DE3817627A1 - Direction finding arrangement for passive acoustic location of target - evaluates transition time by passing electrical signals of acoustic sensors via filters, sampling circuits, multiplexer - Google Patents

Abstract

An arrangement for the passive location of targets using transition time evaluation feeds the electrical signals of at least two acoustic sensors (H1,H2) to filter banks (11,12) of defined bandwidth. The output signals of the filter banks are simultaneously sampled by sample-and-hold circuits (21,22). The sampled values are digitised by a multiplexer (3) and digital-to-analogue circuit (4) and fed to a microcomputer (50-56) for further processing and evaluation. USE/ADVANTAGE - E.g. for use in sea-bed mines. Achieves better directional and positional accuracy at min. cost.

Description

The invention relates to an arrangement for passive acoustic Locating targets using runtime evaluation.

A passive acoustic location of targets z. B. by electronic facilities held in seabed mines by Runtime evaluation of the noise emitted by a ship, in particular that of the engine room, the direction and the ent Determine the distance of a passing ship and at Vor predetermined criteria trigger the mine to ignite. To do this, the sounds of at least two underwater transform microphones (hydrophones, sensors) into electrical signals delt and evaluated.  

In the SABI process, the hydrophone signals are converted according to their ver strengthening filtered and threshold detectors supplied. About the signal of the one hydrophone goes a predetermined one Threshold, a counter is started. Exceeds the signal of the other hydrophone the predetermined threshold, so the Counter stopped. The difference in the meter readings corresponds to the Running time. However, this procedure only provides undisturbed, usable results for noisy signals.

In the phase sum difference method, two hydrophone Signals the square of the sum signal and the time derivative device of the difference signal is formed. As a measure of the signal If the direction (time difference of the signals) becomes the quotient from the averaged sum signal square and the averaged pro product of sum signal and difference signal change value used. This measure makes the quotient obtained independent of the present noise level. However, there is a non-linear re dependency between the determined quotient and the run time difference, which leads to considerable errors in linearization pretations due to noise signals. The evaluation is also frequency dependent.

The invention is based, the state of the art to improve. In particular, an arrangement of the upper Concept of claim 1 mentioned type the DF and Or tion accuracy can be improved, with the least possible agile and inexpensive solution is desired.

The task is in an arrangement of the in the preamble of Pa Tent claims 1 type mentioned in the characterizing part mentioned features solved. It is now possible, even with ge disturbed signals to perform an almost error-free location, so that the efficiency, especially of seabed mines, increase Lich is improved.  

Advantageous further developments and refinements of the invention are specified in the subclaims. The characteristics of the claims che 4 and 5 can be particularly space-saving with a microphone realize the calculator.

The invention will now be illustrated with reference to one in the figures Embodiment explained in more detail. The individual shows:

Fig. 1 block diagram of the arrangement according to the invention;

Fig. 2 block diagram of a matrix filter.

The arrangement for passive location of targets by means of runtime evaluation is a microcomputer-based system to optimize flexibility and complex signal processing. It consists of at least two acoustic sensors. In the preferred embodiment shown in Fig. 1 for sea mines z. B. four sensors H 1 to H 4 (hydrophones) are used. The sensors follow in the signal flow filter banks 11 to 14 with a second order pass characteristic from 100 to 500 Hz. The filter banks contain amplifiers (not shown) in FIG. 1, which ensure interference-free signal processing. Between the output of the filter bank 11 and the multiplexer 3 , a power detector 10 is connected for power control.

Via so-called sample-and-hold switches 21 to 24 , signal samples are tapped from all sensor channels simultaneously. With the help of an analog multiplexer 3 and an A / D converter 4, the signals are digitized, which are supplied to the microcomputer 50 to 56 as reference values of the analog function in the form of dual numbers equidistantly in the example 0.5 ms grid.

Instead of the filter banks 11 to 14 shown in FIG. 1, it is advantageous to use a matrix filter shown in FIG. 2. This avoids phase errors of the filters without special adjustment of the filters.

When matrix filter according to Fig be. 2, the sum and difference signals respectively of two opposite or substantially opposite sensors such. B. the sensors S 1 and S 2 , the inputs of the filters 111 and 121 supplied.

By summing and subtracting the output signals of the fil ter 111 and 121 , the sensor signals S 1 and S 2 are obtained accordingly, filtered sensor signals S 1 'and S 2 '. The sensor signals S 1 'and S 2 ' are fed to the sample-and-hold scarf terns 21 and 22 via amplifiers 112 and 122, respectively, which can be controlled in the amplification. As already explained in relation to FIG. 1, these are connected to the multiplexer 3 . The amplifiers have the same control characteristics and can be controlled, for example, by utilizing the output signal of the multiplexer 3 such that the A / D converter 4 always works in the optimum range.

The microcomputer 50 to 56 connected to the output of the A / D converter 4 has three buses 50 , an 8-bit data bus, a 16-bit address bus and a 4-bit control bus. The following function groups are connected to this bus structure:

one CPU 55 (Central Processor Unit)
a ROM 53 (Read Only Memory)
a RAM 52 (Random Access Memory = direct access memory)
a real time clock 54 (timer)
a digital signal processor 51
an input / output unit 56 (input / output ports)

At the output of the unit 56 z. B. the ignition amplifier 7 of the sea mine.

The programs for signal evaluation are located in the memory 53 (ROM). The memory 52 (RAM) is used to store variables. The entire signal processing is monitored, controlled and evaluated by the central unit 55 . Particularly quickly through feeding signal processes, such as. B. the real-time processing of digital filter algorithms or data reduction processes are done by means of the signal processor 51 .

All function groups mentioned consist of highly integrated, commercially available components. The digital signal processor is a 1- Chip microcomputer that meets the requirements of digital Signal processing in the low frequency range is tailored. The processor is therefore particularly suitable for calculation rules that repeat at the beginning of each sampling period, i.e. H. which are completely processed in one sampling period ("Sample-by-sample" processing). This calculator initially runs with the digitized real-time signals of the Hydrophone Fast Fourier transforms (FFT) with 128 points and is enough the spectra to the actual computer. It follows through Zeroing of all spectral components outside 100 to 500 Hz steep post-filtering. With the help of the differences of the nor mated phase spectra and conversion arise for each of the two pairs of signals of the orthogonal hydrophone arrangements the barrel time differences. The runtime differences thus obtained determine using complex filters, predictor and trajectory tracking algorithms with the help of the measured water depth the optima len ignition timing.

The goal is determined by evaluating runtime differences determined. Be between direction cosine and term differences has the following relationship:

τ _L = τ _maxcos cos (1)

Here τ _{L is} the bearing angle-dependent transit time and τ _max the maximum transit time, which occurs with an axis-parallel target direction and is determined from the speed of sound c in the water and the basic distance D of the hydrophones.

The transit time can easily be determined with the aid of the Fourier transform. The following applies:
If

so also applies

This follows by substitution of (t-τ _L ) = t ′.

This means:

S (ω, τ _L ) = | S (ω, τ _L ) | _e i Φ (ω, τ _L )

= | S (ω, O) | _e i · (Φ (ω, O) -ωτ _L ) (5)

It follows:

Φ (ω, τ _L ) = Φ (ω, O) - τ _L · ω (6)

Φ (ω, O) = Φ (ω, τ _L ) = ω · τ _L (7)

Equations 6 and 7 state that the phase angle difference between the undelayed signal and the delayed signal is equal to the product of angular frequency and transit time. If one plots the phase angle difference as a function of the angular frequency, then a straight line through the origin with the slope τ _{L must} result.

The Fourier transformation is approximated using the "Discrete Fourier Transform" (DFT) performed.

The DFT can be a Fourier transform of the sequence of regular gene pulse samples of S (t) can be viewed. The DFT is an overlay of an infinite number of periodically shifted Fourier transforms.

To overlap between the different frequency periods too Avoid limiting the spectrum of the signal to be sampled.

A widening occurs due to the finite observation period of the spectrum. The finite observation period can be seen in Time range through a rectangular window function with which the signal is multiplied, describe. After the folding The DFT theorem is a multiplication in the time range of a case equivalent in the frequency domain.

The spectrum of a rectangular window is wide. It is therefore advantageous to use window functions, which in Frequency range have a large sideline suppression, to overlay different spectral lines due to the To avoid folding. Particularly advantageous with regard to rakes The Blackman window function has time and memory requirements proven. It has the following weighting functions:  

This window function has a secondary maximum suppression of approx. 58 dB compared to a maximum rejection of approx 12 dB for the rectangular window function.

The Fourier transformation is carried out using the FET algorithm, for example in the signal processor S 1 .

According to equation (7) in the image area of the DFT there is between the Pha send difference between two different signals and the circular fre quenz a linear relationship.

Assuming white, normally distributed discrete The straight line with the most probable slope will be noisy the pattern function determined.

The probability density for a point on the Δϕ-ω plane calculates itself

Since the measured values from different points in time are decorrelated, be the composite density sits the following representation

The most likely runtime is obtained when P _N becomes maximum; equivalent to this is the requirement that the exponent be minimal. The minimum is obtained by zeroing the derivative of the exponent after the term.

The equivalent is the statement:

The most likely runtime is given by equations (13) or (14) given.

Equation (14) applies provided that the spread the bearing values are the same for all frequencies. Under the hypo thesis that the scatter of the measured values is inversely proportional to the spectral power of the signal at the corresponding Fre is an improvement in the estimation accuracy achieve as follows.  

A _n is the spectral power.

If it is known that interference occurs in a certain frequency range, these interference can be suppressed by not taking them into account when calculating the transit time. This can be expressed as follows using the masking function µ _n

A _n is the spectral power
Δϕ _n is the phase difference
ω _n is the angular frequency
µ _n is the masking function
µ _n = 1 for the undisturbed area, otherwise = 0

To calculate the bearing values, the following operations are carried out in the microcomputer 51 to 55 :

- Reading in the digitized samples
- Windowing the samples
- Determination of the cross power spectrum using a fast Fourier transformation (FFT) (data is reduced by the number of FFT support points)
- Determination of the phase differences from the cross power spectrum
- Weighting and masking of the phase difference values
- Determining the terms
- Determination of the direction cosine values and thus the direction of the signals picked up by the sensors.

Knowing is used to calculate the horizontal distance between mine and target The depth of the mine location in relation to the surface of the sea necessary. The depth measurement is carried out by measurements of the hydro static pressure.

The direction finder delivers the direction cosine values of the two orthogonal DF axes.

Between the direction cosine and the Cartesian vector components The following relationship exists

_x and _{y are} the unit vectors in the x and y directions and the distance vector.

x, y are the x component, the y component of distance and H the mine depth. If you add the direction cosine values geometrically, it follows:

and thus:

The distance values determined from the direction cosines fluctuate around the true value. For the distance estimation is the Least-square polynomial approximation of the distance values in particular suitable.

Assuming an unaccelerated movement of the The target is obtained for the horizontal distance square fol Development over time:

R² = (x ₀ + v _x · t) ² + (y ₀ + v _y · t) ² (21)

After multiplying and ordering according to the powers of t there is the following development over time:

R² = (x ₀ ² + y ₀ ²) + 2 · (x ₀ · v _x + y ₀ · v _y ) · t + (v _x ² + v _y ²) · t² (22)

As you can see, a 2nd order polynomial can be used to full describe constantly.

To calculate the development coefficients is the following Solve minimization problem:  

In equation (23) the origin of the time axis is placed at the time of the last direction bearing. This representation has the advantage that the estimated value of the distance is only efficiently represented by the coefficient a ₀ . To solve the above minimization task, the development coefficients have to satisfy the following equation system.

The system of equations is solved using the Gaussian elimination method. The coefficient a ₀ represents the estimated value of the horizontal distance.

Claims (6)

1. Arrangement for the passive acoustic location of targets by means of runtime evaluation, characterized by the following features:

• - The electrical signals (S 1 , S 2 ) from at least two acoustic sensors (H 1 , H 2 ) filter banks ( 11, 12 ) are fed certain same bandwidth,
• - The output signals of the filter banks are sampled simultaneously using sample and hold switches ( 21, 22 ),
• - The samples are digitized by means of a multiplexer ( 3 ) and a D / A converter ( 4 ) and fed to a microcomputer ( 50-56 ) for further processing and evaluation ( Fig. 1).

2. Arrangement according to claim 1, characterized in that the filter banks are matrix filters, to which the sum and difference signals of two opposite or largely opposite sensors are supplied and with their output signals by summing and forming the sensor signals corresponding filtered sensor signals (S 1 ', S 2 ') who won ( Fig. 2). 3. Arrangement according to claim 2, characterized in that the filtered sensor signals (S 1 ', S 2 ') via adjustable LF amplifier ( 112, 122 ) the sample-and-hold switches ( 21, 22 ) are supplied. 4. Arrangement according to one of the preceding claims, characterized in
that the digitized samples are first subjected to windowing during further processing,
that the cross-power spectrum of the sensor channels is determined using the windowed values by means of a Fourier transformation, that the phase differences of the sensor channels are determined from the cross-power spectrum,
that the phase difference values are weighted and masked,
that the transit times of the sensor signals are determined with the weighted and masked phase difference values and
that the direction of the signals picked up by the sensors is determined from the transit times. 5. Arrangement according to claim 4, characterized in that at their use in seabed mines using the stagnant the between mine and water surface and the direction of that of signals picked up by the sensors the distance of the target is determined. 6. Arrangement according to claim 4, characterized in that for Windowing the Blackman window function is used.