FR2764749A1 - PROCEDURE AND DEVICE FOR IMPROVING THE SIGNAL / NOISE RATIO, IN PARTICULAR TO OBTAIN FIRE INFORMATION FROM THE SIGNAL FLOODED IN THE NOISE OF A ROCKET AMMUNITION - Google Patents

Abstract

The invention relates to a method and a device for improving the signal-to-noise ratio, in particular for obtaining firing information (3) from a detection signal embedded in the noise (4) of a munition ( 2) sensor-guided, to be calculated so that the firing command (3) is triggered with the greatest possible probability exclusively by a target signal (6) contained in the detection signal (4). For this, before the digital filtering adapted to the statistical behavior of the target signal (6), a spectral leveling and the amplitude of the noise signal (5) accompanying the target signal (6) in the detection signal (4) are carried out. by means of a digital prediction filter (15) the coefficients of which are adjusted, by means of a transformation calculator (10), according to the auto-correlation function of the noise signal (5) captured instantaneously from the target area so that the series of coefficients of the matched digital filter (13) correlate as far as possible only with the target signal (6) contained in the detection signal (4).

Description

"Method and device for improving the signal-to-noise ratio, in

  particular for obtaining firing information from the signal embedded in the noise of a homing rocket "The present invention relates to a method of improving the signal-to-noise ratio to separate a useful signal from information input drowned in noise, by adaptive filtering, in particular to obtain firing information from a detection signal drowned in the noise of a research rocket ammunition as well

  than a device for implementing this method.

  We know, for example by the German patent

  28 33 590, adaptive filtering to eliminate the compound

  noise health of useful information, in the field of adjustment of the reproduction in phono montages. We

  find other areas where you have to

  separate meaningful useful information separately

  tive noise superimposed by adaptive filtering, for example, in the case of monitoring and exploitation of the beating of the heart of a living being. In the present case, we are particularly interested in obtaining firing information from the detection signal embedded in the noise of ammunition guided by a sensor, therefore with a research rocket, whose charge d attack breaking the armor

  must be initiated, and should not be initiated until, when,

  that its active, semi-active or passive detector, reacting in particular to radiant energy

  in the infrared or microwave field,

  tect at least once, by exploring the zone of targets by bands, a real target to be effectively attacked, and recognized it as such, in a radiating environment in the same spectral domain, but with a

  characteristic of different amplitude.

  A munition provided with a detector functioning in this way is described in more detail, for example in German patent application P 3333517.6 or in the

  German patent application P 33 45 601.1 of the Applicant-

  se; with regard to the equipment of such ammunition with detectors, reference is made to German applications P 34 10 942.0 and P 33 45 529.5 of the Applicant, and with regard to the possibilities of processing the signals

  such detectors, at request P 332519.8 of the Request-

  resse, to avoid to this extent repetitions in

the present case.

  The object of the invention is to improve measurements of this kind so as to be able to obtain, from a signal

  strongly drowned in noise, whose behavior we know

  statistically of its useful signal component, in improvement

  rant signal-to-noise ratio, a well-defined useful signal suitable for triggering other functions, such as

  in particular to deliver firing information.

  To achieve this objective, according to the invention, a reduction filtering is carried out with the digital filtering means optimized on the statistical behavior of the detection signal consisting of a signal corresponding to the noise and of a signal corresponding to the target. level the noise signal and, taking into account the reductive filtering, an adaptive filtering according to the statistical behavior of the predetermined target signal, and, on the other

  part, the detection signal including the variation of the ampli

  study as a function of time is sampled is sent to a digital uniformization filter after which is mounted a digital filter of coefficients adapted to the statistical behavior of the target signal contained in the detection signal, taking account of the characteristic

  filtering of the leveling filter.

  The solution according to the invention therefore uses, in principle, a digital filter, such as that described, for example, for a specific application case, in DE-A 20 25 098. The filter is suitable there for like an equalization filter, with the known statistical behavior of the useful signal. A prediction filter, such as that described, for another special application, in DE-A 114 078, is used to transform the limited band noise distributed along a Causs curve into white noise; optimal correlation conditions are thus obtained for the useful signal in the filter which is adapted to it. Regarding the dimensioning of such a prediction filter, we can refer to section 5.3.3. from the book "Grundlagen der Theorie statistischer Signale" (foundations of the theory of statistics of

  signals) by Eberhard Hansler and Hans Marko (Editions Sprin-

ger, 1.983).

  The invention makes it possible not only to extract from the noise signal from the environment of the target a signal corresponding to the target without ambiguity, but also

  adapt the useful signal extraction to instantaneous data

  tanned by noise, therefore from the ground, that we encounter in

  the environment of the interesting target.

  We obtain a possibility of differentiation between the useful signals of true targets and those of

  misleading or illusory targets, by providing (after the ob-

  attention of the surrounding noise signal without correlation (white) to a large extent) of the filters adapted to target signals of variable widths, so that their comparison of amplitude with the useful signal is successful

  directly to a real target detection signal.

  Because due to the operation of the digital filter

  adapted, it appears exactly at the end of the monitoring

  nance of a target signal and, consequently, with a time shift defined with respect to the start of the predetermined and sought target signal, it is possible, with a derivation, adjusted correlatively by construction, of the location direction of the relative sensors

  to the direction of action of the munition, to deliver the information

  firing training in precise time relationship

  with the scanning of the center of the real target.

  Other features, variants and advantages

  of the invention will emerge from the detailed description

  which will follow from a preferred embodiment of the invention

  tion shown in the single figure of the accompanying drawing in the form of a block diagram limited to the essentials. The detector 1, with which a research rocket ammunition is equipped 2, must deliver, as a function of the detection of an actual target (not shown in the drawing), firing information 3 to initiate a charge

  directed combat on said target.

  Detector 1 works for example, passive-

  ment as a radiometric reception device which, when the mnution 2 enters the zone of a target, scans the latter by its reception characteristic and searches for a target to attack. The real target, for example an armored vehicle, is characterized, in this case, by a reflection or masking effect determined with respect to the environment, for microwave radiation coming from space and penetrating into the area of the target and reflecting on it towards detector 1. Signal 4 of detector includes

  therefore essentially, in the case of such a radiometer detector

  1, a typical noise signal 5 for the reflective properties of the environment, therefore of the area of the

  target, exhibiting amplitude variations at high fre

  quence which are superimposed on the noise of the electronic circuit.

  At this noise signal 5, it is superimposed, when the character-

  detector reception teristics briefly and additively scans a real target, a signal specific to the target or target signal 6 which differs by its sound

  statistical behavior (as indicated symbolically-

  ment on the drawing by the addition operator 7. In the case of a passive thermal opto-electronic detector (1), it should be assumed that this superposition is substitutive, since the signals coming from the area of the target are less powerful, and of lower frequency than the electronic noise and are distinguished less by their characteristic of amplitude, as regards

  real targets and false targets.

  In order to be able to carry out the detection of the target to deliver the firing information 3 with the means of the digital signal processing and filtering technique, an analog-digital converter 8 is mounted following the detector 1. It samples (explore point by point) the detection signal 4 by observing Shannon's sampling theorem, to transform its instantaneous amplitude values into coded information

  digitally for further signal processing.

  It therefore also determines the rate 9 of the digital filtering,

so discreet.

  While the target signal 6, as well as its statistical behavior, for a typical real target, which is the only interesting one, is already known taking into account the sampling data of the reception characteristic of the detector), it is appropriate, to obtain a possibility of multiple use of the invention, to determine each time separately the statistical behavior data of the noise signal 5, therefore the character of the target area in the environment of the real target sought. For this, for example, in what is called a "transformation calculator", a Fourier or fast correlation analysis of the detection signal 4 is carried out, with data to which the instantaneous non-detection of a real target by the detector 1. Such data appear, for example, when the ammunition 2 is still at a very great distance from the zone of the target, that, as a result, a target which is possibly detected there is very small , and that, in the detection signal 4, the component of the target signal 6 is therefore negligible compared to the noise component 5. Thus, the transformation computer 10 supplies the digital coefficients 11 for the optimal adaptation of the digital filters (13, 15, 22) in order to eliminate the noise signal 5 with respect to the target signal 6 in the detection signal 4, which we will come back to with

more details.

  In the example shown, for reasons of clarity, a filter for eliminating separate continuous signals 12 has been shown to eliminate the continuous components of the detection signal. In the practice of switchgear, its function is however already provided, in general, before the analog-digital converter 8 and carried out in particular, for example, by the high-pass properties of the temperature stabilization of the sensor produced.

in detector 1.

  The improvement of the signal-to-noise ratio, therefore of the ratio between the amplitudes of the target signal 6 and the noise signal 5 in the detection signal 4, is essentially carried out in a suitable digital filter, called "matched" 13. Its response impulse, determined by its coefficients of the filter 13, must, as is known, present as precisely as possible, the values of the curve of symmetrical amplitude with respect to the time of the useful signal, therefore of the target signal 6, explored, applied to the corresponding cadence of rhythm 9, to obtain an optimum correlation result (equalization effect, in the sense of eliminating the components of the noise signal). Since the statistical behavior of the target signal 6 for a determined real target is already known by measurements or calculations, the transmission computer 10 can provide these coefficients 11 on the basis of the known correlations of the statistical processing of the signals (taking account transfer functions

  filters that are in series with it.

  The filter 13 set to the symmetrical function of the target signal 6 to be waited for operates optimally only when the components of the noise signal in the detection signal 4 have as low a correlation as possible, moreover, with the specific target signal 6 , as well as between them. For this, the matched filter 13 is preceded by a filter 14 called here "uniformization filter". It essentially comprises a prediction filter 15. Its coefficients 11 are calculated, according to the instantaneous statistical behavior of the signal

  noise (essentially band noise limited to distri-

  Gaussian bution), which is again analyzed by the cal-

  transformation culateur 10, to transform the composition

  noise health of the filter input signal 16 into a noise with the lowest possible correlation, therefore white. In this context, it must be taken into account that an elimination of the continuous components has been carried out (for example, by a filter 12), or that a continuous component

  signal 4 (via the covarian-

  ce), must be incorporated into the pre-

  diction 15. To eliminate correlated fractions from the signal

  of real noise 5 in detector 4, the result of extra-

  polation or prediction of the prediction filter 15 is still subtracted, in a subtracting member 18, from its input information 17, which gives, by leveling of amplitude, a reduction in the noise component, hardly more correlated then, of the input signal 16 of the matched filter 13. As a result, the filter 13 operates optimally as an equalization filter (with an impulse response which represents the symmetry of the target signal 6 in the signal

  input 16); thus, only the useful components correlated

  The lines, therefore dependent on the target, of the input signal 16 provide a contribution to the operating signal 19 of an improved signal / noise ratio compared to the signal of

detection 4.

  This operating signal 19, which certainly presents

  a pace with unstable oscillations, presenting an ampli

  higher study relatively narrow in the presence of a

  general target 6 in detection signal 4, suitable

  during good, due to its relative signal-to-noise ratio

  tively very good, for an adaptive stage 20 with threshold with peak detector29. Indeed, the good signal / noise ratio of the operating signal 19 makes it possible to set the threshold

  central online at a high level, resulting in direct-

  a reduction in the rate of false alarms (delivery of firing information 3 as soon as an apparent target Misleading neck is detected), for the same detection power of real targets. Indeed, the peak detector 29 constantly adapts the height of the precisely active threshold to the short-term average value of the signal 19, so as not to provide,

  wherever possible, that useful signals 21 cor-

  responding to targets_ Depending on the scenario taking place in the target area, it may however happen that the detector 1 also delivers target signals 6 which do not originate

  not a real target, but false targets (trom-

  few or apparent). They are due, for example, to bodies of water, or to reflective plates intended for camouflage which produce, for example, an effect of

  marking or reflection for microwave energy inci-

  tooth similar to that of a real target interesting to attack. Therefore, it has been found that in practice an essential characteristic of differentiation between the signals 6 of real targets and false targets consists in that, in the latter case, the target signal 6 is clearly shorter or more long only when a real

  target is detected by detector 1.

  In order to be able, in the operating signals 19, to distinguish between signals coming from real targets and signals coming from false targets, the signal 16 supplied by the leveling filter 14 is also applied to at least one other filter. adapted digital 22. The choice of the values of its coefficients 11 is basically the same as for the aforementioned adapted filter 13, but the step-by-step processing now extends, due to the shorter or longer target signal 6

  in the case of a typical false target, to a correct number

  laterally higher or lower of cycles 9. The signal 23 appearing previously is therefore, as a comparison signal 25, transferred into a delay circuit 24, for example a shift register progressing at rhythm 9 and, consequently, memorized, up to the other signal, for example the useful signal 21, appears. For false

  longer targets, signal 21 must therefore be memorized.

  This delay buffer is generally chosen to be long enough so that typical false targets can be recognized

  clearly. When care has been taken, when calculating the coef-

  11 of the matched filters 13 and 22 that their operating signals 19, respectively 23, have the same Àc0 energy for their power density spectra, although the filters 13 and 22 operate with a different number of cycles 9, the signal 25 and the maximum of the useful signal 21 can be sent directly, after a corresponding delay, to the comparison inputs of a

amplitude comparator 26.

  As for an instantaneous target signal 6 (which can come from a real target, but also from a false target) the useful signal 21 filtered on the length of the signal of the real target brings a higher amplitude than the comparison signal 25 coming from the filter 22 not optimized on the real target with regard to its processing rate (but adjusted, for example, to a lower value), the amplitude comparator 26 provides a target detection signal 27 when and only when the detection signal 4 carries a signal

real target.

  Its time of appearance is offset in time by as many cycles 9 as the filter 13 needs for the digital (sequential) filtering process with respect to the effective start of the target signal 6 in its input signal 16 Because, according to the location derivation, the delivery of the firing information 3 of duration increased compared to the exploration of the target area, is predetermined, the delivery of the firing information fire 3 can take place, for the

  typical system sampling process, statistics

  only precisely when detecting the center of the real target. In fact, the delay resulting from the digital filtering chain (therefore known in advance) can be reduced in a prior offset arrangement 28, exactly half of this delay, which means that the firing information 3 corresponding to the real target is delivered, according to the known diversion imposed by the system, during the exploration of the target area, precisely during

  detecting the center of an actual target to be attacked.

  A When the localization bypass imposed by the system is sufficiently large, this is not only the case for short false target signals 6, but also for signals (not shown in the drawing) longer than the actual target signals 6 It is then sufficient, as has been indicated, to take care to correlatively delay the maximum of the signal 21, that is to say that the peak detectors 29 and 24 must both be dimensioned.

  over the maximum foreseeable operating time interval.

REX37END I C A T I ONS

  1. Method for improving the signal ratio

  noise to separate a useful signal from input information, embedded in the noise, by adaptive filtering, in particular to obtain firing information from a detection signal embedded in the noise of homing rocket , characterized in that one performs with the means

  digital filtering optimized on statistic behavior

  detection signal tick consisting of a corresponding signal

  reducing to noise and a signal corresponding to the target, a reduction filtering to level the noise signal and, taking into account the reduction filtering, an adaptive filtering

  according to the statistical behavior of the predefined target signal

completed.

  2. Method according to claim 1, character-

  laughed in that, to carry out the reductive filtering, one

  subtracts the sampled detection signal from each other

  lonné and a prediction signal obtained from it,

  whose coefficients are determined by the self-function

correlation of the noise signal.

  3. Method according to any one of the claims.

  cations 1 and 2, characterized in that two adaptive filtering operations are carried out regulated on different lengths of target signals, for similar energy contents of the power density spectra of the useful signals which they

  provide, which is subjected to an amplitude comparison.

  4. Method according to claim 3, characterized in that the useful signal of the filter adapted to a real target is subjected, before the amplitude comparison, a detection of peaks and it is passed through a stage at

an adapted threshold.

  5. Device for improving the signal / noise ratio for separating a useful signal (21) from input information embedded in the noise by means of adaptive filters (15,13,22) in particular for obtaining information of firing (3) from a detection signal (4) of a research rocket ammunition (2), drowned in noise for the implementation of the

  process according to any one of the claims, characterized

  laughed in that the detection signal (4) whose variation

  amplitude as a function of time is sampled is

  seen in a uniformization filter (14) after which is mounted a digital filter (13) of coefficients (11) adapted to the statistical behavior of the target signal (6) contained in the detection signal (4) taking into account the characteristic of the leveling filter (14) .15 6. Device according to claim 5, characterized in that the leveling filter (14) comprises at its output a subtracting element (18) to which its input signal (17)

  is brought on the one hand directly and, on the other hand, by the in-

  through a prediction filter (15) whose coefficients

  cients (11) are predetermined according to the function of autocorré-

  lation of the band-limited noise signal (5) contained in the

detection signal (4).

  7. Device according to any one of the claims

  and 6, characterized in that from the filtering standardization

  (14), several suitable filters (13,22) are supplied with pa-

  aligners, filters among which one is calculated on a

  general target (6) shorter and / or longer than the first.

  8. Device according to claim 7, characterized in that delay assemblies (24) are mounted following the

  suitable filters (22) calculated at shorter durations.

  9. Device according to any one of the res-

  dications 5 to 8, characterized in that an adaptive threshold stage (20) is mounted following the first adapted filter (13). 10. Device according to any one of claims 7, 8 and 9, characterized in that an amplitude comparator (26) is mounted following the adapted filters

(13, 22).

  11. Device according to any one of the res-

  dications 5 to 10, characterized in that it includes an assembly

prior shift (28).