JPS58156877A - Automatic detector for magnetic abnormality - Google Patents

Abstract

PURPOSE:To enable the automatic detection of the presence or absence of inductive waveforms by a target without the influence of the magnitude of signal levels and to improve the efficiency of detection by providing a threshold calculating part which calculates the dispersion of magnetic field measuring signals. CONSTITUTION:A titled detector is constituted of a Haar converting part 2 for Haar conversion of magnetic field measuring signals, a synchronizing pulse generator 3 for controlling the part 2, a discriminating part 5 for discriminating the presence or absence of the magnetic abnormality of a specific target object by comparing the output of the part 2 with the threshold, etc., detects the presence of the magnetic abnormality by the specific target automatically. A threshold calculating part 4 which calculates the dispersion of the magnetic field measuring signals in the past specified time is provided to the detector, and the presence of the magnetic abnormality of the specific target is detected automatically without receiving the influence of the levels of the magnetic field measuring signals by comparing the extracted signal with the automatically set threshold.

Description

[Detailed Description of the Invention] This invention relates to a device that automates the task of measuring magnetic anomalies in the earth and atmosphere using a highly sensitive magnetic anomaly detector mounted on an aircraft and searching for a 2% target. be.

Conventional 1 For example, when searching for a shipwreck as a target, when passing through the soil of the shipwreck, the magnetic anomaly of the earth's magnetism due to the ferromagnetic material such as iron, which is the constituent material, is generated as a guided waveform in the magnetic field that is the output of the magnetic anomaly detector. This was done by having an operator visually judge the magnetic field measurement signals that are continuously drawn on a recording paper by making use of the fact that they appear in the measurement signals.

The reason for adopting this method is that the above-mentioned magnetic field measurement signal contains a lot of noise caused by distortions in the geomagnetic field caused by the topography and geology of the ocean floor, making it difficult to detect guided waveforms based on magnetic anomalies caused by shipwrecks, etc. 1 This is because a highly sophisticated human recognition ability is required, but on the other hand, there is a drawback that a uniform detection function cannot be maintained over a long period of time due to individual differences and fatigue due to human beings being involved in detection.

In order to improve this situation, devices have been devised that automate the detection operations that were previously performed by humans. One of them is to apply the Hanil transformation to the magnetic field measurement signal, and then perform an appropriate linear combination of the resulting ... Lux vectors.

By calculating the signal obtained by squaring this signal, various noises included in the magnetic field measurement signal are removed, and at the same time, the characteristics of the waveform induced by the target object 9, such as a shipwreck, are effectively extracted and set in advance. There is an automatic magnetic anomaly detector that automatically detects the presence or absence of induced waveforms due to shipwrecks, etc. by comparing it with the thread and socihold, but this woodcutter,
Conventional automatic magnetic anomaly detectors had a fixed threshold, so when the level of the VC magnetic field measurement signal changed, the detection results also changed accordingly.

In this invention, a conventional automatic magnetic anomaly detector is equipped with a threshold calculation section, which calculates the dispersion of the magnetic field measurement signal over a certain period of time in the past, and then automatically sets the threshold based on the above dispersion according to the level of the magnetic field measurement signal. Threshold hold that changes to automatic setting Threshold hold By completely calculating the threshold value, the above-mentioned drawback f that occurred due to the conventional fixed threshold value has been removed, and the waveform Q induced by the target object can be calculated without being affected by the magnitude of the signal level.・It attempts to detect the presence or absence completely automatically.

The principle of the automatic magnetic anomaly detector of this invention will be explained. Furthermore, depending on the configuration of the magnetic anomaly detector, the magnetic field measurement signal measured by the magnetic anomaly detector can take the form of either an analog quantity or a digital cover, but here we will use an automatic magnetic anomaly detector that manually generates the analog quantity magnetic field measurement signal. Let me explain using an example.

Figure 1: An example of the magnetic field measurement signal obtained by measuring disturbances in the earth's magnetic field with a highly sensitive magnetic anomaly detector, and a barr function family used for barr conversion of the barr converter of this kind of conventional automatic magnetic anomaly detector. This is a depiction of the section. In the figure, (a) 1 is a magnetic field measurement signal obtained when the above-mentioned high-sensitivity magnetic anomaly detector is mounted on an aircraft to measure geomagnetic disturbances induced by a sunken ship or the like from above. (Hereinafter, this signal σ will be referred to as the magnetic field measurement signal SG) (+)) is the magnetic field measurement signal 5 Gft with a Sunnor period of τ8, and after A/D conversion, the 0th order will be held and the signal (hereinafter, this signal will be referred to as the ti field U) .

Since the value of the mouth changes every cycle T8, the symbol U(r) is used to represent the value at time γ/1 day. ), and the part BX shows the waveform induced by a shipwreck, etc. Note that the duration of the guided waveform is T. (C) Part of the Barr function family used in the Var transformation. The Barr function family consists of the following sequence of functions defined in the interval 0≦τ≦1, and generally continuous functions in the interval 0≦τ≦1, as well as step-like functions such as the alila boundary, are also defined by the Barr function family. I didn't know that it could be developed uniquely.

In other words, the number of samples N of the guided waveform in the part BX above is N
If the sampling period Ts is chosen to be the value shown by equation 21 so that = 2° (N l 11 t; j positive constant), the magnetic field Ts at N points in the past from the current time 2γ・1 day T/
(N 1 ) =T/ (2n-1) +21
0(r-N+1), mouth(r-N+2), U(
1-N+3).

...A function consisting of two units (r) can be uniquely expanded in the Barr function family as shown in equation (3) as a step function divided into 2n equal parts in the interval [0, 1].

21+1 01-ro(γ-N+1+i)-ri. Xo(2n+, )
(1=Ω+ 'r L '''l 2n' J ``2''
T 2+"'r n1j=1.2...2 to -1) (U of formula 51 (γ-N+l), mouth (γ-N+2),
U(γ-N+3),...Coefficient from mouth (γ). and h
(g) (k-+, 2.

"'t ns j"1+2+-=+2 to -1)
The k-calculated Melkot is called...-Le transformation, and the above coefficients obtained by...-Le transformation are ) and hV' (k=1, 2
, -・n, j-1, 2...2, -1) is called the rule spectrum. In this case, the above Norse vector h. , h(, l>(k=1.2.-,
n, j=i, 2. -, -, 2 to -1) 11 mouths (r-N+1+1) (i=0..1, 2.・2n-1)
It is mathematically known that it can be obtained by the following calculation steps.

ψ,"'=01=U(r-N+1+i) (i=0
．． 1.2 ,...,2n 1) -f41 body -92□
(t-1)+92m+1(t-0(Z"1 + 2 +
'-+nl)""0 *'H2+"'H21
) f5i i.e. (4+ (ψ♀ by 5+ formula)
, ψ (define ρ(61(712) hO, h'jj) (k=1.2..., n
, j=1, 2. ......, 2n-1) can be calculated (m+1) (k-1) (k-1)h
+ bi(92m −ψ2 ko, l/(2kp)(
k"" + 2. ",n+""G+1+,,2
n-1)-(61 h, =t ψ, (n +) 1 ψ(n >ゝ)/2
n - (7) Figure 2 shows the above spectrum h.

, hQ) (k=,.

2,..., n, j-1, 2,..., 2'')
= 3, that is, when the number of sample points N is N=2-8 points, hQ, h(2') (k=1.2.3) is applied to the magnetic field and expressed on the same time axis as in 1. It can be seen from FIG. 2 that the Baar spectrum captures the characteristics of the above-mentioned induced waveform well.

The above Baar spectrum ho, h42k” (h=1
+2,...,n) The features of the guided waveform extracted by
It can be further emphasized by the multiplied function, that is, the feature extraction signal Bs shown in equation (111).

S S = (ao-ho+a t ・h(I" + a
2-h":'10a5-h':'+--= 10a,,
- hL2rL112□ (8) Note that the coefficient ak (k:==Q, l, . . . n) tf is determined based on a large number of actual magnetic field measurement signals SG. Third
The figure shows the above-mentioned % feature extraction signal SS together with the magnetic field aperture, and it can be seen that the above-mentioned feature extraction signal 8B including J1 effectively extracts the feature g of the induced waveform.

If a threshold TH is provided in the feature extraction signal SS, it is possible to automatically determine the presence or absence of a waveform induced by a shipwreck or the like. However, in this method of setting the threshold in advance, the level of the magnetic field measurement signal changes depending on changes in the altitude of the aircraft.9 For example, when the level of the magnetic field measurement signal changes by twice, the feature extraction signal 8S
Although the threshold level is constant, the detection result may change if the % image extraction signal Bs and the threshold level are close to each other. .

In this invention, the dispersion VR of the magnetic field U (γ) in the past time indicated by the Vst91 formula is calculated, and this is multiplied by a coefficient K (positive constant), which is automatically set as the threshold A) 1), the above-mentioned drawbacks caused by conventional fixed thresholds are eliminated. However, the values of distributed VR and automatic setting threshold AH change every cycle T6, so when expressing the values at time γ and T8, VR(γ) and AH(
The symbol γ) shall be used.

VR(r)=-Σ(U(r-N-t)-σ(k))2
= ￥ 11(r+1-i)'L 11:1 -(zhi'1 u(γ+1-i)12 (L play 3 N
) −+911−1 AH(γ)=−VR(γ)□− However, UQ) U is the average value of the magnetic field opening for the past L −T8 hours expressed by the following equation.

+91.01 automatic setting thread hole)” AHi
j Since it changes in proportion to the square of the magnetic field opening, due to this property, when the level of the magnetic field measurement signal SO changes due to changes in the altitude of the aircraft, etc., the 1% characteristic extraction signal SS changes by a factor of the square. Even if it appears, the automatic setting threshold AH will also change by a factor of 2.

It is possible to make the detection result independent of the level of the magnetic field measurement signal SG.

FIG. 4 shows this situation. In the figure, (a) shows the magnetic field measurement signal SG, and (b) shows the relationship between the feature extraction signal 8B corresponding to the magnetic field measurement signal Sa of <) and the automatically set threshold AH. (C) U (a)
Feature extraction signal SS and automatic setting thread hold AH when the level of the magnetic field measurement signal becomes 1/2
Bxl(11) BI3. The BI3 portion is a waveform induced by a shipwreck or the like.

From now on, the level of the magnetic field measurement signal will be reduced to 1/2 as shown in (C).
When the level of the feature extraction signal SS becomes 1/41C at η, the conventional fixed threshold TH shown by the broken line
The detection result changes in the 1 dB I understand. In this way, by calculating the automatic setting threshold from the dispersion of the magnetic field measurement signal over a certain period of time in the past, the level of the magnetic field measurement signal can be changed automatically depending on the level of the magnetic field measurement signal. It is possible to detect guided waveforms of shipwrecks, etc. without being detected. The above is the principle according to this invention.

The automatic magnetic anomaly detector according to the present invention basically consists of five parts, namely an A/D converter, a bar conversion part, a separate part,
It consists of a synchronous pulse generator and a threshold hold calculation section.

(12) Figure 5 depicts a block diagram of the configuration of the automatic magnetic anomaly detector according to the present invention, and in Figure 2, 0) is A/'
D converter, (2I voltage conversion section, (31 is a synchronous pulse generator, (41 is a threshold calculation section, (5) is a discrimination section, and SG is a magnetic field measurement signal (analog) from a magnetic anomaly detector. Signal) One port is a signal obtained by digitizing the magnetic field measurement signal EIG using the A/D converter (1) and holding the sampled signal, and a synchronization pulse output from the pH 1-period pulse generator (31).

SS is the feature extraction signal calculated by the bar conversion section (2), AH is the automatically set threshold calculated by the threshold calculation section, and LT is the discrimination signal obtained by the discrimination section (5). .

The operation of each part will be explained below. FIG. 6 shows a synchronizing pulse P generated by the synchronizing pulse generator (3), which has a pulse width T. ij A/11: Set to be longer than the conversion time of converter (1) and to a value expressed by <9 return period Ts Fj equation (2). Figure 1 shows Aβ converter (1)
When the synchronizing pulse P shown in (a) rises from 0 to 1, the A/1) converter (1) starts converting the magnetic field measurement signal SG, which is an analog signal, into a digital quantity, and then T. The h/ry conversion is completed after a conversion time of , and when the synchronizing pulse P falls from 1 to 0, the magnetic field measurement signal 'fro' converted into a digital quantity is held and output as a magnetic field U.

In this way, the magnetic field measurement signal 5GiJ (
As shown in b), a digital amount of magnetic field Uvc& is converted every Sanhoor circumference M Ts and sampled and held.

FIG. 8 shows the case where n = 3, that is, the sample number N is 11 (=23) points, for the Baar conversion section (21).

The magnetic field output from the A/IJ converter (1) at the mouth,
Synchronization pulse generated by the synchronization pulse generator (3).

161 ij The magnetic field is converted to bar by the +61171 formula (m+1) bar spectrum hl), h (n==3
, k=1.

n-to-1-1 2,3m=0.1. ..., 2-1) with two var converters. 41 units, 149 units, 3. Lower 2 Fi The outputs of the Bahr transformer (61) are the Haars vectors ho, hf, It, h9', h (71
Signal corresponding to +71ij' above signal Uo4, ro1
42ro 33. Lower 2'f manually calculates the feature extraction signal shown in equation 81 (Barl spectrum feature extractor, S
S is the output of the Bahr spectral feature extractor (7).

FIG. 9 shows details of the burl converter (61) formed in the burl converter (+21') 41.

(This is shown for the case where the Sun-Noll number N is 2' = 8 (n = 3).) Exit ij in the figure: Magnetic field output from the A/D converter ill.

P is the synchronization pulse generated by the synchronization pulse generator (31, '01 + Mll +"' + "71 + Mn
2 + M12 + ”' + M72 +MO3, M
15, .=, M33 and Mn2. M14U storage device for storing intermediate calculation results, CMl, (3M2,0
M3 are constants 01 s 02 r 05 respectively
'fr Storage device/AD, i/AD21/”'
r AD41 r AD12-AD15U adder +
"Dll + 5D21 + "'+ 5D41
+5D12. 5D22, 5D13 ti subtractor, Dv, 1. Dv21゜'l ・-
, DV41 , DV 12 , DV
22, DVg3, Dvl3 U
The divider DLi is the subtracter sD1. and divider (1
5) A delay element that delays the synchronization pulse P by a time To (positive constant) that is greater than or equal to the sum of the calculation times of Dvll.

DL2, DL3 ij Delay elements Pl that delay the synchronizing pulse Pf by 2·To and 3·To, respectively.

P2. P3σ, respectively, the delay element DL4. DL2゜D
The output of L, U01 + 口+1 + ``' + 口7.The values 1'021012..., U72 are stored in the storage devices MO2, Mi2, -M72, respectively. 2 values. 6゜"13 + "25 + 35 mouths respectively storage device '05 HU3 rM73, MB
2 K stored values, Uo4r 014 b
Each storage device BM. The value stored in 4+M+4.

WO21W221 W421 "62 H Adders AD11+AD21, , AD3i, AD4.
output e w12 + WS2 + WS2 +W72
are the subtractors 5Di1.5D21 and 5D31゜5, respectively.
D4i output + wo3. w25. ij each D
Outputs of daily calculators AD12 and AD22, W13. W33
are the outputs of subtracters 5D12 and 5D22, respectively, and WO4
tj' output of adder AD15, W14 is subtracter 5D13
The output of AWj2°Aw32. AW52, A
w72 is a divider (16) respectively, DVll, DV21, DV51. , DV4
1 output p Aw1'5+AW3xU respectively divider D V 12 , D V 22 f) Output AWO
4, AW14 u each divider D”05 * Dv
This is the output of l5.

In this configuration, each time the synchronization pulse P is generated from the synchronization pulse generator (3), the memory device '11 + M2 is
1,... 2M storage device M01 with the contents
+ "11 + "' + M611' The magnetic field openings are newly stored in the storage device M71 by sequentially moving in this order.

In this way, the storage device "01 + Mll + "'
+ Value stored in '' U01.U111..., U
71 (91° of formula 41) = 01 (i = 0.1, 2.・
..., 1) (however, n-3) is supported.

The above storage device MOI, M21, M41, M
The price points [11+U21 +0.12lo61] stored in 61 are respectively added by adders AD11, AD2i,
The storage device tMt1 is connected to the storage device tMt1 by AD31 and AD41.
．． M month, MB1, price 11 stored in M71
, U31, U51, U71, and as a result, the signals WQ2, W22, shown in formula 09 are obtained.
W42 and W62 are calculated, but each (
5) ψΔ1′−ψQQ) shown in the formula.

ψ (111-95°′ ten lies 1. ψ so) - to) + ψ (8)
, ψ(11-lie)+ψ≦0). (However, n
=3. t=i) Also, the storage device M old r M21
, M41 , the value stored in M61 , 112
B, 12 ports 41+U6' are each subtractor SD1.
, 8D21 , 5D31 , 5D41 K Yacht above memory device * Mll + '51. MB1, M71
Prices stored in 11+ U31+ o5i 1 U7
1 is subtracted and the result is the signal “2+w” shown in formula 02.
32 + WS2 + w72 is calculated.

Furthermore, the above signals W12, WS2, WS2,
W72 is divided by 1DVr+, DV21,
DVSl, DV41 Vc, constant C+ (=4
) and the result is the signal AW12, A shown in equation 0.
W32, AW52, and AW72 are calculated.

This is (Haarsbek shown in formula 61) hT − (inkstone 0) − ψ) 0ノ) /4, h! 21- (ψ(2
)−ψi0) )/4 , h'7' −(ψ(0
)−ψ(F) ) / 4, h! 41−(ψ(0)
−ψ(?))/4 are effectively the respective signals AW12
, AW32, AW52, and AW72. (However, n-3°k-1). The signal wo2 thus calculated.

W22 + “2 r W62 and Aw12. AW
32, AW521AW72a Each time the pulse delayed by To than the pulse P, that is, the pulse P1 becomes 1, the memory device MO2+ M22 r M42 r ”6
2 and M122M62I' M52 + M7
The value is written to 2. Next, storage devices MO2 and M4
The price stored in 2 is 02, 口. 2 is (1
9) The value stored in the memory device M22゜M62 vc by the adder A I+ 2, A U22 0222 62
As a result, the signal wo31 shown in equation 041 is
W25 is calculated.

This is shown in equation (5) ψW゛−ψen)+ψ11′, ψ1
21-ψSo10ψii+ VC compatible. (However, n
=3. t=2) Also, the above storage device MO2+ M42
The values Uo2 and U42 tri stored in the respective subtracters G U12 and S U221Cj are the values U22° stored in the storage devices M22 and M62, respectively.
2 is subtracted and the resultant signal W13-RO02-guchi22-W
O2-W22, WB2 = U42-U62 =W
42-W62 is calculated, and the above signal W1s,
WB2 u each divider D■12. Dv22 is divided by Nyotte constant 02 (-47”i) and the result (1
A signal AWi3°AW33 shown in equation 9 is calculated.

In addition to this, (Barr spectrum h2-(ψ) shown in equation 61.

−ψ! It) /4σ, hso-(ψψ-even')/*
5 (20) indicates that the signals AWi3 and AJ3 are respectively calculated dynamically. (However, n=3.=2) Signal W calculated in this way. 5 + W23 and AW13. Every time a pulse delayed by 2·To from the AWs3U pulse P, that is, a pulse P2, becomes 1, a value is written to the storage device 31 M2S and M13 + M3S, respectively. In addition, the above storage device MO5+ M23
The value stored in [Jo3, 口, 3 is adder A I) 1
3 and the resulting signal WO4 - 113 + 23 = "05 + W26 is calculated incorrectly, and furthermore, the signal Wo4 is changed to a constant 03 (=8) by the divider DVo3.
As a result, the signal A Wo 4 shown in the curve is calculated.

AW, :4-W04/8=(Wo5+W23)/8-(
W (+2 +W22 +W42 +W62) / 8: (B 1] 1 + B 11 + B 21 + B 31 + Mouth 41 + Mouth 5. Ten Ro 61 + Roa 1 ) / B
□Koreha is shown in equation (7), that is, 2
The bar spectrum h (, shown in the following equation) is effectively the signal AW
This indicates that it was calculated in Q4. (However, n=3.
ni=3) ho-(ψ舒-1) 191(5-1))/2-3-(
ψWl + ψ1f21 ) /B − (21)
,,.

((ψ. 1ψ,)+(T2 +93))/8−((ψ
. . + ψ mark)) + (T2゜ + ψ)) 10 (Tachibana ゛) + ψ Gourd 0) ) + (ψ circle 0) + ψ (,o) Month/8 Also, the price 06 stored in the above storage device is The subtracter 5D13 subtracts the value fL stored in the storage device M25, and the resulting signal w14 [13-[+23-WO3=
W23 is calculated, and the signal W14 is sent to the divider D.
V15 is divided by a constant 03 (=8), and as a result, a signal AW14 shown in the form of a plane is calculated.

AW14=W14/8=(Wo3-W23)/8'
-11'n This is (h(1) shown in equation 61 = (Deaf 2
1-ψ%2')/a is effectively calculated as the signal AW14.

A pulse delayed by 3·TQ from w, 4 and AW14 pulses P calculated in this way, that is, pulse P3
each time becomes 1, the storage devices i1Mo4 and M,
The value is written to 4. However, pulse delay time 3・To
It is assumed that the value is set to a value sufficiently smaller than the repetition period T8 of the nivalus P.

As a result, after the pulse P3 becomes 1, the signal 112+port 322 port 52. Harsbek by Roa 2 respectively)
Le h4'ゝ, h'7', h4'', hp'ka, M number [T13, U33 (22) give bar spectra H411 and 41, respectively.

According to the signal UTIIVC, the /%-lunu vector hPI is
The signal UO4 means that the Burr spectrum ho is effectively equal to one, and as a result, the (61 (shown in equation 71) - Lus vector has been calculated.

FIG. 10 shows details of the Barr spectral feature extractor (7) that constitutes the Barr transformer f21fm. (This is shown for the case where the number of sample points N is 2-3 (n=3).) In the figure, UO4, 0+4, mouth 331 U72 B
A storage device M for each of the Var converters (6). 4 +
'14 rM 2 Signal output from M72, M
YI, MX2゜MX3. MX4. MX5 multiplier,
cy4. CN3°cy3. CN4ij is a storage device storing constants aO+al+a2rasf, Al)D is an adder, ZA1 is the output of the above adder, and 88 is a feature extraction signal obtained as the output of the above multiplier MY5. In such a configuration, the signal U
(+4, mouth 14.U331072 each multiplier M
Y1. MX2, MX3. The above storage devices (,lYl, CN2, CN3,
Constant aOr a1+ a stored in CN4 (23) f+, -C
:2+a3 are respectively multiplied and the signal Jl z2
1z31 z4, and the memo signal z1. Z2
．． Z5. z4ij are added by the adder ADD to form a signal ZA. Sand multiplier MY+,,・MX
4. Storage device tCY,...-(2Y4 and adder A
DD calculates the calculation shown in the formula 2.

ZA=ao −U04+a+ ・mouth i4 + a2,
Port 35 +a3 ・Lower 2-Q) 1 Furthermore, the above signal ZA
is squared by the multiplier MY5, and as a result, a signal 88 shown in equation +11 is calculated.

s 8=ZA=:iag・U04+a1・U14+
a2・J3+a3・lower 2) -us This is the above signal [
J! 14 r ro 149 ro 33 and U72 respectively have bar spectra hO2h (111 1, b441
This effectively means that the feature extraction signal SS shown in equation 81 is calculated for the case where n=3.

In the end, the Nord-Var converter (2) configured as shown in σ゛) converts the input magnetic field into the Var converter (61K) to obtain a power of 2 signal components, that is, (24) (m+1 ) fil +7+ Barr spectrum h01 shown in formula
-1-l (n=3. k=1.2, 3.
m=0.1. −． 2-1) is calculated, and among the Baar spectra, hO+h4Iゝh17+, h
The signal 102'4r133+lower2 corresponding to I3+1 is further sent to the barl spectrum feature extractor (7), where the feature extraction signal SS shown in equation (8) is n
= 3.

In the end, in this way, the barr transformation section (21 performs barr transformation of the magnetic field U and calculates the barr spectrum shown in equation 71, and then obtains the value of the -rus vector., h!2' ) (k=1.2.3) is linearly combined and further multiplied by 2 to finally calculate the feature extraction signal BS shown in equation (81) for n=3.

Although n = 3 is used in this explanation to simplify the explanation, it is actually desirable to use an integer of 4 or more as ijn, and in this case, even if n increases, the calculation process can be achieved in exactly the same way.

FIG. 11 shows details of the threshold calculation section (4). In the figure, the magnetic field output from the U ij A/1 converter (11), the pH synchronized pulse generator (3
) generated from synchronization pulses MD+, MD2+・
..., MDL-1 is a storage device that stores intermediate calculation results.

OB, 0Cij are constants, storage device storing K, AX, AYi adder, 5UBij subtractor, MPo.

MPl, , MPLi, MB, MO multiplier, DD,
1) E is a divider UUl, UO3, , , , U[
JL-1H storage devices MDI, MD2.・、M
Dl,, value stored in i + v+ ”'
* ■2 + ”'r vI-1 are each multiplier M
PO, MP, , , , Output of MPL-1.

SX and SY are the outputs of Calo calculators AX and AY, respectively.

SMX, AVFi ij each divider DD
, DB output, SMYi multiplier MB deba, VRi
j subtractor subtractor S high B, AH is the output of multiplier MO.

In such a configuration, the synchronization pulse generator (31) generates a memory device MDL2, M every time a synchronization pulse P is generated.
DL5. ...The contents of MDl are stored in the adjacent storage device MDL-
1, MDL 2 t..., MB2 in this order.

The magnetic field ports are sequentially transferred and the magnetic field ports are newly stored in the storage device MDi. The signal port and the storage device ffiMD1. MD2゜"'
+ 1uU1. stored in MDL 1. Rolo 2.
...UUL-tFi multipliers MPo and Mp, respectively
1,"'+MPL-1. The above squared signals V, Vl, V2-, Vt,-1
It is added by the 1ltBD calculator AX and becomes the signal SX,
Furthermore, the signal SX is stored in the memory r in the divider DD.
It is divided by the value stored in ItCB to obtain signal SMX. In other words, the multiplier M”Or MPl * ”' +
MPL 1 r adder AX and divider 9 divider DD [Thus, the calculation shown in the inner formula is performed.

1 2 2 2 sMx=sx/L=-(V 1'V1+V2+...+
VL-'1) = 1 (Ro2+Rosho 10... +1JU
L-1) - This indicates that the first term of equation (9), which is effectively the equation for calculating the distributed VR, is calculated as the signal SMX.

10,000, above signal l2 [Jl,..., UULl
are added by the adder AY to form the signal BY, and the above signal SY is divided by the value stored in the memory CB in the divider DE to form the signal AVE. The signal S is squared and multiplied by
Become MY. In other words, the adder AY, the divider DE, and the multiplier MB perform the calculation shown in equation c111.

SMY=SY2-1-! -(010 []110 units 12+
+00L-1) j -f2]IL.

This is the second equation of equation (9), which is effectively the calculation equation for distributed VR.
This indicates that the term was calculated as the signal SMY.

Furthermore, the value of the above SMY is subtracted in the above signal SMX [subtractor SUB, and the result is a variance VR expressed by the equation ω.
is calculated.

VR=SMX'-8MY 222 m-(mouth +roro, +uro, +... 10 mouth L-1)-
(τ(U+UUi 10U:!10,,, +UUL-1
)) −122 Effectively, the variance V shown in equation (9) is added to this.
The signal VR indicates that R has been calculated, and the signal VR is calculated by the constant K stored in the storage Co in the multiplier MO.
is multiplied by , and the automatic setting threshold AH shown in the Kan formula is finally calculated.

FIG. 12 shows the configuration of the discriminator (5).

In the figure, the SS value conversion unit (feature extraction signal obtained in 21, AHi; input from the thread hold calculation unit (28), automatic setting threshold, BBij subtracter
If the output of HzH subtractor SB, RLij input signal is positive, 1. This is a discrimination signal output from the relay LTitTi-RL which generates a logic output that becomes 0 if it is negative or 0. In such a configuration, first, the automatically set threshold AHO value is subtracted from the feature extraction signal SS by the subtracter SB, and a signal mz for comparing the magnitudes of both is calculated, that is, 1CZ=S8-AH, and the 9th relay The sign of the signal BZ is checked by RL. In other words, if the feature extraction signal SS is greater than the automatic setting threshold AH, the signal FiZ becomes positive, and as a result, the output of the relay RL is a logic 1 signal, and conversely, if the signal SE3 is less than or equal to AH. Then, 11gz becomes negative or 0, and as a result, a logic O signal is obtained at the output of relay RL.

In other words, L indicates whether or not the feature extraction signal SS exceeds the automatically set threshold AH using a logic signal of 1 and 0.
T times signal is obtained.

In the end, the automatic magnetic anomaly detector configured as described above digitizes the magnetic field measurement signal ij output from the high-sensitivity magnetic anomaly detector mounted on the aircraft by the A/D converter ill, and converts the digitized magnetic field measurement signal into a digital signal. Based on the bar conversion unit (21), a feature extraction signal is extracted that extracts the features of the waveform induced by a shipwreck, etc.

In addition, the threshold calculation unit (41) requires a threshold whose threshold level automatically changes depending on the level of the magnetic field measurement signal, that is, an automatically set threshold.

Next, the determination unit (5) compares the magnitude of the feature extraction signal and the automatically set threshold, and if the feature extraction signal exceeds the level of the automatically set threshold, the magnetic field measurement signal is a waveform induced by a shipwreck, etc. If the logic output 1 determines that the magnetic field measurement signal is noise, the logic output 0 that determines that the magnetic field measurement signal is noise is output as the discrimination signal LT.

By driving, for example, a hood, a ramp, etc. with the determination signal LT obtained in this manner, the existence of a sunken ship or the like can be automatically detected. In this explanation, we will explain the case where an analog magnetic field measurement signal is input, but when inputting a digital magnetic field measurement signal, it is also possible to input it directly to the var converter without going through an A/D converter. good. The above description is one embodiment of the present invention, and various modifications may be made without departing from the gist of the present invention.

Although the above description has been about automating the detection of magnetic anomalies for searching for underwater shipwrecks using a highly sensitive magnetic anomaly detector mounted on an aircraft, this invention is not limited to this. It can also be used to automate the measurement of the number of cars, etc.

As described above, the automatic magnetic anomaly detector according to the present invention can automate the process of detecting magnetic anomalies caused by the target object 1, such as a sunken ship, without being affected by the magnetic field measurement level.
In addition to eliminating the disadvantages of variations in detection ability due to individual differences among operators and non-uniformity in detection ability due to fatigue, the system also allows operators to set thresholds each time the level of the a-field measurement signal changes. If you fix it, it will be bugged (3)
1) It has the advantage of eliminating harshness.

[Brief explanation of the drawing]

Figure 1 shows an example of the Fi magnetic field measurement signal and a part of the Barr function family used for the Barr transformation. Figure 2 shows how the Barr spectrum extracts the characteristics of waveforms induced by shipwrecks etc. according to the respective spectra. Figure 3 is a diagram showing the feature extraction signal along with the magnetic field. Figure 4 is a diagram showing how the level of the feature extraction signal and the automatic setting threshold changes depending on the level change of the magnetic field measurement signal. Figure 5 is a block diagram showing the configuration of the automatic magnetic anomaly detector according to the present invention, Figure 6 is a diagram showing the synchronization pulse P, and Figure 1 is the A/D.
Fig. 8 shows the structure of the L-nole converter, Fig. 9 shows the var converter constituting the L-nole converter, and Fig. 10 shows how the converter holds the signal. FIG. 11 is a diagram showing a Barr spectral feature extractor constituting the Barr converting section, FIG. 11 is a diagram showing a threshold calculation section, and FIG. 12 is a diagram showing a discriminating section. In the figure, i11 is a φ converter, and 121t is an i-bar converter. (32) (3+ Ia r HA pulse generation 4e ((41ij threshold calculation unit, (5) is a discrimination unit, (6) is a Var converter, (7) ← Var spectrum feature extractor,
MOl, Mll, -=. M/1, Mll2. M+2. -, M72. M.O.
3, M16. M23゜M33, Mo4. M14
, MDl, MD2. -, MDL-1ijid storage t, cMi, 0M2. CM,, cyi
, ay2゜CM3. CM4. CB, 00 Storage device storing u constants, ADll, AD21.・・・
AD41. AD12°AD22. AD, 3. ADD
, AI, AY are adders. 5D11, 5D21, -3D41, 5DI2
, 5D22 *5Di3. BUB, 5Bij subtractor +
Dvll, Dv21. DV31. DV41. DV12. DV22.
DVO5, DV15゜DD, DB divider, MYl
, MY2. MY5゜MY4. MY5. MPa, MPl,
, ,MPL-1,MB. MO multiplier, RL#-t relay. Note that the same or corresponding parts in the figures are indicated by the same reference numerals. Agent Makoto Kuzuno - Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 'ITs-@l

Claims (1)

[Claims] (A var conversion unit that converts the magnetic field measurement signal measured by the 111ia abnormality detector into var, a synchronous pulse generator that controls the var conversion, and a threshold for the output of the var conversion unit. and a discrimination unit that determines the presence or absence of magnetic anomaly caused by a specific target object by comparing with Magnetic anomaly automatic detection that creates a feature extraction signal based on an appropriate linear combination, and compares the feature extraction signal with a threshold in the discriminator to automatically detect the presence of a magnetic anomaly caused by a specific target object. The device is equipped with a threshold calculation unit that calculates the dispersion of the IA field measurement signal over a certain period of time in the past, and the above dispersion multiplied by a real number is automatically set as the threshold, that is, threshold 1 is set depending on the level of the magnetic field measurement signal. As a threshold for automatically changing the novel, the discrimination section compares the feature extraction signal with an automatically set threshold to detect the presence of magnetic anomaly caused by a specific target object, regardless of the level of the magnetic field measurement signal. An automatic magnetic anomaly detector whose main feature is fully automatic detection. An automatic magnetic anomaly detector according to claim 1, wherein a magnetic anomaly caused by a specific target object is effectively extracted by correlating a magnetic field measurement signal with a constant sine cosine wave.