a linear prediction model to the time series among the multiple received signals (61, 62), and in the case of calculating a frequency spectrum by using the cross-frequency spectrum, the cross-frequency spectrum between an optionally selected receiver and the other receivers is compensated to be phased to a desired azimuth before being added (64A) in order to calculate the frequency spectrum (65).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention extracts and displays a frequency spectrum of a signal in a desired azimuth by using signals received by a plurality of receivers as in the fields of voice recognition and sonar systems. The present invention relates to a frequency analysis method for an incoming received signal.

[0002]

2. Description of the Related Art Conventionally, such a technique has been used in the field of sonar systems and the like. The sonar device is for performing search, position measurement, classification, etc. of a target moving on or in water using underwater sound. This device is divided into a passive sonar device and an active sonar device according to the operation mode. The passive sonar device uses sound waves radiated by a target, and the active sonar device uses a reflected wave (echo) of a sound wave emitted from the device side from the target to perform target search, position measurement, and classification. .

The signal detection processing used in the sonar apparatus detects the temporal characteristics (waveform, spectrum, etc.) of the signal and the spatial characteristics (position, shape, moving speed) of the signal. The spatial processing used to Temporal processing includes matched filters, Wiener filters and spectral estimation. Spectral estimation estimates the intensity of a signal with respect to frequency and is used to detect a periodic signal (line spectrum) buried in noise and to classify targets that emit that signal. On the other hand, spatial processing includes beam forming (hereinafter referred to as BF) and delay / phase estimation. The BF uses the directionality of waves propagating in space to phase the output signals of a plurality of wave receivers forming a wave receiver group according to the azimuth, thereby improving the S / N of the signal and the signal. Estimate the incident direction and intensity (spatial spectrum) of. Delay / phase estimation is a problem of estimating a delay or a phase difference occurring between signals received by a small number of receivers, and is mainly used for measuring a position of a target.

Among the passive sonar device and the active sonar device, for example, the passive sonar device is mainly used for detecting the running sound and transient sound of a ship. FIG. 5 is a configuration diagram of a conventional passive sonar device. The passive sonar device shown in the figure includes a receiver group 10 and a sonar receiver 2.
0, an azimuth indicator 30, and a frequency indicator 40. In the passive sonar apparatus of FIG. 5, a receiver group 10 including a plurality of receivers that receive the sound waves emitted by the ship 1.
The sonar receiver 20 is connected to the output side of the sonar receiver. In the sonar receiver 20, the received signals from the receiver group 10 are subjected to phasing processing, a directional beam is formed on the entire circumference to receive the underwater sound from the ship 1, and the bearing of the ship 1 around the reception point, The characteristic (for example, frequency) of the signal of the ship 1 is detected, and the azimuth indicator 30 and the frequency indicator 40 are displayed.
The frequency is displayed by.

In the reception processing of the sonar receiver 20,
In order to detect the characteristics of the signal of the ship 1, for example, frequency analysis in the time domain by the fast Fourier transform method (FFT method) or high resolution frequency analysis capable of detecting a fine frequency structure by increasing the number of data is performed. Then, the frequency is displayed on the two-dimensional plane of frequency-time on the frequency display 40 at a concentration according to the intensity of the spectrum.
Further, the sonar receiver 20 has a function of performing ultra-narrow band analysis capable of detecting a fine frequency structure by narrowing the analysis frequency.

[0006]

However, in the above-mentioned conventional frequency analysis method for incoming signals, the received signals are subjected to phasing processing for each azimuth in advance before frequency analysis. A single process is responsible for realizing two functions or effects. One of the functions and effects of this phasing processing is a direction limiting function of extracting a signal from a specific direction, and the other is a noise reduction effect for detecting a low SN signal. This is because in the conventional method, the noise incident from the entire circumference is reduced by limiting the azimuth, and the azimuth limitation is also noise reduction. However, in order to detect a signal with a weaker intensity (lower SN ratio) than the weak signal that can be detected by the noise reduction that is obtained by the phasing process, further noise reduction is required. However, the conventional method, which requires the reduction of noise and expects only the effect brought by the phasing process, is not sufficient. The point that this noise reduction effect is insufficient is the first problem in the conventional method.

The noise reduction effect of the conventional method is
I expect the effect brought by phasing,
The maximum value of this noise reduction effect depends on the number of receivers used per se, and when the number of receivers is M, the noise reduction effect is 10 log (M). However, in order to detect a weaker signal than the conventional one, a larger noise reduction effect is expected. Especially when the number of usable receivers is limited, the noise reduction effect obtained by the phasing process is also limited. The second problem of the conventional method is that only a small noise reduction effect can be obtained when the number of receivers is small. Therefore, there is a demand for a signal detection processing method for an incoming received signal which solves the above-mentioned problems caused by the conventional method of reducing noise only by phasing processing and has a higher noise reduction effect even with a limited number of receivers. It was

[0008]

SUMMARY OF THE INVENTION A frequency analysis method for an incoming received signal according to the present invention is such that an acoustic signal coming from the outside is received by each of a plurality of receivers, and the received plurality of received signals are time-series. A cross frequency spectrum between each received signal is calculated by using a linear prediction analysis that applies a linear prediction model to a frequency spectrum is calculated using the cross frequency spectrum, and an intensity of the frequency spectrum of an incoming received signal is displayed. In the frequency analysis method, when the frequency spectrum is calculated using the cross frequency spectrum, the cross frequency spectrum between one receiver arbitrarily selected and another receiver is aligned in phase with a desired direction. Is subjected to phase compensation and then added to calculate a frequency spectrum.

[0009]

According to the present invention, the acoustic signals coming from the outside are respectively received by the plurality of receivers, and the linear prediction model is applied during the time series of the received plurality of received signals. In a frequency analysis method for an incoming received signal in which a cross frequency spectrum between received signals is calculated, a frequency spectrum is calculated using the cross frequency spectrum, and the frequency spectrum intensity is image-displayed, the frequency spectrum is calculated using the cross frequency spectrum. When calculating, the frequency spectrum is calculated by adding the cross frequency spectrum of one receiver selected arbitrarily and the other receiver after phase compensation so that the phase is aligned with the desired direction. I decided to do it. Then, as described above, the cross frequency spectrum calculation process using the multivariate predictive analysis method that reduces noise without performing the phasing process and the phasing process that limits the azimuth using this cross frequency spectrum are performed separately. Since it is a signal processing method, in extracting the frequency characteristic of the sound source signal coming from a specific direction,
The frequency of the incoming received signal that can detect a signal with a smaller signal strength can be added by the cross frequency spectrum calculation process provided in the preceding stage of the phasing process that has the azimuth limitation function and the noise reduction function. An analytical method can be provided.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First, the outline of signal processing in the present invention will be described in comparison with the problems of the conventional method. The above-mentioned first problem of the conventional method arises from the fact that the single process of phasing fulfills the two purposes of azimuth limitation and noise reduction. Therefore, in the present invention, in order to enhance the noise reduction effect,
It is considered that the azimuth limiting function and the noise reducing function are realized by different processes by adding a process having no azimuth limiting function but a noise reducing function. Therefore, it is considered that the noise reduction function is preceded by the direction limiting function, the noise is reduced from the received signal by the noise reduction function, and the noise is further reduced by limiting the direction of the signal. This noise reduction function is realized by using multivariate linear prediction analysis. Here, the multivariate prediction is M
Let _m (n) (m = 1, 2, ..., M) be the m-th type n-th time-series signal of these different types of signals, and not only on the time-series of each of these signals. A linear prediction model is applied between other signals in the time series and each m-th type signal is expressed by a linear combination with M types of past K signals.

That is, for x _m (n) (m = 1, 2, ..., M), the multivariate vector X (n) = (x ₁ (n), x ₂ (n),
, X _M (n)) ^t (t represents transposition), and a linear prediction model represented by the following equation (1) is applied to this multivariate vector X (n).

[0012]

[Equation 1]

However, K in the equation (1) is a constant called a prediction order, A (k) is a k-th prediction matrix of M rows and M columns, and U (n) is driving of each kind of signal. It is a signal vector of M rows and 1 column consisting of signals called noise. Where M
As the types of signals, M signals received from M receivers are applied. Now, assuming that the signal comes from a sound source in a certain direction, each received signal is the same signal except for the delay (or phase) according to the relative position of the receiver. Therefore, if the cross frequency spectrum between the signals (which represents the correlation of the frequency components between the signals) is calculated using the prediction coefficients for these signals,
The cross-frequency spectrum between any of the signals shows the same power, differing only in phase.

On the other hand, when the signals received by the receiver are independent noises both in time series and between the mutual signals, there are no significant prediction coefficients because they are unrelated to each other.
Even if the cross frequency spectrum between each signal is calculated using this prediction coefficient, the degree of influence on each other decreases as M increases. This is because the cross frequency spectrum between each signal represents the signal as a linear combination sum with other signals, so if the random signal power is expressed as the sum of the signals independent of it, the contribution of each power The ratio becomes smaller as M becomes larger, and the value of the cross frequency spectrum has a random phase and shows a value smaller than the received power. Therefore, the cross frequency spectrum value using the multivariate predictive analysis method does not change in the power value of the sound source signal, and can be made smaller as the M increases with respect to noise.

The phase of the cross frequency spectrum with respect to the sound source signal indicates how the phase peculiar to the relative arrangement of the sound source signal and the receiver advances. Therefore, if the phase corresponding to the azimuth to be limited is corrected and added, it is adjusted. Phase processing can be performed, which further reduces the magnitude of the noise component cross frequency spectrum because the noise cross frequency spectrum is random. Therefore, in the first embodiment of the present invention, a cross frequency spectrum calculation means using a multivariate predictive analysis method for reducing noise without performing a phase phasing process, and a phasing for limiting an azimuth using this cross frequency spectrum. The processing means is provided separately.

The cause of the above-mentioned second problem of the conventional method is that the number of signals in the phasing process is the same as the number of receivers because the signal to be processed in the phasing is the received signal itself. This is because The phasing process is a kind of averaging process, and basically, by adding the processes to each of a plurality of signals and adding them, the values of the components having the common features from the plurality of signals have the common features. If there is a common feature to be extracted in the required signal that is not in the unnecessary signal, the enhancement effect will be higher as the number of signals having the same feature is higher.
Therefore, in order to enhance the effect of the phasing process, the number of signals having the same characteristics may be increased. As a means for doing so, in order to increase the number of signals that can be the subject of phasing processing before performing phasing processing, multivariate linear prediction analysis is performed on the received signal,
It is considered that the cross frequency spectrum matrix between received signals is calculated using the analysis result and the elements of this matrix are subjected to the phasing process.

Cross-frequency spectrum between signals by using prediction coefficients for M received signals received from M receivers (one showing the correlation of frequency components between signals)
The calculation method is as described in the first problem. The cross frequency spectrum between any signals shows the same power with only different phases, and the cross frequency spectrum between received signals is arranged in a matrix. The frequency spectrum matrix is composed of components having a fixed phase relationship. On the other hand, if the signal received by the receiver is independent noise both in time series and between mutual signals, there is no meaningful prediction coefficient because it is a signal that is not related to each other in the first place. As described above, even if the cross frequency spectrum of is calculated, the degree of influence on each other decreases as M increases.

This is because the cross frequency spectrum between each signal represents the signal as a linear combination sum with other signals, so if the random signal power is represented by the sum of the independent signals, The power contribution becomes smaller as M is increased, and the value of the cross frequency spectrum is
It shows a value smaller than the received power. Further, those phases do not have a fixed phase relationship like the sound source signal, and the cross frequency spectrum matrix becomes a component showing different phases. However, since the cross-frequency spectrum matrix shows the complex conjugate relationship of the symmetric components among the non-diagonal components, the only elements that have unique information are the non-diagonal components on one side of the non-diagonal components. is there. The total number of non-diagonal components is M (M-1) / 2.

That is, the M (M-1) / 2 components have a common phase relationship determined by the receiver arrangement and the sound source arrival direction for the sound source signal, and have a common phase for the noise. It has no relation and shows a random phase.
Therefore, if phasing is performed using these components, the number of signals that can be subjected to phasing processing can be increased, and M (M-1) / 2 signals for M receivers can be used. A phasing process can be performed. Therefore, in the second embodiment of the present invention, before the phasing process, one of the cross frequency spectrum matrix calculating means using the multivariate predictive analysis method and the non-diagonal component of the cross frequency spectrum matrix is provided. And a phasing processing means using a non-diagonal component on the side.

Example 1. 1 is a block diagram for explaining a frequency analysis method for an incoming received signal according to a first embodiment of the present invention, in which 50 is a receiver group, 60A is a frequency analysis unit, and 7A is a frequency analysis unit.
Reference numeral 0 is a display unit. The frequency analysis unit 60A includes a cross-correlation matrix calculation unit 61, a prediction coefficient matrix calculation unit 62, a cross frequency spectrum matrix calculation unit 63, a phasing processing unit 64A, and a frequency spectrum calculation unit 65. In FIG. 1, a receiver group 50 is composed of M receivers that convert acoustic signals from a ship or the like into electrical signals, and a frequency analysis unit 60A is connected to the output side thereof. The frequency analysis unit 60A performs frequency spectrum analysis using the multivariate linear prediction analysis method from the M received signals input from the wave receiver group 50, and outputs the frequency spectrum to the display unit 70. Display 7
At 0, the frequency spectrum of each frequency is displayed according to its strength and, if necessary, its time passage is displayed.

In the frequency analysis unit 60A, first, in the cross-correlation matrix calculation unit 61, a cross-correlation matrix composed of cross-correlation coefficients between M received signals input from the receiver group 50 is calculated, and the prediction coefficient matrix is calculated. It is output to the calculation unit 62. The prediction coefficient matrix calculation unit 62 calculates the prediction coefficient matrix from the input cross-correlation matrix using the multivariate linear prediction analysis method, and outputs the prediction coefficient matrix to the cross frequency spectrum calculation unit 63. The cross frequency spectrum calculation unit 63 calculates a cross frequency spectrum matrix for each analysis frequency using the prediction coefficient matrix, and inputs the cross frequency spectrum matrix to the phasing processing unit 64A. Phase adjusting unit 64A
Then, using the non-diagonal components of the cross frequency spectrum matrix, the phase corresponding to the desired azimuth is added to calculate the phased cross frequency spectrum, and this is calculated.
Enter in 5. The frequency spectrum calculation unit 65 calculates a frequency spectrum from the phasing cross frequency spectrum for each analysis frequency and outputs it to the display unit 70.

The operation of the first embodiment shown in FIG. 1 will be described in detail.
The frequency analysis unit 60A converts the sample time series of the M received signals input from the receiver group 50 into a fixed number N of samples.
It is divided into analysis frames consisting of, and the following processing is performed for each analysis frame. In addition, s (n, m) (n = 1,2) using the sample time series number n in the analysis frame in which the sample time series signal of the m-th received signal input from the receiver group 50 is to be analyzed. , ..., N, m = 1, 2, ..., M)
It is written as. The cross-correlation matrix calculation unit 61 in the frequency analysis unit 60A uses the sample time series r (n, m) to generate K M
Cross-correlation matrix R (k) (k = 0, 1,
, K-1) is calculated by the following equation (2).

[0023]

[Equation 2]

Here, K is the predicted order (constant), and is the kth
The element r in the i-th row and the j-th column of the th cross-correlation matrix R (k)
_{i, j} (k) is calculated based on the following equation (3).

[0025]

(Equation 3)

In the prediction coefficient matrix calculating section 62, the cross correlation matrix R (k) (k =
0, 1, ..., K−1) is used to solve the K equations of the following equation (4), and the prediction coefficient matrix A (k) (k = 0, , ..., K-1) are calculated.

[0027]

[Equation 4]

The above equation may be solved by using, for example, a multivariately extended Levinson algorithm. In the cross frequency spectrum matrix calculation unit 63, first, the prediction coefficient matrix A (k) (k = 0,
, ..., K−1), the inverse matrix H (f) of the frequency response, which is a complex matrix of M rows and M columns with respect to the analysis frequency f, is calculated based on the following equation (5).

[0029]

(Equation 5)

Here, h _ij (f) in the equation (5) is calculated based on the following equation (6). However, q in the equation (6) is an imaginary unit, a
_{i, j} (k) is a component in the i-th row and j-th column of A (k), and C _{i, j} is a coefficient that satisfies the following expression (7).

[0031]

(Equation 6)

[0032]

(Equation 7)

Next, using H (f) of the above equation (5), a cross frequency spectrum matrix P _C (f) which is a complex matrix of M rows and M columns for the analysis frequency f is calculated based on the following equation (8). The calculated value is output to the phasing processing unit 64A.

[0034]

(Equation 8)

However, * in the equation (8) represents a conjugate transposition, and W
Is given by the following equation (9).

[0036]

[Equation 9]

In the phasing processor 64A, out of the components of the cross frequency spectrum matrix P _C (f), a non-diagonal element that is a cross frequency spectrum component, that is, a component p _{ci, j for} which i ≠ _j
calculating a phased cross frequency spectrum p _c (f) is added to align the reference phase using the phase compensation determined according phases for stabilizing partner position θ of (f). If the receivers are arranged linearly at equal intervals d and the direction of the sound source signal is θ (the direction of the arrangement is 0 °), the reference phase is the phase of the first receiver signal. , P _c (f) is calculated by the following equation (10).

[0038]

[Equation 10]

However, b _j (θ) in the equation (10) is the phase compensation for the received signal at the j-th receiver, and the following equations (11), (12), (13) are used, where q is an imaginary unit. Given in.

[0040]

[Equation 11]

[0041]

(Equation 12)

[0042]

(Equation 13)

The frequency spectrum calculating unit 65 calculates the frequency spectrum P (f) based on the following equation (14) using _pc (f) of the equation (10) and outputs it to the display 70.

[0044]

[Equation 14]

The display unit 70 displays the frequency spectrum S (f) for each analysis frame according to its strength.

According to the first embodiment, since the frequency analysis method for the incoming received signal has the structure in which the cross frequency spectrum calculating means and the phasing processing means are separately provided as described above, the sound source signal coming from a specific direction is obtained. In the extraction of the frequency characteristic of, the noise reduction effect can be further added by the cross frequency spectrum calculation means provided in the preceding stage of the phasing processing means having the azimuth limitation function and the noise reduction function, and a signal with a smaller signal strength can be obtained. It is possible to provide a frequency analysis method for the incoming signal that can be detected. Further, since it is based on the linear predictive analysis, it has a better analysis capability in a shorter time than the frequency analysis converted into Fourier, and it becomes possible to accurately extract the transient sound of a ship or the like and the sound source signal in the narrow band analysis.

Example 2. FIG. 2 is an explanatory diagram of a frequency analysis method of an incoming received signal according to the second embodiment of the present invention. The frequency analysis unit 60B of FIG. 2 is a matrix instead of the phasing processing unit 64A in the frequency analysis unit 60A of FIG. Phase adjusting unit 64
B is provided, and other parts of 61 to 63 and 65 are the same as those in FIG. Frequency analysis unit 6 of FIG.
In 0B, the cross-correlation matrix calculation unit 61, the prediction coefficient matrix calculation unit 62, and the cross frequency spectrum matrix calculation unit 63.
Performs the same operation as in FIG.

That is, in the frequency analysis unit 60B, first, in the cross-correlation matrix calculation unit 61, a cross-correlation matrix composed of cross-correlation coefficients between M received signals input from the receiver group 50 is calculated, and the prediction coefficient is calculated. It is output to the matrix calculation unit 62. The prediction coefficient matrix calculation unit 62 calculates the prediction coefficient matrix from the input cross-correlation matrix using the multivariate linear prediction analysis method, and outputs the prediction coefficient matrix to the cross frequency spectrum calculation unit 63. The cross frequency spectrum calculation unit 63 calculates a cross frequency spectrum matrix for each analysis frequency using the prediction coefficient matrix, and inputs it to the matrix phasing processing unit 64B. The matrix phasing processing unit 64B calculates the phasing cross frequency spectrum by adding the phase corresponding to the desired direction using the non-diagonal components of the cross frequency spectrum matrix, and inputs this to the frequency spectrum calculation unit 65. The frequency spectrum calculation unit 65 calculates a frequency spectrum from the phasing cross frequency spectrum for each analysis frequency and outputs it to the display unit 70.

The operation of the second embodiment shown in FIG. 2 will be described in detail.
However, the cross-correlation matrix calculation unit 61 described in the first embodiment
Since the operation description including Expressions (2) to (9) from to the cross frequency spectrum matrix calculation unit 63 is the same in Embodiments 1 and 2, duplicate description will be omitted. In the matrix phasing processor 64B, the cross frequency spectrum matrix P
Of the components of _C (f), the phase of one symmetric component of the non-diagonal elements that is the cross frequency spectrum component, that is, the component p _{Ci, j} (f) where i <j, is set according to the phasing direction θ. Phased cross frequency spectrum p _c (f) is calculated by aligning and adding to the reference phase using the defined phase compensation.

Here, as the reference phase, a diagonal component with i = j is used. This component becomes an auto power spectrum between the same receiver signals (not a cross (mutual) but a power spectrum of an auto (self) receiver) and can be treated as a real number, so that the phase is 0 degree regardless of the row number i. Become. The reason why i <j is used for the phasing is as follows. First, the cross frequency spectrum matrix P
_{Since C} (f) can be treated as a matrix in which diagonal elements are conjugate, the elements of the i-th row and the j-th column and the elements of the j-th row and the i-th column have the same peculiarity of information on the strength and phase, and thus One having information has one of i ≦ j. In the cross frequency spectrum component with i ≠ j, the noise component is reduced from the auto power spectrum component with i = j due to the mutual correlation of noises.
A component that is j may be used, but i ≠
It is advantageous to use the component that is j in order to reduce noise from the phased output.

Further, in the case of a sound source, the relative phase between the receivers is unique to the relative position between the receivers (and is unique depending on the direction of arrival of the sound source). The number is only (M-1) out of a total of M (M-1) / 2 elements where i <j, where M is the number of receivers. If the elements of the cross-frequency spectrum matrix for noise are random values that are determined only by the relative positions between the receivers, the elements that are candidates for the phasing process are (M-1) (diagonal) of any one row. If components are also used, only M) is meaningful. However, when the signal received by the receiver is independent noise both in time series and between the mutual signals, there is no meaningful prediction because they are unrelated signals in the first place,
Regarding the phase of the cross frequency spectrum, even if the relative position of the receiver is the same, if the row number is different, the phase is also different.

That is, the phase difference between the phase of the j-th column of the i-th row and the reference phase of the i-th column and the phase difference of the phase of the j'th column of the i'th row with respect to the reference phase of the i'th column are Difference (j-i)
And (j'-i ') are equal or different. Therefore, if phase compensation according to the direction of the sound source and the relative position of the receiver is performed on the element of i <j,
All are phased to the same phase, and each element of noise has a different phase, and meaningful phasing processing can be performed over M (M-1) / 2 elements. For the above reasons, it is calculated on the basis of the phasing cross frequency spectrum p _c (f) in the following equation (15). Here, the receivers are linearly arranged at equal intervals d, and the direction of the sound source signal is θ (the direction of the arrangement of the receivers is 0 °).

[0053]

(Equation 15)

However, in equation (15), b _ij (θ, f) is the i-th
This is phase compensation for the component in the j-th row and is given by the following equation (16), where q is an imaginary unit.

[0055]

[Equation 16]

The frequency spectrum calculation unit 65 calculates the frequency spectrum P (f) based on the equation (14) using _pc (f) of the equation (10) and outputs it to the display 70. The display unit 70 displays the frequency spectrum S (f) for each analysis frame according to its strength. The operation according to the second embodiment of the present invention will be shown below by simulation. The analysis frequency band is 100 Hz, the noise is band-limited to this band, and the receivers are linearly arranged at intervals of 1/2 of the wavelength of 100 Hz. The sound source is a tone signal with a frequency of 50 Hz that is incident from the 0 degree direction. The frequency spectrum shown is an average of the frequency spectra obtained in each analysis frame for 10 frames.

FIG. 3 is a diagram showing a frequency spectrum analysis result when the number M of receivers is set to 1 in the second embodiment. The horizontal axis represents frequency (Hz) and the vertical axis represents frequency spectrum value (dB). ). In FIG. 3, since M = 1, there is no phasing process, and the elements extracted from the cross frequency spectrum matrix are those of the first row and first column. Therefore, this output will be the same as the linear predictive analysis performed with one receiver. The average level of the frequencies excluding the frequency of the sound source from this analysis result is -8 dB when the power of the sound source frequency is 0 dB.

FIG. 4 is a diagram showing a frequency spectrum analysis result when the number M of receivers is 4 in the second embodiment. In FIG. 4, the average level of frequencies excluding the frequency of the sound source is −19 dB. Therefore, the noise reduction effect in this case according to the present invention is 11 dB (= 19 dB-8d).
B). This is the noise reduction effect obtained by four receivers in the conventional phasing process. 6 dB (= 10 log
(4)) A higher noise reduction effect.

According to the second embodiment, since the frequency analysis method of the incoming received signal is configured as described above, in the extraction of the frequency characteristic of the sound source signal coming from a specific direction, the phasing having a high noise reduction effect is performed. It is possible to provide a frequency analysis method of an incoming received signal that can be processed and can detect a signal having a smaller signal strength. Further, since it is based on the linear predictive analysis, it has a better analysis capability in a shorter time than the frequency analysis by the Fourier transform, and it is possible to accurately extract the transient sound of a ship or the like and the sound source signal in the narrow band analysis.

[0060]

As described above, according to the present invention, acoustic signals coming from the outside are respectively received by a plurality of receivers, and a linear prediction model is applied during the time series of the received plurality of received signals. In the frequency analysis method of the incoming received signal, the cross frequency spectrum between the received signals is calculated using linear prediction analysis, the frequency spectrum is calculated using the cross frequency spectrum, and the cross frequency spectrum is image-displayed. When calculating the frequency spectrum using the frequency spectrum, the cross frequency spectrum between one receiver arbitrarily selected and the other receiver,
Since the frequency spectrum is calculated by adding after performing phase compensation so that the phase is aligned with the desired direction, in the extraction of the frequency characteristics of the sound source signal coming from a specific direction, the direction limiting function and noise The noise reduction effect can be further added by the cross frequency spectrum calculation process provided separately from the phasing process having the reduction function.
It becomes possible to detect a signal having a smaller signal strength. Furthermore,
Since it is based on the linear predictive analysis, it has a better analysis capability in a shorter time than the frequency analysis by the Fourier transform, and enables accurate extraction of the transient sound of a ship or the like and the sound source signal in the narrow band analysis.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of a frequency analysis method for an incoming received signal according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram of a frequency analysis method for an incoming received signal according to a second embodiment of the present invention.

FIG. 3 is a diagram showing a frequency spectrum analysis result when the number of receivers is 1 in the second embodiment.

FIG. 4 is a diagram showing a frequency spectrum analysis result when the number of receivers is 4 in the second embodiment.

FIG. 5 is a configuration diagram of a conventional passive sonar device.

[Explanation of symbols]

 50 receiver group 60A, 60B frequency analysis unit 61 cross-correlation matrix calculation unit 62 prediction coefficient matrix calculation unit 63 cross frequency spectrum calculation unit 64A phasing processing unit 64B matrix phasing processing unit 65 frequency spectrum calculation unit 70 display unit

Claims (2)

[Claims] 1. An inter-reception signal is received using a linear prediction analysis in which an acoustic signal coming from the outside is received by a plurality of receivers and a linear prediction model is applied between time series of the received plurality of reception signals. In the frequency analysis method of the incoming received signal for calculating the frequency spectrum using the cross frequency spectrum and displaying the frequency spectrum intensity as an image, and calculating the frequency spectrum using the cross frequency spectrum. At this time, it is possible to calculate the frequency spectrum by adding the cross frequency spectrums of one receiver arbitrarily selected and other receivers after phase compensation so that the phases are aligned with respect to the desired azimuth. A characteristic frequency analysis method for incoming signals. 2. An acoustic signal arriving from the outside is received by each of a plurality of receivers, and a linear prediction model is applied between time series of the received plurality of received signals. In the frequency analysis method of the incoming reception signal for calculating the cross frequency spectrum matrix, phasing the components of the cross frequency spectrum matrix, calculating the frequency spectrum using the phasing value, and displaying the frequency spectrum intensity as an image. , When performing phasing using the components of the cross frequency spectrum matrix, the pair included in each row or column for the non-diagonal component on one side of the cross frequency spectrum matrix or the component including the diagonal component Using the angular component as the reference phase, the other components contained in the same row or column are aligned in the desired direction, and this is applied to all rows or columns. Frequency analysis method of the incoming received signal and obtaining a phasing adding and outputting the phasing values obtained by performing.