GB2267755A

GB2267755A - Passive accoustic range finding method

Info

Publication number: GB2267755A
Application number: GB8820419A
Authority: GB
Inventors: Michel Rivet
Original assignee: Thomson CSF SA
Current assignee: Thales SA
Priority date: 1987-09-01
Filing date: 1988-08-26
Publication date: 1993-12-15
Anticipated expiration: 2008-08-26
Also published as: IT1241398B; IT8867682A0; GB2267755B; GB8820419D0; DE3829532A1

Abstract

An acoustic range finding system has three aligned sensors (A, B, C) and operates by correlating the signals received by these sensors and uses two additional sensors (A1, C1) the signals of which are correlated with those of a central sensor (B). The comparison between these correlations permits to eliminate the ambiguities produced when acoustic pulses in a narrow frequency range are received, which in particular permits one to determine the position of naval vessels and weapons by means of their own sonar transmissions. <IMAGE>

Description

2267755 PASSIVE ACOUSTIC RANGE FINDING METHOD The present invention

relates to the passive acoustic range finding methods which permit, by measuring by means of at least three aligned sensors, the curvature of the wave front of a signal formed, for example, by pulses of a frequency between 1 and 50 kHz, transmitted by a source, the determination of the direction and distance from this source.

Already known systems use, as shown in figure 1, 3 aligned sensors, 1, 2 and 3 a distance L apart. A signal source E is at a distance D from the central sensor 2, on a bearing from the perpen dicular to the alignment of the 3 sensors of (D 0.

The wave front which arrives at the base formed by the 3 sensors is, for practical purposes, circular and the moments of arrival at each of the sensors are different. Thus, in the direction considered here, the wave front arrives at sensor 1 first, then sensor 2 with a delay Of T 12 with reference to sensor 1 and finally at sensor 3 with a delay Of T 23 with reference to sensor 2. These two time of arrival differences, at the sensors, permit to obtain E) 0 and D according to the formula:

G = arc sine CC 12 + T 23)l 0 2L and L cos D 0 c 23 T 12) where c is the speed of acoustic waves in the ambience in question, generally water To ine 1 asure - 12 and T 23' we carry out, in 2 processing circuits 10 and 11, the intercorrelation of the signals from sensors 1 and 2 on one hand 2 and 3 on the other hand. The positions of the maxima of the 2 intercorrelation functions provide the readings T 12 and T23 This processing is generally carried out digitally, as is the calculation of () 0 and D in circuit 27.

When the transmetter E produces a noise, the noise of propellers for PxpMnle the f-r-eqitenQy r-aUe is wide and the 2 intereorrelation functions provide a single maximum reading.

There is therefore no possible ambiguity in determining the distance from-the noise source.

When the transmetter E produces acoustic pulses, those of the sonar system of a boat or a torpedo for example, the fre quency band is narrow and the intercorrelation functions have as many maxima as there are in phase signal receptions at the pairs of sensors.

As shown in figure 2, a pulse B, of frequency f arrives, in the case of this figure, first at sensor 2 and then at sensor 1 after having covered an additional distance LsinG; 0 corresponding to a delay Of T = L sinE) 0 c The signal phase lag at these sensors is therefore 2 f representinf a series of successive maxima where + 2k.,T, that is to say T = + k, k being a whole number.

7 As distance L is greater than the wave length X c/f to have me,:asurable delays, finding the maxima of the 2 intercorrela tion functions gives rise to an ambiguity that corresponds to an uncertainty of + k/f, where f is the central frequency of the pulse.

This ambiguity increases with a reduction of l/f, that is to say an increase in f. - According to the known technique, this ambiguity is is eliminated by measuring, roughly, the distance. This rough estimation is carried out by measuring the 3 moments of arrival at the 3 sensors. To do this one detects the rising front of the pulse as it exceeds a predetermined level. This technique is sometimes referred to as "thresholding". However, such techniques require rising fronts to be steep and this implies the existence of good rectangular pulses not distorted ones. This does not correspond to true operational conditions in that the pulses may frequently be smoothed out, for example into gaussien form, and the ambience upsets the rising front to a considerable extent even when the signal to noise ratio is good.

3 An idea of the problem can be gained if we consider that when length L is 100 m and distance D 10 km, the difference between the times of arrival at one of the 2 extreme sensors and the central sensor equals 165 microseconds. It can therefore be seen that one must be capable of determining the position of the rising front of the pulses to the nearest 10 microseconds, something which is very difficult to obtain especially when dealing with smoothed out pulses.

To remove the ambiguity from such a system, the invention proposes correlating the signal from the central sensor with those from 2 other sensors slightly offset with reference to the outer sensors.

One thus obtains 2 distinct series of correlation maxima and removes the ambiguity by taking those maxima of the 2 series which coincide with one another.

Thus, according to the present -invention there is provided a method of passive acoustic range finding wherein three sensors receive acoustic pulses in a narrow frequency range transmitted by a source at a distance D on a bearing eo, wherein the first correlations between the signal of one of these sensors are made to obtain a first sequence of delays only one of which corresponds to the required reading, wherein in addition the said pulses are received by two additional sensors each one a short distance from the other two sensors and a second series of correlations are obtained between the signal from the first sensor and those of the other two additional sensors to obtain a second sequence of delays only one of which corresponds to the reading required, and wherein the first and the second sequences are compared to determine the points at which the values coincide. e 1 1 Other/ ea ures and advantages of the invention will become apparent during the following description which is given merely by way of example and refers to the two figures which represent: - in the case of figure 1, the diagram of a system set up according to the known method. 5 - figure 2, a pulse of frequency f exciting sensors 1 and 2 in figure 1. figure 3. the diagram of a system set up for operation of a method according to the invention, - figure 4, a comparison between the two series of maxima determined by the system shown in figure 3.

Figure 3 is an overall diagram of a range finding system that applies the principle of the invention. There is a support, for-example a naval vessel or a submarine, 3 panels, Pl. P2 and P 3 themselves carrying 3 sensors A, B and C. These panels are mounted on each side of the vessel and secured so that the misalignment is as small as possible.

The signals from these antenna are processed in correlators 11 and 21 to obtain 2 intercorrela tion functions CAB(T) and CCB(-r)' If fo is the operating frequency obtained, for example, by filtering the signals produced by the antenna around frequency fo.

1 these 2 functions will show maxima the periods of which in T equal i/f 0 as already described: the position of the maximum of C AB (T) is given by:

L sinE) + L 2 cos 2 0 + m (where m is a whole number) C 0 U_c 0 7 whereas the position of the maxima of C CB (T) is given by:

L sinE) = L 2 cos 2 0 + n (where n is a whole number) C 0 U_c 0 0 The position difference is therefore also a sequence of periodic values 1 such as:

= 2 2 - k T1 L cos 0 0 + 1 (1) (k 1 whole number) Dc f 0 The "true" maximum corresponds to the exact value of D which is that which corresponds to ki = 0.

The calculations required to obtain this sequence of values T 1 are carried out in circuit 13.

Other sensors A 1 and C 1 are mounted on panels P 1 and P 3 They are identical to sensors A and C. These 2 sensors A 1 and C 1 are mounted at the same vertical height in line with the 3 sensors A, B and C but are offset, horizontally, by a length 1 so that they constitute, with sensor B, a second measurement base at a distance L 1 between the different sensors. In the case shown in the figure L 1 < L but it could just as well be L 1 > L.

The panel alignment errors are the same for both measurement bases. So no additional errors caused by misalignment are added.

To be able to mount both sensors on one panel, 1 is much smaller than L and it will also be noted that L =cc L 1 and this makes a + 1 L 1 In the same way as before, the signals from sensors A13 B, C 1 are processed in 2 correlators 12 and 22 to obtain 2 intercorrelation functions C A1B (T) and C CIB (T). The position of the maxima of these functions is obtained from the same expressions, replacing L by L The difference in the positions of the maxima of the 2 functions is therefore a sequence of periodic values such as:

T' = L 2 cos 2 E) + k 2 (where k is a whole number) 1 0 - - 2 uc- f 0 This sequence is also obtained by means of circuit 13.

One therefore multiplies the sequence of values of T' by cc 2 to obtain a sequence of values T 2 such as:

T = L 2 cos 2 E) 2 k (2) 2 0 a 2 Dc f 0 The coincidence of k 1 k 2 = o permits to obtain a distance D by means of:

is D = L cos0 C(T 0 1 k i The other points of coincidence occur at:

k 1 = a 2 k 2 # o. 2 As can be seen above cc, and theref ore also a is slightly higher than 1. An one can therefore writes:

a 2 = 1 + F_ where small E is less than 1.

The condition for the values of the two sequences Tly T 2 to coincide is therefore written as K 1 = k 2 (1 +E:). These coinci dences set up ambiguities but there are fewer of them than when you consider only a single sequence.

One determines therefore the coincidence in a circuit 15 by measuring the differences separating the values of the 2 sequences -1 1 and T 2 and by determining the pairs of values at which the differences are zero or are at least within a band that allows for measuring errors. In other words, one uses only those values from the sequence T 1 that are nearest to any given value of sequence T 2 The angle S 0 is measured, incidentally, in a system 16.

This measurement can be taken with a separate sonar antenna or by taking "thresholding" readings on the rising fronts of the pulses, this procedure being less critical for the measurement of E) 0 than it is for D.

If one therefore has 0 and the values of that corres pond to the coincidences, one can calculate, in a circuit 17, the corresponding distance values by means of:

D = L cos E) 0 CT 1 To obtain the actual distance, one takes a rough estimate of this distance, for example by "thresholding" the pulse fronts, in system 18. This permits to select the correct value of D from those already obtained by making a comparison, in circuit 19i between this estimated value and the values provided by circuit 17.

To do this, the distances D corresponding to the remai ning ambiguities must differ, clearly, from the estimated distance.

This is obtained by taking out a sufficiently large number of ambiguiies between 2 cons ecutive ambiguities.

Between each coincidence of the maxima of the 2 sequences T131 T one has 1 = 1 maxima of T that correspond to the same 2 1 2_ number of ambiguities that we shall call k A. The smaller E therefore, the larger the number of ambiguities taken. As c cannot be infini tely small, because it cannot be zero, the distances obtained between 2 reMaining ambiguities must be sufficiently different for allowing to eliminate those that are not plausible.

To be able to take k A ambiguities, it must be possible to measure the difference between, for example, the two positions k 1 = 0, k 2 = 0 and k 1 = 1, k 2 = 1. One has shown, diagramatically on figure 4, the case of 2 series of maxima T 1 and - 2 for which the first non zero coincidence occurs when k 1 = 5 and k 2 4. For T 1 the difference between two maxima is 1 and for T 2 it is 2.

0 0 Each value of T 1 and T 2 is obtained with a given degree of measurement error. The necessary sufficient condition for k A ambiguities to be taken is therefore for the difference in the values Of T 1 and T 2 when k 1 = k 2 = 1 to be such that when the errors are removed the remaining difference will be at least equal to that obtained when the errors added up for k 1 = k 2 = 0. This condition gives 1/k A f o > 4AT if it is assumed that the measurement errors are equal on both sequences.

The maximum operating frequency of the system is therefore given by:

f 1 0 max 4k A LS T The number of ambiguities to be taken at k A is determined so that between the true distance (k 2 = 0) and the first remaining ambiguous distance (k 2 = k A) there is such a difference that this last figure would not be plausible.

The system is completely defined by determining L 1 as follows:

L 1 = L: k A 1/2 k A + 1 d As an example, a range finding system permits to determine 4 ambiguities, that is to say k A 4. One obtains:

01 2 = 5 and L 1 L 112 4 (1,125) Thus where L = 25 m for example, L 1 = 22,36 m, that is to say 1 = 2,64 m.

The maximum frequency f o max can be obtained by estimating the measurement error.ST. This error comes only from errors caused by noise and the lack of accuracy of the calculations in that the 2 sensors Al. C 1 are affected in an identical way by the errors caused by misalignment in that they are mounted on the same panels as sensors A and C.

ot The erros caused by noise are estimated by the CRAMER-RAO point. This point is provided, approximately, by the expression:

AT R 2,/2-- f 0 where R is the signal to noise ratio.

If AT c is the error caused by the innaccuracy of calcula tion AT= AT R + ATCY and the maximum frequence at which the system can be used is given by:

1 V2k A f o max 4 k A AT On sonar interception, the signal to noise ratio is in the region of 20 dB, that is to say R = 10. If one takesAtc 1 us, one obtains for k A = 4, f o max = 27 kHz.

The first residual ambiguity is to be found at a value T = 5/fo of the true value, that is to say, for 27 kHz, at microseconds. For a distance of 10 km, the true delay is 42 microseconds where L = 25 m for 0 0 = 0. With a delay of 227 microseconds, the distance obtained is 1,8 km something which is not plausible. - Furthermore, the error on the direction 0- 0 must not give rise to too great a distance error.

As direction 0 is known to within.!'- the error on the distance D AD. The relative error on this distance is therefore such that:

IA Pj = 2 tgE) 0 A0E 0 D When 0 0 is obtained by "thresholding" the pulses, the relative error on distance remains small. For pulses that have a steep front, AE) 0 is less than a degree for G7 0 s of between -45 and to +45, that is to say AD/D < 3%. For smoother pulses, experience has shown that 0- 0 is in the region of + 2,5', that is to say AD/D = 8,5%, an error which is still tolerable.

Claims

1. A method of passive acoustic range finding wherein three sensors receive acoustic pulses in a narrow frequency range transmitted by a source at a distance D on a bearingGop wherein thefirst correlations between the signal of one of these sensors are made to obtain a first sequence of delays only one of which corresponds to the required reading, wherein in addition the said pulses are received by two additional sensors each one a short distance from the other two sensors and a second series of corre lations are obtained between the signal from the first sensor and those of the other two additional sensors to obtain a second sequence of delays only one of which corresponds to the reading required, and wherein thie first and the second sequences are compared to determine the points at which the values coincide.

2. A method according to the claim 1 wherein to deter is mine the points at which the values coincide we retain the values of the first sequence that are nearest to a given value on the second sequence, calculate the corresponding distances and select the true distance.

3. A method according to the claim 2 wherein the five sensors are aligned, the first sensor being in a central position, two other sensors are situated on either side of.the first one at a distance L.and the two additional sensors are situated on either side of the first one at a distance L 1 so that L-Li=l, the value of.1 being small when compared with L and L 1

4. A method according to cla-im 3 wherein ope subtracts the maxima of the first correlations to obtain the first sequence of delays, one subtracts the maxima of the second correlations to obtain another sequence of delays, one multiplies this other sequence by a coefficient a 2 = (1 + 1+1 1) 2 to obtain the second sequence of delays and one compares the first and second sequence to remove at least - 1 ambiguities.

a 2_ 1

5. A method according to claim 4 wherein one removes the remaining ambiguities by comparing the calculated values of the distances with a distance value estimated by another means.

6. A method according to claim 5, characterised by the fact that this value is estimated by thresholding the fronts of the pulses received at the sensors.

7. A method of passive acoustic range finding substantially as hereinbefore described with reference to figures 3 and 4 of the drawings.

7. A method of passive acoustic range finding substantially as hereinbefore described.

Amendments to the claims have been filed as follows A method of passive acoustic range finding wherein three sensors receive acoustic pulses in a narrow frequency range transmitted by a source at a distance D on a bearing 909 wherein first correlations between the signal of one of these three sensors and the signals of the two other sensors are made to obtain a first sequence of delays only one of which corresponds to a. 1 reading usable to determine the distance D, wherein in addition the said pulses are received by two additional sensors each one being located at a distance from the other two sensors which is small in relation to the distance between these other two sensors and a second series of correlations are obtained between the signal from the first sensor and those of the other two additional sensors to obtain a second sequence of delays only one of which corresponds to the usable reading and wherein the first and the second sequences are compared to determine the points at which the delays coincide.

2. A method according to the claim 1 wherein to determine the points at which the values coincide the values of the first sequence that are nearest to a given value on the second sequence are retained, the corresponding distances are calculated and the true distance is selected.

3. A method according to the claim 2 wherein the five sensors are aligned, the first sensor being in a central position, two other sensors are situated on either side of the first one at a distance L and the two additional sensors are situated on either side of the first one at a distance L so that L-L 4. A method according to claim 3 wherein one subtracts the maxima of the first correlations to obtain the first sequence of delays, one subtracts the maxima of the second correlations to obtain another sequence of delays, one multiplies this 2 other sequence by a coefficient..,, = (1 + 1/L 1)2 to obtain the second sequence of delays and one com ares the first and second sequence to remove at least ambiguities.

w,2_ 1 5. A method according to claim 4 wherein one removes the remaining ambiguities by comparing the calculated values of the distances with a distance value estimated by another means.

6. A method according to claim 5, characterised by the fact that this value is estimated by thresholding the fronts of-the pulses received at the sensors.