JP2001343448A - System for measuring location of generation of impulse sound - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a system for measuring the location of the occurrence of impulse sound, capable of outputting only a relative sound source location computed from the combination of arrival time differences of only impulse sound as the location of the occurrence of impulse sound by preventing a relative sound source location of noise from being outputted as the location of the occurrence of impulse sound even if wrong detection occur due to noise. SOLUTION: In the system for measuring the location of the occurrence of impulse sound, acoustic sensors are constituted as an acoustic sensor array, and the measuring system is provided with a means for computing a first arrival location of impulse sound at the acoustic sensor array through the use of the acoustic sensor array group, a means for geometrically computing a second arrival direction at the acoustic sensor array from the computed relative sound source location, and a means for determining the reliability of the relative sound source location by comparing the direction difference between the first arrival direction and second arrival direction. When the direction difference between the first arrival direction, computed by the acoustic sensor array and the second arrival direction computed from the relative sound source location, which is the point of intersection of hyperbolas, lies within a fixed error range, it is decided as being a correct sound source location.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention uses an acoustic sensor group in which an impulse-like sound (impulse sound) such as an explosion sound is superimposed on surrounding background noise at intervals of several kilometers. From the hyperbolic group calculated from the arrival time difference of the impulse sound between any two acoustic sensors,
The present invention relates to an impulse sound generation position measurement system that calculates a relative sound source position, which is a position where an impulse sound is generated for a plurality of acoustic sensor groups.

[0002]

2. Description of the Related Art As one of the impulse sound generation position measuring systems, an impulse sound such as an explosion sound is received by using a group of acoustic sensors arranged at intervals of several kilometers, and a set of two arbitrary acoustic sensors is provided. The arrival time difference between the acoustic sensors of the impulse sound is calculated, and a hyperbola where the arrival time difference between the pair of acoustic sensors is constant is obtained from the arrival time difference. The intersection
There is an impulse sound generation position measuring system that calculates a relative position of the impulse sound with respect to the acoustic sensor group.

FIG. 6 is a configuration diagram of a conventional impulse sound generation position measuring system when the number of acoustic sensors is three.
In the figure, 1-1-1 to 1-3-1 are input terminals from an acoustic sensor, 3-1 to 3-3 are pulse detectors, and 5-1.
5-2 is an arrival time difference calculator, 6 is a sound source position calculator, 9
Is an output terminal.

FIG. 2 is a detailed block diagram of the k-th pulse detector. In the figure, reference numeral 10 denotes an input terminal 1 of the acoustic sensor.
A connection terminal of the pulse detector 3-k connected to -k-1,
11 is a short-time integrated power calculator, 12 is a background noise level estimator, 13 is a detection threshold value calculator, 14 is an impulse sound detector, 15 is a waveform extractor, 16 is arrival time difference calculators 5-1 and 5 -2 or a connection terminal of the pulse detector 3-k connected to either of them.

Next, the operation of the impulse sound generation position measuring system will be described.

[0006] The acoustic signals received by the three acoustic sensors are converted into input terminals 1-1 as time series data of digital signals.
Through 1-1-3-1, the pulse detectors 3-1 to 3-3
Respectively.

The pulse detector 3-1 converts the time-series data of the acoustic signal received by the acoustic sensor from the input terminal 1-1-1 through the connection terminal 10 to the short-time integral power calculator 11 and the waveform extractor 15 as digital signals. take in.

The short-time integrated power calculator 11 has an input terminal 1
-1-1 time-series data S of the audio signal input from
₁ (n) is subjected to a square / integral process by the following equation (1) to calculate a short-time integrated power, and the result is used as the background noise level estimator 12 and the impulse sound detector 1
4 is output. Here, the integral coefficient α of the integral processing of the following equation (1) is set to a value within a range where the peak level of the impulse sound does not significantly decrease.

P (n) = α · S ₁ (n) ² + (1−α) · P (n−1) (1) where α: integration coefficient P (n): shortness of the n-th sample Time integrated power Background noise level estimator 12 is short-time integrated power calculator 1
The background noise level NL (n) is calculated by further performing an integration process on the short-time integrated power P (n) input from 1 by the following equation (2), and the result is used as a detection threshold value calculation. Output to the container 13. Here, the integration coefficient β in the following equation (2) is set to a value within a range in which the peak level of the impulse sound is sufficiently small and the fluctuation of the background noise level can be followed.

NL (n) = β · P (n) + (1−β) · NL (n−1) (2) where β: integration coefficient NL (n): background noise estimation of the n-th sample The level detection threshold calculator 13 uses the background noise level NL (n) from the background noise level estimator 12 to calculate a threshold THL for detecting an impulse sound according to the following equation (3).
(N) is calculated, and the result is output to the impulse sound detector 14. Here, the constant γ is set in consideration of the detection probability and the false alarm probability.

THL (n) = γ · NL (n) (3) where γ: constant set in consideration of detection probability and false alarm probability THL (n): detection threshold value of n-th sample The sound detector 14 is a short-time integrated power calculator 1
The short-time integration power P (n) from 1 and the detection threshold value THL (n) from the detection threshold value calculator 13 are input, and n becomes THL (n) <P (n) (4). When the sample is continuous for a certain period, that is, when the sample in which the short-time integrated power exceeds the detection threshold value THL (n) is continuous for a certain period, a starting sample number satisfying the expression (4) assuming that an impulse sound exists at that time. k
s and the end sample number ke are output to the waveform extractor 15.

The waveform extractor 15 receives the acoustic signal S received by the acoustic sensor from the input terminal 1-1-1 through the connection terminal 10.
₁ (n) and the impulse sound existence section [ks, ke] from the impulse sound detector 14 are input, the impulse sound existence section [ks, ke] is included, and the processing of the subsequent arrival time difference calculator 5-1 is performed. Sound signal S by the amount required for
₁ (n) cut _{out, S 1 (n) n =} D 1, D 1 + 1, ..., E 1 -1, E 1, N 0 = E 1 -D 1 +1 ... (5) connecting the result terminal 16 Arrival time difference calculator 5-
Output to 1.

The same processing is performed for the pulse detectors 3-2 and 3-3, and the sound signals S ₂ (n) n = D ₂ , D ₂ +1,..., E cut out by the pulse detector 3-2. _{2 -1, E 2, n 0} = E 2 -D 2 +1 ... (6) is the arrival time difference calculator 5-1 and 5-2, the acoustic signal S ₃ which is cut by the pulse detector 3-3 (n _{) n = D 3, D 3} + 1, ..., E 3 -1, is output to the _{E 3, n 0 = E 3} -D 3 +1 ... (7) is the arrival time difference calculator 5-2.

The arrival time difference calculator 5-1 includes the pulse detector 3
-1 and 3-2, the time-series data S ₁ (n) (n = D ₁ , D ₁ +1,..., E ₁ −1) of the audio signal cut out
E ₁ ) and S ₂ (n) (n = D ₂ , D ₂ +1,..., E ₂
−1, E ₂ ), the arrival time difference τ ₁₂ between the two waveforms is calculated in consideration of the cut-out times D ₁ and D ₂ , for example, by using the correlation processing method of the following equation (8). The result is output to the sound source position calculator 6.

[0015]

(Equation 1)

Where S ₁ '(i): S ₁ (n) is a relative number i (i = 0,
1,..., N ₀ -1) S ₂ ′ (i): S ₂ (n) is replaced by a relative number i (i = 0,
1, 1, N ₀ -1) S ₁ ′ (i) = 0 i <0 or i> N ₀ −1 τ ₁₂ = [time difference at peak point of correlation curve of C ₁₂ (τ)] + (D ₂ −D ₁ ) (9) Similarly, the arrival time difference calculator 5-2 also outputs the acoustic signal S ₂ (n) (n = D) cut out from the pulse detectors 3-2 and 3-3.
_{2, D 2 + 1, ...} , E 2 -1, E 2) and S ₃ (n)
Using (n = D ₃ , D ₃ +1,..., E ₃ −1, E ₃ ), the arrival time difference τ ₃₂ between the two waveforms is calculated, and the result is output to the sound source position calculator 6.

The sound source position calculator 6 has an arrival time difference calculator 5-
Using the arrival time difference τ ₁₂ and τ ₃₂ from 1 and 5-2, the intersection of the hyperbola where the arrival time difference τ ₁₂ is constant and the hyperbola where the arrival time difference τ ₃₂ is constant are obtained by solving the following simultaneous equations, The coordinates of the intersection are output to the output terminal 9 as a relative sound source position, that is, an impulse sound generation position.

Hereinafter, the relationship between the arrangement of the acoustic sensors and the sound source position will be described with reference to FIG.

As shown in FIG. 8, the acoustic sensor 2 is
At the origin (0,0) as the reference position for the
Let the coordinates of sensor 1 be (x₁, 0), the coordinates of the acoustic sensor 3
(X_Three, Y_Three) Does not lose versatility. Relative sound
The source position is (x_B, Y_B), The relative sound source position
(X_B, Y_B) Is the five simultaneous equations (10) to (14)
It is obtained by solving the equation. These equations are unknown 5
(X_B, Y_B, T₁, T_Two, T _Three), The independent equation
Since there are five, we can solve it.

[0020] ^{_{(x B -x 1) 2 +}} y B 2 = (c · T 1) 2 ... (10) x B 2 + y B 2 = (c · T 2) 2 ... (11) (x B -x 3 _{^{) 2 + (y B -y 3}} ) 2 = (c · T 3) 2 ... (12) T 1 -T 2 = τ 12 ... (13) T 3 -T 2 = τ 32 ... (14) However, T ₁ : Propagation time for the impulse sound to reach the acoustic sensor 1 T ₂ : Propagation time for the impulse sound to reach the acoustic sensor 2 T ₃ : Propagation time for the impulse sound to reach the acoustic sensor 3 c: Sound speed In the above prior art, the case where the number of acoustic sensors is 3 has been described, but when the number of acoustic sensors is 4 or more, the acoustic sensor is set to M
Then, there are _three sets of _M C that take three sensors out of the number of M sensors. After calculating the relative sound source positions of the _three sets of _M C by the above method, the average value is output as the impulse sound generation position. A method is generally taken.

[0021]

However, the conventional impulse sound generation position measuring method has the following problems.

FIG. 7 is a diagram for explaining a problem of the conventional impulse sound generation position measuring system.

For example, when the acoustic sensor receives background noise together with an impulse sound, the pulse detectors 3-1 to 3-
3 not only detects the received impulse sound,
Noise may be erroneously detected. As background noise, wind noise of the acoustic sensor, local sound around the acoustic sensor,
For example, a sound of an animal or the like or a sound of a small firearm can be considered, and these are likely to be erroneously detected as impulse sounds such as explosion sounds from the frequency and waveform.

FIG. 7 shows output signals 3-1-1 to 3-1-1-3 of the pulse detectors 3-1 to 3-3 when one impulse sound and background noise generated at a certain sound source position are received by the acoustic sensors 1 to 3.
3-3-1 is shown. Output signals 3-1-1 and 3-3
-1, the pulse detectors 3-1 and 3-3 detect only the impulse sound, and the pulse detector 3-2 detects the impulse sound from the output signal 3-2-1, and the noise is erroneously detected. .

The operation of the impulse sound generation position measuring method when the impulse sound is detected and the background noise is erroneously detected will now be described.

In such a situation, when the conventional method for measuring the position at which an impulse sound is generated is applied, the arrival time difference τ calculated by the arrival time difference calculators 5-1 and 5-2 becomes the impulse sound detection times T ₁ , T ₂ , T ₃ is not only the time difference by the time difference between the noise false detection time T ₄ also calculates the time difference between the plurality of combinations of the following is calculated.

Arrival time difference calculator 5-1: T ₂ -T ₁ , T ₄ -T ₁ Arrival time difference calculator 5-2: Calculated by sound source position calculator 6 by T ₂ -T ₃ and T ₄ -T ₃ The relative occurrence position is: Relative occurrence position 1: intersection of hyperbolas calculated by (T ₂ −T ₁ , T ₂ −T ₃ ) Relative occurrence position 2: (T ₄ −T ₁ , T ₄ −T ₃ ) The two relative sound source positions at the intersection of the hyperbolas calculated by are calculated. As a result, only the relative occurrence position 1 calculated from the combination of the arrival time differences of the impulse sounds alone should be output as the impulse sound occurrence position, but the existence of the sound source calculated from the combination of the arrival time differences between the impulse sounds and the noise There is a problem that the relative generation position 2 which is not output is also output as the impulse sound generation position.

According to the present invention, when means for calculating the direction of arrival of the impulse sound detected by each acoustic sensor is provided,
Focusing on the fact that the actual impulse sound has a correlation between the direction of arrival measured by each acoustic sensor and the relative sound source position, whereas there is no correlation with noise, The acoustic sensor is configured as an acoustic sensor array, and the direction of arrival of the impulse sound with respect to each acoustic sensor array is calculated using each acoustic sensor array, while the relative point, which is the intersection of hyperbolas at which the arrival time difference between each acoustic sensor array is constant, is calculated. From the sound source position and the position of each acoustic sensor array, the geometric arrival direction of the relative sound source position with respect to each acoustic sensor array is calculated, and the azimuth difference between the arrival direction of the impulse sound and the geometric arrival direction is used. It is determined whether the relative sound source position is the relative sound source position of the impulse sound or the relative sound source position of the noise, and only the relative sound source position of the impulse sound is subjected to the impulse sound. There is provided an impulse sound generating position measurement system to output as a generation position.

That is, the present invention, in view of the above situation,
Even if erroneous detection due to noise occurs, the relative sound source position of the noise is not output as the impulse sound generation position, and only the relative sound source position calculated from the combination of the arrival time differences of the impulse sounds only is set as the impulse sound generation position. It is an object of the present invention to provide an impulse sound generation position measuring system capable of outputting.

[0030]

According to the present invention, in order to achieve the above object, [1] an impulse sound such as an explosion sound is received by using a group of acoustic sensors arranged at intervals of several km, and is optionally selected. The arrival time difference between the acoustic sensors of the impulse sound is calculated using the two acoustic sensors as one set, and a hyperbola that makes the arrival time difference between the one set of acoustic sensors constant is calculated from the arrival time difference. In the impulse sound generation position measurement system that calculates the intersection of the hyperbolic curve obtained between the acoustic sensors as the relative sound source position of the impulse sound with respect to the acoustic sensor group, the acoustic sensor is configured as an acoustic sensor array, and the acoustic sensor array is Means for calculating a first direction of arrival of the impulse sound with respect to the acoustic sensor array using the acoustic sensor array based on the calculated relative sound source position. Means for geometrically calculating a second direction of arrival with respect to the array, and means for determining the reliability of the relative sound source position by comparing the direction difference between the first direction of arrival and the second direction of arrival. Equipped with the first calculated by the acoustic sensor array
Is determined as a correct sound source position when the direction difference between the second direction of arrival calculated from the relative sound source position, which is the intersection of the arrival direction and the hyperbola, is within a certain error range.

[2] In the impulse sound generation position measuring system according to [1], the acoustic sensor array may be used as a means for calculating a first direction of arrival of the impulse sound with respect to the acoustic sensor array. Calculating the arrival time difference between the acoustic sensors of the impulse sound as a set of any two acoustic sensors in the array;
The arrival direction of the impulse sound to the acoustic sensor array is calculated from the one or more arrival time differences.

[3] In the impulse sound generation position measuring system according to [1], the acoustic sensor array may be used as a means for calculating a first arrival direction of the impulse sound with respect to the acoustic sensor array. Perform standby phasing in multiple directions using an array,
The arrival direction of the impulse sound with respect to the acoustic sensor array is calculated by determining the direction in which the phasing processing output of the standby phasing processing is maximized.

[0033]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a functional block diagram of an impulse sound generation position calculation system according to a first embodiment of the present invention. In this embodiment, the number of acoustic sensor arrays is 3, and the number of acoustic sensors in each acoustic sensor array is 2. Has been taken up. Three acoustic sensor arrays are installed at a distance of several [km], and two acoustic sensors in each acoustic sensor array are several [m].
It is installed at intervals of up to several tens [m].

In FIG. 1, 1-1-1 and 1-1-2 are input terminals from the acoustic sensors in the acoustic sensor array 1;
-2-1, 1-2-2 are input terminals from the acoustic sensors in the acoustic sensor array 2; 1-3-1, 1-3-2 are input terminals from the acoustic sensors in the acoustic sensor array 3; -1 to
3-3 is a pulse detector, 4-1 to 4-3 are arrival direction calculators, 5-1 and 5-2 are arrival time difference calculators, 6 is a sound source position calculator, 7 is a direction error amount calculator, 8 Is a decision processor, 9
Is an output terminal. Details of the configuration of the pulse detectors 3-1 to 3-3 will be described with reference to FIG.

Next, the operation of the impulse sound generation position calculation system will be described with reference to FIGS.

The acoustic signals received by the two acoustic sensors in each acoustic sensor array are passed through the input terminals 1-1-1 to 1-3-2 as time series data of digital signals, and output from the pulse detectors 3-1 to 1-3-2. 3-3 and arrival direction calculator 4-1 to 4
-3 respectively.

Here, along with the description of each processor, the situation described in the problem to be solved by the present invention, that is, as shown in FIG. 7, the case where the detection of an impulse sound and the erroneous detection of noise occur. Will be described as necessary.

The pulse detector 3-1 converts the time series data of the acoustic signal received by the acoustic sensor from the input terminal 1-1-1 through the connection terminal 10 into a digital signal, and calculates the short-time integrated power calculator 11 and the waveform extractor. Take in to 15.

The short-time integrated power calculator 11 has an input terminal
Sound signal S input from 1-1-1 _1-1For the expression
By performing the square and integration processing according to (1), short-time integration can be performed.
Power P (n) is calculated, and the result is estimated as the background noise level.
Output to the constant detector 12 and the impulse sound detector 14. here
And the integration coefficient α of the integration processing of the equation (1) is
Set a value within the range where the peak level does not decrease much.
Set.

The background noise level estimator 12 outputs the short-time integrated power P (n) input from the short-time integrated power calculator 11.
In addition, the background noise level NL (n) is calculated by further executing the integration processing according to the above equation (2), and the result is output to the detection threshold value calculator 13. Here, equation (2)
Is set to a value within a range where the peak level of the impulse sound is sufficiently low and the fluctuation of the background noise level NL (n) can be followed.

The detection threshold calculator 13 uses the background noise level NL (n) from the background noise level estimator 12 to calculate
The threshold value THL (n) for detecting the impulse sound is calculated by the above equation (3), and the result is output to the impulse sound detector 14. Here, the constant γ is set in consideration of the detection probability and the false alarm probability.

The impulse sound detector 14 receives the short-time integrated power P (n) from the short-time integrated power calculator 11 and the detection threshold value THL (n) from the detection threshold value calculator 13 and calculates If n that satisfies (4) is continuous for a certain period, that is, if samples in which the short-time integrated power P (n) exceeds the detection threshold value THL (n) continue for a certain period, an impulse sound exists at that time. As a result, a start sample number ks and an end sample number ke that satisfy Expression (4) are calculated, and the impulse sound existence section [ks, k
e] to the waveform extractor 15.

The waveform extractor 15 receives the time series data S _1-1 (n) received by the acoustic sensor from the input terminal 1-1-1 through the connection terminal 10 and the impulse sound existence section [ks, enter the ke], impulse noise present interval [ks, encompasses ke], and cut by an amount acoustic signals required TDOA processing calculator 5-1 subsequent, S _1-1 (n) n = D ₁ , D ₁ +1,..., E ₁ −1, E ₁ , N ₀ = E ₁ −D ₁ +1 (15) The extracted acoustic signal S _1-1 (n) and the extraction time D ₁ ,
And it outputs the E ₁ to the arrival time difference calculator 5-1 and the arrival direction calculator 4-1 through the connection terminal 16.

The same processing is performed for the pulse detectors 3-2 and 3-3, and the acoustic signal S _2-1 (n) input from the input terminal 1-2-1 and cut out by the pulse detector 3-2. _{n = D 2, D 2 +} 1, ..., E 2 -1, and _{E 2, n 0 = E 2} -D 2 +1 ... (16) and cut the time D _2, E ₂ is the arrival time difference calculator 5-1 5 -2, and an input terminal 1-3 to the arrival direction calculator 4-2.
Acoustic signal cut out by the input from -1 pulse detector _{3-3 S 3-1 (n) n =} D 3, D 3 + 1, ..., E 3 -1, E 3, N 0 = E 3 -D ₃ +1... (17) and the cutout times D ₃ and E ₃ are output to the arrival time difference calculator 5-2 and the arrival direction calculator 4-3.

The arrival direction calculator 4-1 includes a pulse detector 3-
1, the sound signal S _1-1 (n) (n = D ₁ ,
D ₁ +1,..., E ₁ −1, E ₁ ) and the cutout times D ₁ ,
Enter the E _1, the input terminal 1-1-2 to input time-series data of the acoustic signals received by the acoustic sensor, cut out from cutout time D _1, E ₁ (input terminal 1-1-1 With the same cut-out time as the sound data S _1-1 (n), the sound signal is cut out from the time-series data of the sound signal from the input terminal 1-1-2.

[0047] _{S 1-2 (n) n = D} 1, D 1 + 1, ..., E 1 -1, E 1, N 0 = D 1 -E 1 +1 ... (18) Next, for example, the following equation (19 ), The time series data S _1-1 (n) and the time series data S
_1-2 Calculate the arrival time difference ζ ₁ of the (n) waveform,

[0048]

(Equation 2)

[0049] _{However, S 1-1 '(i):} S 1-1 (n) is the relative number i (i = 0,1, ..., N 0 -1) those were replaced with S _1-2' (i ): S _1-2 (n) is replaced by relative number i (i = 0,
1, ..., N ₀ -1) to replace ones _{S 1-1 '(i) = 0} i <0 or i> N 0 -1 ζ 1 = [ time difference at the peak point of the correlation curve of the C ₁₂ (tau) ] ... (20) by using the installation coordinates of two acoustic sensors of the acoustic sensor array 1 which is connected to the arrival time difference zeta ₁ and the input terminal 1-1-1 and 1-1-2, the following equation (21 ), The arrival direction CB1 of the impulse sound is calculated, and is output to the direction error amount calculator 7.

[0050]

(Equation 3)

Where, d _11-12 : the distance between two acoustic sensors in the acoustic sensor array 1 ofs1: the angle between the normal direction of the baseline of the acoustic sensor array 1 and the north (or conversion correction in the true direction) Angle) c: Sound speed The same processing is performed for the arrival direction calculators 4-2 and 4-3, and the arrival direction calculator 4-2 has an input terminal 1-2.
1 from which the audio data S _2-1 (n) (n = D
_{2, D 2 + 1, ...} , E 2 -1, time D Log and E _2)
2 , E ₂ and the arrival direction CB2 calculated using the time-series data of the acoustic signal from the input terminal 1-2-2 are output to the direction error amount calculator 7, and the arrival direction calculator 4-3 outputs Time series data S cut out from the input terminal 1-3-1
_{3-1 (n) (n = D} 3, D 3 + 1, ..., E 3 -1,
E ₃ ), the cut-out times D ₃ and E _3, and the input terminal 1-2.
The azimuth CB3 calculated using the time-series data of the acoustic signal from the azimuth 2 is output to the azimuth error calculator 7. Here, in the case of Figure 7, since the calculated arrival direction for all of the detection signals, the calculation result B1 of the arrival direction CB1 to an impulse sound arrival time T ₁ from the arrival direction calculator 4-1, arrival direction the noise calculation result B4 of the calculation result B2 and arrival time T ₄ in the arrival direction CB2 to an impulse sound arrival time T ₂ are the calculator 4-2, arrival direction calculator 3
From -3 calculation result B3 of the arrival direction CB3 to an impulse sound arrival time T _3, but are output.

The arrival time difference calculator 5-1 outputs the sound signal S _1-1 (n) (n = n) cut out from the pulse detector 3-1.
_{D 1, D 1 + 1,} ..., E 1 -1, E 1) and the time D _1, E ₁ cut, acoustic signal S _2-1 cut out from the pulse detector 3-2 (n) (n = _{D 2, D 2 + 1,} ..., E 2
−1, E ₂ ) and the cut-out times D ₂ , E _2, and taking into account the cut-out times D ₁ and D ₂ by using, for example, the correlation processing method according to the equations (8) and (9). Calculates the arrival time difference τ ₁₂ between the two waveforms, and outputs it to the sound source position calculator 6.

The same processing is performed for the arrival time difference calculator 5-2, and the sound signal S _2-1 (n) (n = D ₂ , D ₂ +1,...) Cut out from the pulse detector 3-2. E
₂ -1, E ₂₎ and the time D _2, E ₂ cut, the sound signal S _3-1 cut out from the pulse detector 3-3 (n)
Using (n = D ₃ , D ₃ +1,..., E ₃ −1, E ₃ ) and the cutout times D ₃ , E ₃ , the arrival time difference τ ₃₂ between both waveforms is calculated, and the sound source position calculator 6 Output to

Here, in the case of FIG. 7, all the detection signals
Arrival time difference τ ₁₂And τ₃₂Calculate
Therefore, the arrival time difference calculated by the arrival time difference calculator 5-1
τ₁₂And arrival time difference calculated by arrival time difference calculator 5-2
τ₃₂Is the time difference τ₁₂: T_Two-T₁, T_Four-T₁ Time difference τ₃₂: T_Two-T_Three, T_Four-T_Three Becomes

The sound source position calculator 6 has an arrival time difference calculator 5-
And the arrival time difference tau ₁₂ from 1, the incoming type from the time difference calculator 5-2 and arrival time difference tau _32, hyperbolic intersections hyperbolic and TDOA tau ₃₂ of TDOA tau ₁₂ is constant is constant formula ( 10)-It is determined by solving the simultaneous equations of Expression (14). When a plurality of arrival time differences are input from each of the arrival time difference calculators 5-1 and 5-2, the intersection of the hyperbolas is calculated for each combination of the arrival time differences, and the combination information of the arrival time differences is used to calculate the azimuth error amount. The coordinates of the intersection of the hyperbola are output to the azimuth error calculator 7 and the determination processor 8 as the relative sound source position of the impulse sound.

Here, in the case of FIG. 7, the sound source position calculator 6 sets the combination information 1: (τ ₁₂ , τ ₃₂ ) = (T ₂ −T ₁ , T
₂ −T ₃ ) Combination information 2: (τ ₁₂ , τ ₃₂ ) = (T ₄ −T ₁ , T
₄ -T ₃₎ of the outputs of two, a relative sound source position 1 which is calculated by using a set of TDOA tau ₁₂ and tau ₃₂ combination information 1, a set of TDOA tau ₁₂ and tau ₃₂ of combination information 2 And outputs two relative sound source positions of the relative sound source position 2 calculated using.

The azimuth error amount calculator 7 uses the relative sound source position and combination information from the sound source position calculator 6 and the arrival directions at the respective acoustic sensor arrays from the arrival direction calculators 3-1 to 3-3. For example, the azimuth error amount is calculated by the following method, and is output to the determination processor 8 as the azimuth error amount corresponding to the relative sound source position.

An azimuth (hereinafter referred to as a geometric arrival azimuth) PB when the relative sound source position is viewed from each of the acoustic sensor arrays 1 to 3
1, PB2, and PB3 are geometrically determined from the relative sound source position coordinates and the installation coordinates of each of the acoustic sensor arrays 1 to 3, and are geometrically determined for each of the acoustic sensor arrays 1 to 3 as in the following equation (22). The difference between the arrival directions PB1, PB2, and PB3 and the arrival directions CB1, CB2, and CB3 in each acoustic sensor array from the pulse detectors 3-1 to 3-3 is calculated, and the average value of the three difference values is calculated. Azimuth error amount D corresponding to relative sound source position
It is calculated as err.

[0059]

(Equation 4)

Here, in the case of FIG. 7, two relative sound source positions, a relative sound source position 1 and a relative sound source direction 2, and arrival directions B1 to B
4 and the azimuth error amount 1 and the azimuth error amount 2
Are output. When calculating the azimuth error amount, the relative sound source position (geometric arrival direction) and the arrival direction are selected using the combination information.
As shown in (3) and (24), in order to calculate the azimuth error amount 1, the relative sound source position 1 (geometric arrival azimuth PB1
1, PB12, PB13) and arrival directions B1, B2, B3
In order to calculate the azimuth error amount 2, the relative sound source position 2 (geometric arrival directions PB21, PB22, PB2
3) and the arrival directions B1, B4, B3.

[0061]

(Equation 5)

The decision processor 8 receives the relative sound source position from the sound source position calculator 6 and the azimuth error amount from the azimuth error amount calculator 7 so that the azimuth error amount falls within the range of the azimuth error threshold _Dth. It is determined whether or not it is included, and only the relative sound source position within the range of the azimuth error threshold is output to the output terminal 9 as the impulse sound generation position. The azimuth error threshold _Dth is set in consideration of the detection rate and the false alarm probability.

In the case of FIG. 7, the judgment processor 8 inputs the relative sound source position 1 and the relative sound source position 2 from the sound source position calculator 6 and the azimuth error amount 1 and the azimuth error amount 2 from the azimuth error amount calculator 7. , It is determined whether the azimuth error amount is within the range of the azimuth error threshold value. Since the azimuth error amount 2 of the relative sound source position 2 in which the combination includes noise increases, only the relative sound source position 1 in which the combination is an impulse sound and the azimuth error amount is small is output to the output terminal 9 as the impulse sound generating position. You.

FIG. 3 is a functional block diagram of an impulse sound generation position calculating system according to a second embodiment of the present invention. Elements common to those in FIG. 1 according to the first embodiment are denoted by the same reference numerals. Have been. In FIG. 3, 1-1-1 to 1-1
-N is an input terminal from N acoustic sensors in the acoustic sensor array 1, 1-2-1 to 1-2-N is an input terminal from N acoustic sensors in the acoustic sensor array 2, 1-3. -1 to
1-3-N is an input terminal from N acoustic sensors in the acoustic sensor array 3, and 2-1 to 2-3 are standby phase adjusters. Details of the configuration of the standby phase adjusters 2-1 to 2-3 will be described with reference to FIG. The details of the configuration of the pulse detectors 3-1 to 3-3 will be described with reference to FIG. 5 because the operation of the impulse sound detector 14 in the related art and the first embodiment is slightly different.

Next, an operation will be described focusing on differences from the first embodiment. The acoustic signals received by the N acoustic sensors in each acoustic sensor array are converted into digital signal time-series data as input terminals 1-1-1 to 1-1-1.
N, 1-2-1 to 1-2-N, 1-3-1 to 1-3-N
The acoustic signals input to the standby phase adjusters 2-1 to 2-3 and received by one acoustic sensor in each acoustic sensor array are input to the input terminal 1-1-1 as digital signal time-series data. , 1-2-1 and 1-3-1 are input to the pulse sound detectors 3-1 to 3-3, respectively.

The standby phase adjuster 2-1 has a connection terminal 31-.
Input terminals 1-1-1 to 1-1-N through 1-31 to N
Received by acoustic sensors in the acoustic sensor array from the
Signal S_1-1(N)-S_1-N(N) is a digital signal
Then, they are taken into the time delay units 32-1 and 32-L. General
The phasing of the received signal of the acoustic sensor array is expressed by the following equation
As shown in (25), the reception signal S of each acoustic sensor_1-i
(N), i = 1, 2,._jPhasing against
Time delay τ_i(Θ_j) Delay processing
And add the delay-processed received signal to obtain the azimuth θ._j
Phasing output b for _1-j(N) can be obtained.

[0067]

(Equation 6)

In this embodiment, the phasing process for one direction is realized by the time delay unit 32-1 and the adder 33-1. Therefore, the time delay units 32-1 to 32-L and the adder 33 are used. the L sets of the phasing process by -1~33-L, phasing output b for orientation theta ₁ through? _L of the plurality (L number)
_1-1 (n) to b _1-L (n) can be calculated. Further, the respective phasing outputs b _1-1 (n) to b _1-L (n) are input to the short-time integrators 34-1 to 34-L, respectively, and are squared by the following equation.
By performing the integration processing, the short-time integration power of the phasing output (hereinafter, referred to as phasing power) BF _1-1 (n) to BF
_1-L (n) is calculated and output to the arrival direction calculator 4-1 through the connection terminals 35-1 to 35-L. Here, the integral coefficient γ of the integral processing of the following equation (26) is set to a value within a range where the peak level of the impulse sound does not significantly decrease.

BF _1-j (n) = γ · b _1-j (n) ² + (1−γ) · BF _1-j (n−1), j = 1, 2,. (26) where γ: integration coefficient b _1-j (n) ² : short-time integrated power standby phase adjuster 2-2 and standby phase adjuster for the n-th sample of the phasing output b _1-j (n) The same processing is performed for 2-3, and the standby phase adjuster 2-2 sets the azimuths θ _{1 to} θ _L of the acoustic sensor array 2.
L phasing powers BF _2-1 (n) to BF
_2-L (n) is output to the arrival direction calculator 4-2, waiting phasing unit 2-3 L number of delay-power BF _3-1 for orientation theta ₁ through? _L of the acoustic sensor array 3 (n ) To BF
_3-L (n) is output to the arrival direction calculator 4-3. The pulse detector 3-1 is connected to the input terminal 1-1 through the connection terminal 10.
The time series data of the acoustic signal received by the acoustic sensor from 1 is taken into the short-time integrated power calculator 11 and the waveform extractor 15 as a digital signal.

The short-time integral power calculator 11 has an input terminal
Sound signal S input from 1-1-1 _1-1For (n)
By performing the square / integral processing by equation (1), the product
Calculate the minute power P (n) and use the result as the background noise level
Output to the estimator 12 and the impulse sound detector 14. This
Here, the integration coefficient α of the integration processing of the equation (1) is represented by an impulse
A value within a range that does not cause a significant decrease in sound peak level
Set.

The background noise level estimator 12 outputs the short-time integrated power P (n) input from the short-time integrated power calculator 11.
Then, the background noise level NL (n) is calculated by further performing an integration process according to Expression (2), and the result is output to the detection threshold value calculator 13. Here, the integration coefficient β in the equation (2) is set to a value within a range where the peak level of the impulse sound is sufficiently low and the fluctuation of the background noise level can be followed.

The detection threshold value calculator 13 uses the background noise level NL (n) from the background noise level estimator 12 to calculate a threshold value THL (n) for detecting an impulse sound according to equation (3). Is calculated, and the result is output to the impulse sound detector 14. Here, the constant γ is set in consideration of the detection probability and the false alarm probability.

(The above operation in the pulse detector is exactly the same as that of the conventional and the first embodiment of the present invention).

The impulse sound detector 14 has the short-time integrated power P (n) from the short-time integrated power calculator 11 and the detection threshold THL (n) from the detection threshold calculator 13.
When n satisfying the expression (4) is continuous for a fixed period, that is, when samples in which the short-time integrated power P (n) exceeds the detection threshold value THL (n) continue for a fixed period, the time , The start sample number ks and the end sample number ke satisfying the equation (4) are calculated, and the impulse sound existence section [k
s, ke], the short-time integrated power P of the impulse sound
(N) (n = [ks, ke]) is cut out, and the impulse sound existence section [ks, ke] is output to the waveform extractor 15;
The impulse sound existence section [ks, ke] and the short-time integrated power P (n) of the cut out impulse sound are output to the peak detector 17.

The waveform extractor 15 receives the time-series data S _1-1 (n) of the acoustic signal received by the acoustic sensor from the input terminal 1-1-1 through the connection terminal 10 and the impulse sound existence section from the impulse sound detector 14. Enter [ks, ke]
Impulse sound presence interval [ks, ke] include, and cut by an amount acoustic signals required TDOA processing calculator 5-1 subsequent by equation (15), an acoustic signal that cut out S _{1- 1} (n) and the cutout times D ₁ and E ₁ are output to the arrival time difference calculator 5-1 through the connection terminal 16.

The peak detector 17 is an impulse sound detector 1
4, the impulse sound existence section [ks, ke] from P.4 and the short-time integrated power P (n) (n =
[Ks, ke]), and taking into account the starting sample number ks of the impulse sound, the peak time kp1 (k) at which the short-time integrated power P (n) of the impulse sound becomes the maximum value.
s <kp1 <ke) is calculated and output to the arrival direction calculator 4-1 through the connection terminal 18.

The same processing is performed for the pulse detectors 3-2 and 3-3, and the sound signal input from the input terminal 1-2-1 and cut out by the pulse detector 3-2 [Equation (16)] Is output to the arrival time difference calculators 5-1 and 5-2, and the peak time kp2 of the impulse sound is output to the arrival direction calculator 4-2, input from the input terminal 1-3-1 and cut out by the pulse detector 3-3. The obtained acoustic signal [Equation (17)] is supplied to the arrival time difference calculator 5-2 to obtain the peak time k of the impulse sound.
p3 is output to the arrival direction calculator 4-3.

The arrival direction detector 4-1 is a pulse detector 3-
The peak time kp1 impulse sound from 1, waiting the L phasing power BF _1-1 for orientation theta ₁ through? _L from the phasing 2-1 and _{(n) ~BF 1-L (} n) input, obtains the instantaneous output value BF _1-1 of KakuSeisho power at peak time kp1 impulse sound _{(kp1) ~BF 1-L (} kp1), the instantaneous output value BF _1-1 (kp1) ~BF _{1- L} (kp
The phasing azimuth which becomes the maximum value in 1) is determined by the acoustic sensor array 1
Is output to the azimuth error amount calculator 7 as the arrival azimuth CB1 of the impulse sound.

The pulse detectors 3-2 and 3-3 are also
The processing is exactly the same, and the arrival direction calculator 5-2 outputs a pulse
Peak time kp2 of the impulse sound from the detector 3-2
And the direction θ from the standby phase adjuster 2-2.₁~ Θ_LAgainst
L phasing powers BF_2-1(N)-BF_2-L(N) and
In the acoustic sensor array 2 calculated using
The arrival direction CB2 of the sound is output to the direction error amount calculator 7,
The arrival direction calculator 4-3 receives the signal from the pulse detector 2-3.
The peak time kp3 of the impulse sound and the standby phase adjuster 2-
Azimuth θ from 3₁~ Θ_LL phasing powers BF for
_3-1(N)-BF _3-L(N), calculated using
The direction of arrival CB3 of the impulse sound at the sensor array 3
Output to the position error amount extractor 7.

The subsequent operations, that is, the operations of the arrival time difference calculators 5-1 and 5-2, the sound source position calculator 6, the azimuth error amount calculator 7, and the decision processor 8 are exactly the same as in the first embodiment. is there.

In the first and second embodiments, only the case where the number of acoustic sensor arrays is three is described. However, when the number of acoustic sensor arrays is four or more, if the number of acoustic sensor arrays is M, the pulse detector is used. And the arrival direction calculator is M
Stage, TDOA calculator is 2 × _M C ₃ pairs (however, TDOA calculator for calculating the same arrival time difference can also be used in common), it construction of later may be the configuration of _M C ₃ pairs . At this time, as mentioned in the prior art, since the combination of the sound source position calculator 6, the M number of sensors takes the three sensors is _M C ₃ sets, _M C ₃ when receiving a single impulse sound Are calculated, _M
Whether to output an average value of C ₃ pieces of relative sound source position, if such to integrate _M C ₃ pieces of relative sound source position in one relative position of the sound source output using the combination information, to obtain the same effect Can be.

Further, in the first embodiment, the case where the number of acoustic sensors in each acoustic sensor array is 2 has been described. However, as the number of acoustic sensors is 3 or more, the arrival direction between each acoustic sensor is calculated. The calculated average value of the direction of arrival may be output to the direction of arrival calculator, or the calculated direction of arrival may be selected based on the S / N ratio or the direction of arrival of the acoustic signal received by each acoustic sensor, and the direction of arrival may be selected. The output of the calculator may be used.

In the first and second embodiments, the pulse detector attempts to detect an impulse sound using an acoustic signal from one input terminal for each acoustic sensor array. It does not matter.

In the second embodiment, the pulse detector attempts to detect an impulse sound using an acoustic signal from one input terminal for each acoustic sensor array. A sound signal from an arbitrary input terminal may be used by selecting a signal having a good S / N ratio from among the above. Further, instead of the sound signal from the N input terminals, a sound from a standby phase adjuster may be used. L phasing outputs may be used.

In the second embodiment, the pulse detector attempts to detect an impulse sound using an acoustic signal from one input terminal for each acoustic sensor array.
A sound signal from an arbitrary input terminal may be used by selecting a signal having a good S / N ratio from among the sound signals from the N input terminals, and a sound signal from the N input terminals may be used. May be used instead of L phasing outputs from the standby phasing device.

Further, in the second embodiment, the arrival azimuth calculator calculates the phasing azimuth having the maximum value among the L power estimation values of the respective phasing azimuths θ _{1 to} θ _L as the arrival azimuth of the impulse sound.
The phasing azimuth which has been output to the azimuth error amount calculator 7 and which has the maximum value among the L power estimation values and the phasing azimuth adjacent to the phasing azimuth are selected. The phase azimuth and the power estimation value may be used to interpolate between the phasing azimuths, and the more accurate arrival azimuth may be calculated by obtaining the peak point of the interpolation curve, and may be output to the azimuth error calculator 7. .

In the first and second embodiments, the azimuth error amount calculator 7 calculates three geometrical arrival directions as viewed from the relative sound source positions from each acoustic sensor array, and calculates the geometrical arrival direction and the arrival direction. The average value of the azimuth differences from the three arrival directions from the calculators 3-1 to 3-3 has been output to the determination processor 7 as the azimuth error amount.
The maximum value, the minimum value, and the root mean square value of the azimuth difference may be obtained as the azimuth error amount and output to the determination processor 7. In this process, various methods for obtaining the azimuth error amount can be used.

Various techniques for determining the time difference can be used.

In the first embodiment, the means for calculating the time difference of arrival in the direction-of-arrival calculators 5-1 and 5-2 has been described using the correlation processing in the time domain.
After calculating the cross spectrum of both, correlation processing in a frequency domain for obtaining a correlation curve by performing inverse FFT may be used.

In the first and second embodiments, the pulse detector 3
Although the background noise level is calculated by further integrating the output of the short-time integration power in the internal configuration of -1 to 3-3, the background noise level may be directly squared from the time-series data and integrated. Alternatively, the output of the square processing in the short-time integrated power calculator 11 may be extracted, and the square output may be integrated to be obtained. Although the integration of the short-time integrated power calculator 11 and the background noise level estimator 12 has been described by exponential integration, simple addition-type integration may be used.

In the first and second embodiments, the calculated position of the sound source is calculated as a relative position with respect to the acoustic sensor group. However, by measuring the absolute position of each acoustic sensor,
The sound source position can be calculated as absolute coordinates.

The devices of the first and second embodiments may be constituted by individual circuits using integrated circuits, or may be constituted by software using a digital signal processor (DSP) or the like. .

The present invention is not limited to the above embodiments, but various modifications can be made based on the spirit of the present invention, and these are not excluded from the scope of the present invention.

[0094]

As described above, according to the present invention, the following effects can be obtained.

(A) When means for calculating the arrival direction of the impulse sound detected at each acoustic sensor position is provided, for the actual impulse sound, the arrival direction measured by each acoustic sensor and the relative sound source position are calculated. While there is a correlation between
Focusing on the fact that there is no correlation with respect to noise, each acoustic sensor is configured as an acoustic sensor array, and while the arrival direction of impulse sound for each acoustic sensor array is calculated,
Calculate the geometric arrival direction from the relative sound source position, which is the intersection of the hyperbolas where the arrival time difference between the acoustic sensor arrays is constant, and use the azimuth difference between the arrival direction of the impulse sound and the geometric arrival direction. , It is possible to determine whether the relative sound source position is the relative sound source position of the impulse sound or the relative sound source position of the noise. Is not output as the impulse sound generation position, and only the relative sound source position calculated from the combination of the arrival time differences of only the impulse sounds can be output as the impulse sound generation position.

(B) A standby phasing process is performed in a plurality of directions using time-series data of an acoustic sensor array including N acoustic sensors in the acoustic sensor array, and the phasing process output of the standby phasing process is performed. Since the direction in which the impulse sound arrives with respect to the acoustic sensor array is calculated by finding the direction in which is the maximum, the same effect as in the first embodiment can be expected.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of an impulse sound generation position calculation system according to a first embodiment of the present invention.

FIG. 2 is a detailed configuration diagram of a k-th pulse detector.

FIG. 3 is a functional block diagram of an impulse sound generation position calculation system according to a second embodiment of the present invention.

FIG. 4 is a standby phase adjuster 2 according to a second embodiment of the present invention.
It is a figure showing the example of the detailed composition of k.

FIG. 5 shows a pulse detector 3-k according to a second embodiment of the present invention.
FIG. 3 is a diagram showing a detailed configuration example of the embodiment.

FIG. 6 is a configuration diagram of a conventional impulse sound generation position measuring system.

FIG. 7 shows a pulse detector 3-k in the case of erroneous detection of noise.
FIG. 4 is a diagram showing output signals of the first embodiment.

FIG. 8 is a diagram showing the relationship between the arrangement of acoustic sensors and the position of a sound source.

[Explanation of symbols]

1, 2 Acoustic sensor array 1-1-1 to 1-1-N, 1-2-1 to 1-2-N, 1
-3-1 to 1-3-N input terminals 2-1 to 2-3 Standby phase adjusters 3-1 to 3-3 Pulse detectors 4-1 to 4-3 Arrival direction calculators 5-1 and 5- 2 Arrival time difference calculator 6 Sound source position calculator 7 Azimuth error amount calculator 8 Judgment processor 9 Output terminal 10, 16, 18, 31-1 to 31-N, 35-1 to 3
5-L connection terminal 11 short-time integrated power calculator 12 background noise level estimator 13 detection threshold calculator 14 impulse sound detector 15 waveform extractor 17 peak detector 32-1-23-L time delay device 33- 1-33-L Adder 34-1-34-L Short-Time Integrator

Claims (3)

[Claims] 1. An impulse sound such as an explosion sound of several km.
Receiving using the acoustic sensor group arranged at intervals, calculating the arrival time difference between the acoustic sensors of the impulse sound as a set of two arbitrary acoustic sensors, and calculating the arrival time difference between the one set of acoustic sensors from the arrival time difference In the impulse sound generation position measuring system for obtaining a hyperbola with a constant arrival time difference at the time and calculating an intersection of the hyperbolas obtained between two or more sets of acoustic sensors as a relative sound source position of the impulse sound with respect to the acoustic sensor group, a) configuring the acoustic sensor as an acoustic sensor array, and calculating a first direction of arrival of the impulse sound with respect to the acoustic sensor array using the acoustic sensor array; and (b) calculating a first direction from the calculated relative sound source position. Means for geometrically calculating a second direction of arrival with respect to the acoustic sensor array; and (c) calculating a first direction of arrival and a second direction of arrival. Means for determining the reliability of the relative sound source position by comparing the positional differences, and (d) the relative sound source position is calculated from the relative sound source position which is the intersection of the first arrival direction and the hyperbola calculated by the acoustic sensor array. An impulse sound generation position measuring system characterized in that a case where the azimuth difference between the second arrival directions falls within a certain error range is determined as a correct sound source position. 2. The impulse sound generation position measuring system according to claim 1, wherein the means for calculating a first direction of arrival of the impulse sound with respect to the acoustic sensor array using the acoustic sensor array is provided in the acoustic sensor array. The arrival time difference between the acoustic sensors of the impulse sound is calculated using the arbitrary two acoustic sensors as a set, and
An impulse sound generation position measuring system, wherein the direction of arrival of the impulse sound with respect to the acoustic sensor array is calculated from the arrival time difference of the set or more. 3. The impulse sound generation position measuring system according to claim 1, wherein the acoustic sensor array is used as a means for calculating a first arrival direction of the impulse sound with respect to the acoustic sensor array using the acoustic sensor array. The standby phasing process is performed in a plurality of directions by using the phasing process output of the standby phasing process, and the direction in which the output of the impulse sound with respect to the acoustic sensor array is calculated by calculating the direction in which the output of the phasing process is maximized. An impulse sound generation position measuring system characterized by the following.