DE3446658C2 - - Google Patents

Description

The invention is based on an adaptive filter according to the preamble of claim 1.

Such filters are in the Naviga solution tion tasks required in order to make multiple pre lying measured values the position of a to be measured Object, for example a landmark or of a target vehicle. Un in particular Watercraft use for position determination the passive acoustic bearing of target vehicles from noise sources to your own location not to reveal active transmission signals; you are therefore rely heavily on such filters position coordinates with those from passive bearings can be determined.

It's from literature, Wade H. Foy. Position-Location Solutions by Taylor-Series Estimation, IEEE Trans. AES. No. 2, March 1976, pp. 187-194, an adaptive filter, i.e. H. a measuring device downstream signal processing system, known, with that from the strongly scattering measured values Initial values are determined using a known or determinable statistical security Positions of the measured objects are. By the The filter becomes the smoothing of scattering measured values, e.g. B. bearings, and the implementation of the measured values in states of the object, e.g. B. Position koor dinates, accomplished. These states are also called  denotes the components of a target vector.

A signal is used to implement this filter processing system in which a measured value estimation and the required filter coefficients from one estimated input vector, namely the target position coordinates, and the associated measurement position determined will. Through the signal processing system, which is a model for the implementation of the measured values in which corresponds to object states, the error between the measured value and the measured value estimation are minimized, so that the actual filter input value is the difference between measured value and measured value estimation with the the respective filter coefficient is multiplied. The sum of all multiplication results is the Value of the respective component of the output vector, in this case a position vector error vector to correct the input vector. The so certain output vector becomes component by component fed to a comparison device. Gives way to Output vector from a predetermined threshold, so becomes the one corrected by the output vector gangsvector the signal processing system in the Kind of an iteration loop fed again. The multiple iterative filtering of the measured values results then finally the target states, so here the Target position coordinates, with an error deviation, the components are smaller than the specified one Is threshold.

However, such a filter is for processing unsuitable from target states that moved for Yield goals. The other target dates, course and speed, are not taken into account in any form. The filter even fails after a few iterations  also when determining the target position coordinate because the error vector no longer converges.

Furthermore, in US Pat. No. 38 48 509 described a system used to control a cannon processed data of a target measured by a radar. With this system there is also a value for the Speed of the target is taken into account, but only around the hit point of the cannon fired Projectile aimed to determine. About the type and Are ways of speed measurement, except that they are according to the state of the art, no further Information included. Because the system of improvement Is used, the projectile is measured and Muzzle velocity and deviation in one Missed in a Kalman filter processed to the Correct control of the cannon. Such a system is not for adaptive measurement of target data set up.

The invention has for its object a filter of the type mentioned so that the target data with little circuitry of moving targets with as few iterations as possible certainly.

With a filter, this task is described in the preamble of claim 1 defined type according to the invention by those specified in the characterizing part of claim 1 Features solved.

This is the goal of the filter according to the invention vector memory expanded to include the movement descriptive of the target according to speed and course Record speed components. From the Target position coordinates at the start of the measurement, too referred to as static target position coordinates, the speed components and the respective Measurement time are by means of multipliers and adders so-called dynamic target position coordinates are formed, with which the coefficient calculator is controlled. From the first determined by the coefficient calculator Measurement coefficients, the so-called static part of the measurement coefficients, are in the downstream multiplier stages by multiplying it again with the Measurement time now further measurement coefficients, the so-called dynamic part of the measurement coefficients. In a computing device for matrix operations the conversion of the measurement coefficients into the required ones Filter coefficients with which then the target data  be determined.

The filter according to the invention has the advantage that to determine the dynamic part of the measurement coefficients the same circuits, computing devices, as for the static measurement coefficients to be used. There are only a few additional multipliers Multiplication levels necessary to the time dependence of the dynamic coefficients capture. For the downstream computing device, the accumulation circuit, the comparator etc. are considered according to the number of considered Vector components, however, multiplications the internal processing circuit necessary. At low processing time requirements therefore also processing circuits used in the multiplex, d. H. "Component filter", realizable.

Further advantageous embodiments of the invention result from the subclaims.

So it is particularly advantageous the invention to develop according to claim 3, because of the Determination of additional support coefficients with a support coefficient circuit that essentially the coefficient circuit to determine the measurement coefficient corresponds to the filter base values can be entered by observation or other measuring methods have been determined are. The support coefficients and the associated one Base value estimates are made from the vector components the input vector and additional if necessary Entry of basic values determined. The measurement coefficient matrix  H is additionally with the support coefficient expanded and as a result also one extended to control the accumulation circuit Filter coefficient matrix F is formed. The The consequence of this is that the filtering of the measured values including that in addition in the accumulation circuit processed difference values faster, d. H. with a significantly smaller number of iterating system runs, to the exact target data vector leads.

It is also advantageous, according to claim 6 Difference values and the base coefficient to evaluate with a weight factor. Quality Input of basic values, e.g. B. radar measurements considered with great weight, so that a quick The filter settles to the true values of the target data vector. Less good base values, e.g. B. inaccurate periscope observations Considered and changed with less weight then the transient response hardly. To this Wise is the filter of the respective measurement and observation situation optimally adaptable.

Such filters for determining target data by Passive bearing measurement filtering converge a stable result only when the measuring Vehicle itself a maneuver, the so-called self-maneuver, drove. Here is the consideration of base values for determining the extended Filter coefficient matrix F particularly advantageous, since the maneuver can be avoided, if very good measurements of the base values with high Weighting can be evaluated.  

The structure of the filter according to the invention points out addition the advantage that the consideration of Baseline values controlled differently for each iteration can be because in every iteration the filters coefficient matrix is recalculated.

The invention is based on one shown in the drawing Embodiment in more detail below described. It shows

Fig. 1 is a block diagram of an apparatus for determination of target data,

Fig. 2 is a block diagram of a coefficient circuit,

Fig. 3 is a block diagram of a distance computation circuit,

Fig. 4 is a block diagram of a tap coefficient circuit,

Figure 5 is a block diagram of a device for determining the target data of a moving target maneuver.,

Fig. 6 is a block diagram of a coefficient circuit for Meßkoeffizientenbestimmung during a maneuver moving target,

Fig. 7 is a representation of a target trajectory of a maneuver moving target.

Fig. 1 shows the block diagram of a device with a filter for determination of target data from measured values B _mi. The vector components _o , _o , _x , _{y of} this target data are stored as input vectors in a target vector memory 10 with memory elements 11, 12 for target position coordinates _o , _o and memory elements 13, 14 for speed components _X , _Y. A coefficient circuit 20 is connected downstream of the target vector memory 10 , to which an input memory 60 with its partial memories 61, 62 for measuring positions X _Ei , Y _Ei and a measuring time memory 70 are connected on the input side. The coefficient circuit 20 is connected to the transfer of measurement coefficients h _{i 1} , h _{i 2} , h _{i 3} , h _{i 4} to a computing device 80 and to transfer a measurement estimate _Mi to a differential unit 85 . The differential unit 85 is also connected on the input side to a measured value memory 86 , in which the bearings recorded by a direction finder 87 at the measuring times T _Mi are stored as measured values B _Mi.

The computing device 80 is designed such that the complete matrix H of the measurement coefficients h _{i 1} , h _{i 2} , h _{i 3} , h _{i 4 is} stored and into a matrix F of the filter coefficients f _{i 1} , f _{i 2} , f _{i 3} , f _{i 4 is} converted, with which an accumulation circuit 88 can be controlled.

The accumulation circuit 88 is also connected on the input side to the differential unit 85 , so that each filter coefficient f _{i 1} , f _{i 2} , f _{i 3} , f _{i 4 is} multiplied and summed up with the associated difference δ _Mi from the measured value B _Mi and the measured value estimate _Mi. and forms an output vector Δp of the accumulation circuit 88 . On the output side, the accumulation circuit 88 is connected to a comparison circuit 90 and, via an addition stage 94 connected to the target vector memory 10 , to a gate circuit 95 , via which the target vector memory 10 can be controlled with the input vector corrected by the output vector Δp of the accumulation circuit 88 . The comparison circuit 90 is connected on the output side to the gate circuit 95 and to an output interface 96 for outputting the input vector, ie the target data. The interconnection of the target vector memory 10 and the output interface 96 takes place via a connecting line 97 , to which the addition stage 94 also accesses.

However, the target vector memory 10 can be controlled not only via the gate circuit 95 with the input vector corrected by the output vector Δp, but also from a basic data memory 15 for vector components _o , _o , _x , _y .

When determining the target data, targets are assumed which move uniformly in sections. The destination therefore drives unaccelerated, ie at a constant speed, on a constant course. This constant speed and the course are determined by the filter, ie the filter coefficients f _{i 1} , f _{i 2} , f _{i 3} , f _{i 4,} which are dependent on the speed and time, are determined and the bearings are filtered in such a way that the target position is not reached coordinates _o , _o the speed components _x , _{y are also} output, from which current estimates for distance, course and speed of the target result.

For this purpose, the filter has a target vector memory 10 for the target data, in which, in addition to the static target position coordinates _o , _o , the speed components _x , _{y are also} stored. Since with passive locating devices for foreign targets, the real but only so-called estimated target data is never available as the target vector, the components are identified by an additional roof-shaped symbol (ˆ). The errors .DELTA.X, .DELTA.Y, .DELTA.V _x , .DELTA.V _y occurring then indicate the differences between the estimated and the actual target data.

The filter is the transmission system with which the vector of the target data _o , _o , _x , _y from a plurality of i = 1 to N measured bearings B _Mi

= ( _o , _o , _x , _y ) ′

is determined.

From the static target position coordinates _o , _o , the speed components _x , _y and all the stored measurement times T _Mi , at which the bearings B _{Mi have been} measured, component by elements are multiplied by multipliers 25.1, 25.2 and adders 26.1, 26.2 dynamic target position coordinates _i = _o + _x * T _Mi and _i = _o + _y * T _Mi generated and fed to a coefficient calculator 50 . The coefficient calculator 50 determines on the basis of the measurement geometry for a target position _i , _i , taking into account the associated stored measurement position _Ei , _Ei, a so-called measurement value estimate _Mi , ie the estimate of a bearing from the measurement position X _Ei , Y _Ei to the estimated one dynamic target position _i , _i .

The coefficient calculator 50 emulates a filter system in which from a single static target position, and the associated measurement position X _E , Y _E, the estimated bearings _M and first measurement coefficients h 1 and h 2 are determined.

If the measurement model implemented for static states in the above-mentioned manner is applied to the dynamic target position coordinates _i , _i , then the estimated bearing _{Mi obtained} from the estimated target states including the speed components is obtained

or with the abbreviations for the spacing _{components Xi} and _Yi

With

_yi = _o + _y * T _Mi -X _egg, (3.1)

_yi = _o + _y * T _Mi -Y _egg, (3.2)

² _i = ² _xi + ² _yi , (3.3)

where ² _{i represents} the distance square between the target position _i , _i and the measuring position X _Ei , Y _Ei .

With the first measurement coefficients h _{i 1} and h _{i 2} determined by the coefficient computer 50 on the basis of the dynamic target position _i , _i , the first elements of a sequence of measurement coefficients have already been determined, the number of elements of which depends on the number of components of the target vector. Further measurement coefficients h _{i 3} and h _{i 4} necessary for determining the complete target vector are additionally determined by multiplying the first measurement coefficients h _{i 1} and h _{i 2} by the measurement time T _Mi.

To prove these relationships, one of Eq. (1) corresponding relationship for the measured bearing B _Mi the actual target state vector by the estimated target state vector and an error vector Δp

Δp = (ΔX _o , ΔY _o , ΔV _x , ΔV _y ) ′ (4)

replaced. From the Taylor series development of this Relationship after the error vector Δp results by omitting the higher order terms linearized measurement model for the dynamic target states to

with the measurement coefficients

If one takes the measurement coefficients h _{i 1} ,. . . , h _{i 4} in a matrix H and the differences between the measured value B _Mi and the measured value estimate _Mi in the vector δ, the measurement equation gives

δ = HΔp (6)

the dependence of the bearing difference δ _Mi = B _Mi - _Mi on the error vector Δp and the measurement coefficient matrix H. The signal processing system to be realized results in a value of minimum variance for the error vector Δp if the error vector Δp is used as the output vector from the difference vector δ and a filter coefficient matrix F by the signal processing system

Δp = Fδ (7)

is determined and if for the filter coefficient matrix F

F = (H ′ · H) ^-1 · H ′ (8)

applies, where H 'is the transpose of the measurement coefficient matrix H is.

The entirety of all measurements can thus be determined by the matrix equation system, the so-called measurement equation, according to Eq. (6) can be evaluated. The signal processing system to be implemented gives an optimal estimate for the output vector Δp if it is calculated according to Eq. (7) and Eq. (8) is determined from the vector δ of the differences between measured values B _Mi and measured value estimates _Mi , the matrix F representing the filter coefficient matrix.

This filter coefficient matrix F according to Eq. (8) is determined in the computing device 80 , to which the entirety of the measurement coefficient sequences, the measurement coefficient matrix H, is supplied.

The computing device 80 is then connected to the Akkumula tion circuit 88 connected together δ in the component-wise differences _Mi between the measured value B _Mi and the Meßwertschätzung _Mi with the filter coefficient matrix F is multiplied, and are respectively summed over all the products, so that the output side to the accumulation circuit 88 of the From the starting vector Δp for the target data.

The output vector Δp is fed component by component to a comparison circuit 90 connected to the measurement time memory 70 , in which a vector standard of the output vector Δp is formed from the statistical and dynamic components and compared with a threshold. If the comparison yields an inadmissible deviation, an input vector corrected for the output vector Δp is transferred to the target vector memory 10 , and the stored signal values B _Mi and observation times T _Mi in iterations are used to determine improved output vectors Δp until the comparison conditions are met.

The input memory 60 provided in the device according to FIG. 1 also has partial memories 63, 64 for observation coordinates X _R , Y _{R of} an additional measuring position, a reference time memory 65 , a weight factor memory 66 , a reference value memory 67 and a control memory 68 . To control the individual storage units, the input memory 60 is connected to an input interface 69 , to the input side of which a position measuring device 72 , a speed measuring device 73 , a compass unit 74 and a data display device 75 are connected.

The device for determining target data is completed by a support coefficient circuit 120 for determining support coefficients h _{S 1} , h _{S 2} , h _{S 3} , h _{S 4} , which on the input side with the storage elements 11 to 14 of the target vector memory 10 , with the partial memories 63, 64 for observation coordinates X _R , Y _R with the time memory 65 , the Ge weight factor memory 66 and the control memory 68 is connected. On the output side, the support coefficient circuit 120 is connected to the computing device 80 for transferring the support coefficients h _{S 1} , h _{S 2} , h _{S 3} , h _{S 4} . The support coefficient circuit 120 also determines a support value estimate which, in connection with the support value input _S, results in a difference value δ _{S in} a differential circuit 185 connected downstream of the support coefficient circuit 120 and the support value memory 67 . The differential circuit 185 and the differential unit 85 can be connected alternately to the accumulation circuit 88 via a switch 181 which can be controlled by the control bus 78 , so that either the differences δ _Mi from measured values B _Mi and measured value estimate _Mi or the difference values δ _S from the base value estimate and base value input S in the accumulation circuit 88 are filtered by means of the filter coefficient matrix F and the output vector Δp is thus determined.

In a central control device 77 , the necessary clock signals are generated in particular for reading in and reading out the memories 10, 60, 70 ,, the coefficient circuits 20, 120 , the computing device 80 and the interfaces 69, 96 and transmitted via the control bus 78 .

Fig. 2 shows in detail the coefficient circuit 20 with inputs 21.1, 22.1 for target position coordinates _o , _o , inputs 21.2, 22.2 for speed components _x , _y , inputs 21.3, 22.3 for measuring position coordinates X _Ei , Y _Ei and a measuring times input 23 and with measuring coefficient outputs 36.1 to 35.4 for the measurement coefficients h _{i 1} to h _{i 4} and with an output 37 for the measurement value estimation _Mi. The coefficient circuit 20 has multipliers 25.1 and 25.2 , each of which is connected on the input side to the input 21.2 or 22.2 and both to the measurement time input 23 and which are each followed by an adder 26.1 or 26.2 on the output side, which is also connected on the input side to the input 21.1 or 22.1 is connected. The adders 26.1, 26.2 are connected on the output side to a coefficient computer 50 with target position coordinate inputs 51.1, 52.1 , which is also connected on the input side to the inputs 21.3 and 22.3 of the coefficient circuit 20 . The coefficient computer 50 is connected on the output side with the output 37 for the measurement value _Mi and on the one hand directly with the measurement coefficient outputs 36.1 and 36.2 and on the other hand via multiplier stages 39.3 and 39.4 with the measurement coefficient outputs 36.3 and 36.4 .

For the calculation of the measurement coefficients h _{i 1} and h _{i 2} and the measurement value estimate _Mi , the coefficient calculator 50 has a distance calculation circuit 40 which is connected to the adder stages 26.1 and 26.2 and to the inputs 21.3 and 22.3 of the coefficient circuit 20, at which the measurement position coordinates X _Ei , Y _{Queue egg is} connected. In addition, 50 divider circuits 54.1 to 54.3 are provided in the coefficient calculator, which are connected to the distance calculation circuit 40 with their outputs for distance components 46.1 and 46.2 and a distance square output 42 in such a way that the quotient of the distance component _xi at the dividing circuits 54.1 and 54.2, respectively or _yi and distance square ² _i and at the dividing circuit 25.3 the quotient of both distance _{components xi} and _{yi are} present. The divider circuit 54.3 is followed by an arctangent calculator 55 , which forms the measured value estimate _Mi from the quotient of the divider circuit 54.3 . The output signals of the arc tangent circuit 55 and the divider circuits 54.1 and 54.2 are identical to the output signals _Mi , h _{i 1} , h _{i 2 of} the coefficient calculator 50 .

The distance calculation circuit 40 shown in FIG. 3 has subtraction circuits 43.1 and 43.2 for each component of the input data, in which the measuring position coordinates X _Ei , Y _{Ei are} subtracted component by component from the target position coordinates _i , _i . On the output side, the subtraction circuits 43.1 and 43.2 are connected on the one hand to the component outputs 46.1 and 46.2 and on the other hand to squarers 44.1 and 44.2 , which in turn are connected to a summing circuit 45 . The summing circuit 45 , on the output side of the distance square between dynamic target position _i , _i and measuring positions _Ei , _Ei , is connected to the distance square output 47 of the distance calculation circuit 40 .

The support coefficient circuit 120 shown in FIG. 4 results from the expansion of the coefficient circuit 20 as shown in FIG . The support coefficient circuit 120 therefore has, in addition to a support coefficient computer 150 comparable to the coefficient calculator 50 , the multiplication circuits 124.1, 124.2 and 124.3, 124.4 connected downstream of the inputs 121.1, 121.3 and inputs 122.1, 122.3 , which have a control input 128 with the control memory 68 are connected. The support coefficient calculator 150 , which is connected on the one hand to the output 137 of the support coefficient circuit 120 , is also followed by weight multipliers 135.1, 135.2 , which are connected on the output side to two multiplier stages 139.1, 139.3 and 139.2, 139.4 in pairs. Here, the the Stützkoeffi zientenausgängen 136.1 and 136.2 upstream Mul tiplizierstufen 139.1, 139.2 input side tenschaltung to the control input 128 of the Stützkoeffizien also connected 120, whereas the outputs 136.3 and 136.4 upstream Multipli ornamental step 139.3, 139.4, in addition to the obser-up time input 123 of the support coefficient circuit 120 is connected are.

The support coefficient calculator 150 has the distance calculation circuit 40 , whose distance square output 47 is connected to the output 157 of the support value estimation. The outputs 46.1 and 46.2 for the distance components are followed by the summation circuits 153.2 and 153.1 with their interconnected inputs for doubling, with the processing of a distance support S _R the negating and for processing a speed support S _V the direct inputs of the summation circuits 153.1 and 153.2 can be controlled. The outputs of the summation circuits 153.1, 153.2 form the outputs 156.1, 156.2 for first support coefficients of the support coefficient calculator 150 .

The interaction of the components shown in Fig. 1 to Fig. 4 under the control of the central control device 77 and thus the function of the filter is shown in the following functional description.

The filter is a digital filter, the entire signal processing of which is clock-controlled by the central control device 77 . The generation of the clocks in detail results from the dependencies of the data processing circuits indicated below and is easy to implement with known logic circuits. In addition, the central control device 77 contains a measurement timer in order to output measurement times T _Mi to the measurement time memory 70 and the observation time memory 65 .

As a direction finder 87 , for example, a solar system of "sufficient intelligence" is provided, ie, such a sonar system that is set up for automatic target tracking of the detected line. This direction finder 87 , which has measured values of the pursued target at every instant, is queried by suitable scanning pulses from the control device 77 and the bearings determined as measured values B _Mi are written into the measured value memory 86 . At the same time, the associated measurement time T _Mi generated in the control device 77 is stored in the measurement time memory 70 . With the query from the direction finder 87 , the query of the position measuring device 72 is triggered via the input interface 69 and the respectively associated measuring position coordinates X _Ei , Y _{Ei are transferred} to the partial memories 61 and 62 of the input memory 60 . In this way, i-measurements (i = 1, 2,..., N) are triggered by the control device 77 , so that the bearing B _Mi , the measuring times T _Mi and measuring position coordinates are associated as values which are additionally identified by the index i X _egg , Y _egg are stored.

Simultaneously with the storage, for example, of the second sensor B _{M 2} , the subsequent signal processing is activated by the control device 77 . For this purpose, initial values of the components _o , _o , _x , _{y of} the target vector are first read into the elements 11 to 14 of the target vector memory 10 from the basic data memory 15 . These initial values are arbitrary and e.g. B. for the target position components _o , _o due to the detection range of the sonar system 87 and for the speed component _x , _y due to the bearing difference B _{M 1} -B _{M 2} or known average speeds of ships fixed.

Although not shown in more detail in FIG. 1, a modification of the base memory 15 via the input interface 69 and the data display device 75 could also be implemented in a simple manner, with which initial values can then be written interactively from the data display device 75 into the base memory 15 and thus filtering the Measured values B _Mi for determining the target data _o , _o , _x , _{y are} carried out, which are based on changeable _or more precise initial values.

Initial values are only adopted at the beginning of a measurement cycle, namely when there are no values of components _o , _o , _x , _y in target vector memory 10 that are more precise than the initial values of base memory 15 .

The input vector = ( _o , _o , _x , _y ) 'stored in the target vector memory 10 contains, in addition to the static target position coordinates _o , _o, the speed components _x , _y , from which, taking into account the measuring times T _Mi and the measuring positions X _Ei , Y _Ei the coefficient circuit 20 determines the measurement coefficients h _{i 1} , h _{i 2} , h _{i 3} , h _{i 4} . For this purpose, the multipliers 25.1 and 25.2 and the adders 26.1 and 26.2 are provided in the coefficient circuit 20 , on which dynamic target position coordinates are formed from the input vector and the measuring times T _Mi. In the distance calculation circuit 40 of the coefficient calculator 50 , the distance components _Xi and _Yi according to Eq. Are then determined from the dynamic target position coordinates, taking into account the measurement position coordinates X _Ei and Y _Ei . (3.1) and (3.2) and the distance square ² _i according to Eq. (3.3) determined. The first measurement coefficients h _{i 1} , h _{i 2} and the measurement value estimation _{Mi are} determined from the output signals of the distance calculation circuit 40 by the divider circuits 54.1 to 54.3 and the Arcus Tangens calculator 55 . With the multiplication stages 39.3 and 39.4 , the further measurement coefficients h _{i 3} and h _{i 4} are then determined from the first measurement coefficients h _{i 1} , h _{i 2} by renewed multiplication by the measurement time T _Mi. On the output side of the coefficient circuit 20 there is therefore a sequence of four measurement coefficients h _{i 1} , h _{i 2} , h _{i 3} , h _{i 4} , which is assigned to the corresponding measurement value B _Mi via the index i.

Given the complexity of the filter, the filter coefficients f _{i 1} , f _{i 2} , f _{i 3} , f _{i 4} cannot be directly and essentially directly based on the model according to Eq. (7) can be determined, which calculates the input vector or the error vector Δp as a function of filter coefficient matrix F and measured values B _Mi , but it is necessary to first measure the measurement coefficients h _{i 1} , h _{i 2} , h _{i 3} , h _{i 4} des Determine measurement model and then convert the complete measurement coefficient matrix H into the filter coefficient matrix F. For this purpose, the computing device 80 is provided, in which the successively determined sequences of the measurement coefficients h _{i 1} , h _{i 2} , h _{i 3} , h _{i 4} are stored. The conversion of the measurement coefficient matrix H into the filter coefficient matrix F according to the matrix equation (8) is then enabled by the control device 77 when the number of sequences of measurement coefficients h _{i 1} to h _{i 4} determined by the index i is stored. Such conversions generally require a greater amount of time, so that here asynchronous processing takes place with a start by the control device 77 and a completion message to the control device 77 by the computing device 80 . With the completion message, all filter coefficients f _{i 1} , f _{i 2} , f _{i 3} , f _{i 4} assigned to the measured values B _Mi are then stored in the computing device 80 .

Simultaneously with the transfer of the measurement coefficients h _{i 1} to h _{i 4} to the computing device 80 , the measurement value estimate _Mi determined in the coefficient circuit 20 has been transmitted to the difference unit 85 , in which the respective differences δ _{Mi are} formed and stored.

The actual output signal of the filter, the output or error vector Δp of the target data, is obtained component by component by multiplying the filter coefficients f _{i 1} to f _{i 4} by the difference δ _Mi and the summation of all these products over all i in the accumulation circuit 88 educated. This output vector Δp is then checked in the comparison circuit 90 in such a way that its vector standard || Δp ||

|| Δp || = [(ΔX _o + ΔV _x · T _Mi ) ² + (ΔY _o + ΔV _x · T _Mi ) ²] ^½ (9)

is determined and compared with a predetermined threshold stored in the comparison circuit 90 . Is the threshold of the vector norm || Δp || according to Eq. (9), the output vector Δp is added to the input vector in the addition stage 94 , the addition result is written as an improved input vector into the target vector memory 10 and its output via the output interface 96 z. B. unlocked to the data display device 75 . Otherwise, only the corrected input vector is written into the target vector memory 10 and the determination of the filter coefficient matrix F is restarted with the same measured values B _Mi , but with this already improved input vector.

The iterative determination of the target data converges faster, ie with fewer iterations, if the support coefficients h _{S 1} , h _{S 2} , h _{S 3} , h _{S 4} and a support value estimate are determined in an additional support coefficient circuit 120 , with which additional measurements or observations, for example Distance measurement with radar, speed monitoring due to the propeller noise or course observation through periscope observation, are included by the computing device 80 in the calculation of the filter coefficient matrix F.

In the support coefficient circuit 120 , which is largely similar to the coefficient circuit 20 , the support coefficients h _{S 1} to h _{S 4} assigned to a support value _S and the square of the support value estimate 2 are determined and are preferably fed to the computing device 80 after all measurement coefficients h _{i 1} to h _{i 4} . The Re chenvorrichtung 80 calculates an advanced filter coefficient matrix F, and subsequent to the accumulation of the products of differences δ _Mi and filter coefficients f _{i 1} to _{i 4} is supplied via the switch 181, the difference δ _S of the square of the support value S² and the square of the support appreciation ² connected to the accumulation circuit 88 , multiplied by the filter coefficients f _{S 1} to f _{S 4} and thus an output vector Δp with the components

educated.

Here, i = N indicates the number of measured values, up to one Time, i. a. the current measuring time will.

The support coefficient circuit 120 shown in FIG. 4 can be modified with little effort for various support value inputs S or their squares S 2, for this purpose the multiplication circuits 124.1 to 124.4 are provided, which can be controlled by the control memory 68 via the control input 128 .

By entering a squared distance support S² _R and belonging to the reference time T _R observation position X _R, Y _R is input to a one in the control memory 68 and thus the switching of the target position coordinates _{o, o} to the adders 26.1, 26.2 and the observation position X _R and Y _R on the support coefficient calculator 150 and the activation of the outputs 136.1 and 136.2 by the multiplier stages 139.1 and 139.2 . In the reference routing memory 65 , the reference time T _{R of} the distance support S _{R is} stored, so that the support coefficient calculator 150 outputs the outputs of the adders 126.1 and 126.2 as observed dynamic target position coordinates _o + _x * T _R and _o + _y * T _R and the observation positions X _R , Y _{R are} supplied.

In the support coefficient calculator 150 shown in FIG. 4, first support coefficients result from this

= -2 _x ; = -2 _y (11.1)

With

_x = _{y o} + T _R · X and _y = _{R o} + _y · T _R Y _R,

since the multiplier levels 139.1 and 139.2 only result in a multiplication by one, and the square of the base value estimate ² _R increases

² _R = ²,

in which

² = ² _x + ² _y

is. The other support coefficients

= · T _R and = · T _R (11.2)

result as output signals of the multiplier stages 139.3, 139.4.

With the input of a squared speed support S² _V via the data display device 75 , the control memory 68 is set to zero and the reference time memory 65 to one. The inputs 121.1, 121.3 and 122.1, 122.3 are thus deactivated and the outputs 136.1 and 136.2 are set to zero, so that only the coefficient sequence associated with the speed support S _V is assigned

is determined.

For the input of a course support S _K , the support coefficient calculator 150 in the support coefficient circuit 120 would have to be replaced identically by the coefficient calculator 50 in accordance with an embodiment not shown. Is then set at the input of the price support S _K by the data display unit 75 of the control memory 68 to zero, and the reference time memory 65 on one, so are in the support coefficient circuit 120, the tap coefficients

formed, which are supplied via the outputs 136.1 to 136.4 of the computing device 80 to complete the measurement coefficient matrix H.

When filtering measured values B _Mi using a distance support S _R or a speed support S _V , the differences δ _SR or δ _SV from the squared support values S _R ² or S _V ² and estimates _R ² or _V ², ie

δ _SR = S _R ² - _V ² or δ _SV = S _V ² - _V ² (14.1)

weighted with the filter coefficients f _S to or to. Whereas with a course support the difference δ _SK from linear course support S _K and estimate _K

δ _SK = S _K - _K (14.2)

is filtered.

The embodiment according to FIGS. 5 and 6 is a modification of the embodiment shown in FIGS . 1 and 2, in which, for simplification, the processing of base values has been dispensed with. In the description of the already described in FIG. 1 and 2 indicated like components, which are shown in Fig. 5 and Fig. 6 with identical reference numerals to this embodiment will only be received by the description, when it is necessary for the representation of the function.

Due to the modifications of the filter according to FIG. 5, the target data of a vehicle are determined by the filter in a longer time interval even if the target moves on a target path with different speed components. These target path sections are generally referred to as "legs", it being assumed that the speed components belonging to the respective leg are constant.

To detect the times at which the target vehicle has performed a maneuver, ie a change in its speed components, a maneuver detector 76 is provided, which is connected to the direction-finding device 87 and the control device 77 on the input side. The times for detected target maneuvers are stored in a downstream maneuver time memory 71 , which has a counter 71.5 for the number of detected maneuvers and which has maneuver memory locations 71.0, 71.1 , the start time T _{A0 of} all measurement times T _Mi , dhia T _{M 1} = 0, in the maneuver memory location 71.0 , is saved. This maneuver time memory 71 and the measurement time memory 70 are connected to a time comparison circuit 79 , the time outputs 79.0 and 79.1 of the coefficient circuit 20 and the comparison circuit 90 are closed.

In the target vector memory 10 , additional memory elements 16 and 17 are provided for the speed components after the first maneuver _{x 1} and _{y 1} , which on the output side are connected on the one hand to the coefficient circuit 20 and on the other hand to the output ion interface 96 and on the input side as well as the memory elements 11 to 14 are controlled by the now expanded gate circuit 95 and for initialization by the basic data memory 15 . To the expanded number of vector components to be determined _o , _o , _x , _y , _{x 1} , _{y 1} , the computing device 80 , the accumulation circuit 88 , the comparison circuit 90 and the addition stage 94 are also adapted, which can be seen directly from the expanded number of connecting lines .

The coefficient circuit 20 of FIG. 6 has further inputs 21.4 and 22.4 for the Geschwin digkeitskomponenten _{x 1} and _{y 1} and a further Meßzeiteneingang 23.1. At the measuring times inputs 23 and 23.1 , the time signal values T ' _Mi and T' _{Mi 1} are available as measuring times. In parallel to the multipliers 25.1 and 25.2 , further multipliers 25.3 and 25.4 are provided, the multiplier 25.3 on the input side from the input 21.4 with the speed component _{x 1} and from the input 23.1 with measuring times T ' _{Mi 1} and the multiplier 25.4 on the input side from the input 22.4 with the Speed component _{y 1} and the input 23.1 are controlled with measuring times T ' _{Mi 1} . On the output side, the multipliers 25.3 and 25.4 are each connected to the adders 26.1 and 26.2 .

The coefficient circuit 20 also contains further multiplier stages 39.5 and 39.6 , which are connected to the coefficient outputs 56.1 and 56.2 and respectively to the input 23.1 for measuring times T ' _{Mi 1} . With these multiplier stages 39.5 and 39.6 , the additional measurement coefficients h _{i 5} and h _{i 6 are} formed, which are present at the measurement coefficient outputs 36.5 and 36.6 , which are connected to the computing device 80 in FIG. 5.

In FIGS. 5 and 6, although an embodiment for determining the Filterkoeffizientenf _i shown at a single target maneuver, but by similar parallel extensions is to realize a filter with which the target data of a target for a random number j are determined from maneuvers. This is also pointed out in several of the following sections.

The modified filter according to FIG. 5 determines the target data of a target which is moving on a target path according to FIG. 7. The representation of the target path in FIG. 7 is based on a Cartesian X, Y coordinate system, in the origin of which the measuring position X _Ei , Y _{Ei is} at a point in time, generally the starting point in time of the measurements. The target path begins, as shown, at time T _{A 0} at the target position X _o , Y _o and the target has the speed components _x , _y . At time T _{A 1} , the maneuver detector 76 , which continuously subjects the target noise to a signal or frequency analysis, has detected a maneuver from a signal or frequency change. From this point in time T _{A 1 the} target has the speed components _{x 1} , _{y 1} . The section of the target path in the time interval T _{A 0} , T _{A 1} is also referred to as Leg 0 and the section beginning at time T _{A 1} as Leg 1. When further maneuvers are detected, the target path is supplemented accordingly, as indicated in FIG. 7 by the dashed representation of the target path for leg 2 from the time T _{A 2} onwards. The associated speed components would then be _{x 2} and _{y 2} .

To determine the vector components of the target on the different legs, the maneuver times T _{Aj are written} into the maneuver time memories 71 after the initial time T _{A 0} . The detection of maneuvers always takes place only after a number of measurements due to the construction of the maneuver detector 76 , that is to say with a constructional, time delay. The consequence of this is that the measurements can also be evaluated by the filter only after this design-related delay.

The speed components _x , _y and _{x 1} , _{y 1 represent} the correct target state in pairs only in the time interval associated with the leg between two maneuvers. Therefore, the time at the beginning of a leg indicates the lower interval limit and at the end of a leg the upper interval limit. In the example in FIG. 7, the time T _{A 0 is} the lower and the time T _{A 1} the upper interval limit for Leg 0, whereas the lower interval limit for Leg ₁ is designated by T _{A 1} . The upper interval limit T _{A 2} would only be determined by the detection of a further maneuver. The memory locations provided for this purpose in the maneuvering time memory 71 - they are not shown in FIG. 5 - are occupied with the maneuvering times T _Aj = ∞ just like the memory locations 71.0 and 71.1 immediately before the start of the measuring cycle.

In the time comparison circuit 79 in FIG. 5, the measurement times T _Mi in the measurement time memory 70 are compared with the maneuver times T _Aj in the maneuver time memory 71 and assigned to the intervals, ie the legs of the target path. An output of the time comparison circuit 79 is assigned to each leg. This means that time signal values T ′ _Mi for Leg 0 are present at time output 79.0 and time signal values T ′ _{Mi 1} for Leg 1 at time output 79.1 . These time signal values have the value zero at each output, as long as the measuring times T _{Mi are} less than or equal to the detected maneuver time T _Aj , the starting time T _{A 0 being} understood in this sense as the maneuver time, or they have the difference signal value between the measuring time T _Mi and lower interval limit T _Aj if the measuring time T _{Mi is} greater than the lower interval limit T _Aj and smaller than the upper interval limit T _{Aj + 1} , or they have the difference signal value between the upper interval limit T _{Aj + 1} and the lower interval limit T _Aj if the measuring time T _{Mi is} greater than the upper interval limit T _{Aj + 1} . The function of the time comparison circuit is 79 for any number j of maneuvers for each time output j by Eq. 15

T ′ _Mÿ = max (0, min (T _Mi -T _Aj , T _{Aj + 1} -T _Aj )) (15)

to describe.

In the coefficient circuit 20 , the time signal values T ' _Mi at the time output 79.0 are multiplied by the speed components _x , _y and the time signal values T' _{Mi 1} at the time output 79.1 are multiplied by the speed components _{x 1} , _{y 1} , coordinate by component to the components _o and _o added, so that the extended target position coordinates _i , _i result. Taking into account the stored measurement position coordinates X _Ei , Y _Ei , the result is analogous to Eq. (3.1) and (3.2) the extended distance _{components xi} , _yi

_xi = _o + _xT ′ _Mi + _x 1T ′ _{Mi 1} - X _Ei (16.1)

_yi = _o + _yT ′ _Mi + _{y 1} · T ′ _{Mi 1} - Y _Ei . (16.2)

With these distance components _xi , _yi , the measurement coefficients h _{i 1} ,. . ., h _{i 4} according to Eq. (5) determines, the measurement coefficients h _{i 3} and h _{i 4} resulting from the multiplication of the measurement coefficients h _{i 1} and h _{i 2} by the time signal values T ' _Mi at the input 23 . The for determining the vector components of the target state vector expanded by the speed components _{x 1} , _{y 1}

= ( _o , _o , _x , _y , _{x 1} , _{y 1} ), (17)

required measurement coefficients h _{i 5} and h _{i 6} are additionally calculated by multiplying h _{i 1} and h _{i 2} by the measurement times T ′ _{Mi 1} at input 23.1 and transmitted to the computing device 80 via outputs 36.5 and 36.6 .

As can be seen from the conditions according to Eq. (15) for the time signal _values T ' _Mÿ at the time _outputs 79. j, the time signal _value T' _{Mÿ is} always identical when the measuring time T _{Mi is} less than the maneuver time T _Aj , but also when no maneuver is detected has been. As a result, the calculation of the extended filter coefficient matrix F by the computing device 80 is only necessary when the filter coefficient matrix F is determined for a current measurement time T _Mi that is greater than the maneuver times T _Aj . For this purpose, the number j of maneuver times T _Aj , each of which has been detected up to the current measurement time T _Mi , is counted in the counter 71.5 of the maneuver time memory 71 provided for addressing the maneuver memory locations 71.0 , 71.1 and transmitted to the computing device 80 via the control bus 78 . As a result, the computing device 80 designed to determine coefficients h and f for a maximum number J of target maneuvers is controlled such that the degree of the matrices H and F is limited in proportion to the number j of maneuvers detected. In the exemplary embodiment specified, the coefficient sequences are maneuver-related, that is to say they are each expanded by the measurement coefficients h _{i, 2j + 3} and h _{i, 2j + 4} or the filter coefficients f _{i, 2j + 3} and f _{i, 2j + 4} .

In the computing apparatus 80 according to Eq. (8) determined from the measurement coefficients h _{i 1} to h _{i 6,} the filter coefficients f _{i 1} to f _{i 6} , with which the accumulation circuit 88 is controlled. The accumulation circuit 88 is therefore provided for the component-based accumulation of the vector components to be determined on the basis of a maximum predetermined number of detectors to be detected, ie with a maximum number of J maneuvers 2J + 4 parallel accumulation levels are present. The testing and processing of the extended output vector Δp is fundamentally carried out in a manner which is already stated in the filter without additional maneuver detection according to FIG. 1. Only in the comparison circuit 90 is a vector standard || Δp || for the comparison with the threshold of the output vector Δp

to determine the components of the output vector Δp and now the time signal _values T ' _Mÿ at the time signal outputs 79. j are taken into account.

The embodiment of FIG. 5 is to be improved by the input of tap coefficients h _S. The support coefficients h _S result in an improvement in the vector components of one or more legs and are classified in the measurement coefficient matrix H in accordance with these legs. For those legs for which no further supporting observations are available, the associated elements of the measurement coefficient matrix H are then to be filled with zero values.

Claims (17)

1. Adaptive filter for determining target data from recorded measured values (B _Mi ), e.g. B. bearings, with a target vector memory ( 10 ) for an input vector (), the target position coordinates ( _o , _o ) as vector components, with an input memory ( 60 ) containing the coordinates of each of the measurement positions (X _Ei , Y _Ei ), with one with the target vector memory ( 10 ) and the input memory ( 60 ) connected coefficient calculator ( 50 ) for determining a sequence of measurement coefficients (h _{i 1} , h _{i 2} ) associated with each measured value (B _Mi ) and a measured value estimate ( _Mi ) from the input vector () and the associated measuring position (X _Ei , Y _Ei ), with a downstream of the measurement coefficient outputs of the coefficient calculator ( 50 ) computing device ( 80 ) for filter coefficients (f _{i 1} , f _{i 2} ), in which the matrix (F) of the filter coefficients (f _{i 1} , f _{i 2} ) from the matrix (H) and the transposed maxtrix (H ′) of the measurement coefficients (h _{i 1} , h _{i 2} ) according to the matrix equation F = (H ′ · H) ^-1 · Is determined with a r for determining an output vector (Δp) of the target data accumulation circuit ( 88 ) in which for each vector component of the output vector (Δp) the weighted differences (δ _Mi ) between the measured value (δ _Mi ) with the filter coefficients (f _{i 1,} f _{i 2} ) B _Mi ) and the respective measured value estimate ( _Mi ) are added up component by component, and with a comparison circuit ( 90 ) for the output vector (Δp), which on the one hand inscribes the input vector () corrected by the output vector (Δp) into the target vector memory ( 10 ) and which, on the other hand, triggers an output of the corrected input vector () if the threshold is undershot or repeats the determination of the target data if the threshold is exceeded,
characterized in that the input vector () stored in the target vector memory ( 10 ) has speed components ( _x , _y , _{x 1} , _{y 1} ) as a further vector component, in that a measuring time memory ( 70 ) connected to a measuring timer which is part of a control device ( 77 ) ) it is provided that for each speed component ( _x , _y , _{x 1} , _{y 1} ) a multiplier ( 25.1, 25.2, 25.3, 25.4 ) is provided, which is connected on the input side to the target vector memory ( 10 ) and the measurement time memory ( 70 ) that a target position coordinate input ( 51.1, 52.1 ) of the coefficient calculator ( 50 ) is preceded by an adder ( 26.1, 26.2 ), the input side with the multiplier ( 25.1, 25.2, 25.3, 25.4 ) and with the target vector memory ( 10 ) for transfer the target position coordinates ( _o , _o ) is connected to each measuring coefficient output ( 56.1, 56.2 ) of the coefficient calculator ( 50 ) at least one with the measuring times memory ( 70 ) connected multiplier stage ( 39.3, 39.4, 39.5, 39.6 ) for forming further measurement coefficients (h _{i 3} , h _{i 4} , h _{i 5} , h _{i 6} ) is connected and that the multiplier stages ( 39.3, 39.4, 39.5, 39.6 ) are connected on the output side to the computing device ( 80 ). 2. Adaptive filter according to claim 1, characterized in that the coefficient calculator ( 50 ) has a distance computing circuit ( 40 ) which can be controlled with target position coordinates ( _o , _o ) and measuring position coordinates (X _Ei , Y _Ei ) and outputs ( 46.1, 46.2 ) for the distance components ( _yi , _xi ) and a distance square output ( 47 ) for the sum of the squares of the distance components ( _yi , _xi ) and that the coefficient calculator ( 50 ) is designed such that the determination of the measurement coefficients ( h _{i 1} , h _{i 2} ) according to and the determination of the measured value estimate ( _Mi ) as a function of the distance components, preferably according to he follows. 3. Adaptive filter according to claim 1 or 2, characterized in that the computing device ( 80 ) is additionally controllable with support coefficients (h _{s 1} , h _{s 2} , h _{s 3} , h _{s 4} ) of a support coefficient circuit ( 120 ) that the matrix (F) the filter coefficients (f _{i 1} to f _{i 4} , f _{S 1} to f _{S 4} ) from the matrix (H) expanded by the support coefficients (h _{S 1} , h _{S 2} , h _{S 3} , h _{S 4} ) the measurement coefficient (h _{i 1} , h _{i 2} , h _{i 3} , h _{i 4} ) is determined, and that the accumulation circuit ( 88 ) has the differences (δ _Mi ) from the measured values (B _Mi ) and the measured value estimates ( _Mi ) and at least a difference value (δ _S ) from a support value input (S) into a support value memory ( 67 ) and a support value estimate () determined by the support coefficient circuit ( 120 ) in the same order in which the computing device ( 80 ) with the measurement coefficients ( h _{i 1} , h _{i 2} , h _{i 3} , h _{i 4} ) and the support coefficients (h _{S 1} , h _{S 2} , h _{S 3} , h _{S 4} ) has been expensive to be supplied. 4. Adaptive filter according to one of claims 1 to 3, characterized in that the input memory ( 60 ) for storing at least one base value (S), z. B. an observed distance value (S _R ), a speed value (S _V ) and / or a course value (S _K ), a reference time (T _R ) and observation coordinates (X _R , Y _R ) is formed and that the support coefficient circuit ( 120 ) with their inputs ( 121.1, 121.2, 122.1, 122.2 ) for input vector components ( _o , _o , _x , _y ) with the target vector memory ( 10 ) and with their inputs ( 123, 121.3, 122.3 ) for the reference time (T _R ) and the observation coordinates (X _R , Y _R ) are connected to the input memory ( 60 ). 5. Adaptive filter according to one of claims 3 to 4, characterized in that the support coefficient circuit ( 120 ) has a support coefficient calculator ( 150 ), the inputs ( 151.1, 152.1 ) for target position coordinates in the same way as the coefficient calculator ( 50 ) via multipliers ( 125.1, 125.2 ) and adders ( 126.1, 126.2 ) can be controlled, and its inputs ( 151.2, 152.2 ) for measuring position coordinates are connected to the partial memories ( 63, 64 ) for observation positions of the input memory ( 60 ) that the support coefficient calculator ( 150 ) outputs ( 156.1, 156.2 ) for first support coefficients (h _{S 1} , h _{S 2} ) and an output ( 157 ) for the support value estimation (), that the multiplier ( 125.1, 125.2 ) via the observation time input ( 123 ) with the input memory ( 60 ) and in that the support coefficient circuit (120) multipliers (139.3, 139.4), which on the input side with the outputs nts (156.1, 156.2) for the first tap coefficients (h _{S 1,} h _{S 2)} and time input to the observation (123) are connected to which of the aisle side further supporting coefficients (h _{S 3,} h _{S 4)} are pending, which together with the form first support coefficients (h _{S 1,} h _{S 2} ) output signals of the support coefficient circuit ( 120 ). 6. Adaptive filter according to one of claims 3 to 5, characterized in that the outputs ( 156.1, 156.2 ) of the support coefficient calculator ( 150 ) for evaluating the first support coefficients (h _{S 1,} h _{S 2} ) and the differential circuit ( 185 ) each Ge Weight multipliers ( 135.1, 135.2, 183 ) are connected downstream and that the weight multipliers ( 135.1, 135.2, 183 ) are connected to a weight factor memory ( 66 ) of the input memory ( 60 ). 7. Adaptive filter according to one of claims 3 to 6, characterized in that the support coefficient circuit ( 120 ) multiplier units ( 124.1, 124.2, 124.3, 124.4, 139.1, 139.2 ), which on the one hand via a control input ( 128 ) with the input memory ( 60 ) and on the other hand each with the inputs ( 121.1, 122.1 ) for target position coordinates ( _o , _o ) and the inputs ( 121.3, 122.3 ) for observation coordinates (X _R , Y _R ) or the outputs for first support coefficients (h _{S 1 ,} h _{S 2} ) are connected downstream. 8. Adaptive filter according to one of claims 3 to 7, characterized in that the support value input (S) is a distance support (S _R ), that the support coefficient calculator ( 150 ) has the distance calculation circuit ( 40 ), the distance component outputs ( 46.1, 46.2 ) with the outputs ( 156.2, 156.1 ) of the support coefficient calculator ( 150 ) for first support coefficients (h _{S 1,} h _{S 2} ) and their distance square output ( 47 ) is connected to the output ( 157 ) for the support value estimation (), and that in the control memory ( 68 ) a one is stored. 9. Adaptive filter according to one of claims 3 to 7, characterized in that the support value input (S) is a speed support (S _V ), that the support coefficient calculator ( 150 ) has the distance calculation circuit ( 40 ), the distance component outputs ( 46.1, 46.2 ) with the outputs ( 156.2, 156.1 ) of the support coefficient calculator for first support coefficients (h _{S 1,} h _{S 2} ) and their distance square output ( 47 ) are connected to the output ( 157 ) of the support coefficient calculator ( 150 ) for the support value estimation (), and that a zero is stored in the control memory ( 68 ). 10. Adaptive filter according to one of claims 3 to 7, characterized in that the support value input (S) is a course support (S _K ), that the support coefficient circuit ( 120 ) as the support coefficient calculator ( 150 ) has the coefficient calculator ( 50 ) and that a zero is stored in the control memory ( 68 ). 11. Adaptive filter according to one of claims 1 to 9, characterized in that the computing device ( 80 ) for determining the filter coefficients (f) is a field or matrix processor. 12. Adaptive filter according to one of claims 1 to 11, characterized in that the control device ( 77 ) for clock generation and synchronization and the associated measuring timer are provided for generating the measuring times (T _Mi ). 13. Adaptive filter according to one of claims 1 to 12, characterized in that a maneuver detector ( 76 ) connected to the direction finder device ( 87 ) and the control device ( 77 ) and a maneuver time memory ( 71 ) are provided, in which a number (j ) of maneuver times (T _{A 0} , T _{A 1} ) of a target having different speed components ( _x , _y , _{x 1} , _{y 1} ) in successive time intervals are stored, that between the measurement time memory ( 70 ) on the one hand and the multipliers ( 25.1 to 25.4 ) and the multiplier stages ( 39.3 to 39.6 ), on the other hand, a time comparison circuit ( 79 ) is switched on, which is connected on the input side to the maneuver time memory ( 71 ) and has time interval-specific time outputs ( 79.0, 79.1 ) for time signal values (T ′ _Mi , T ' _{Mi 1} ) has that the multiplier ( 25.1, 25.2 or 25.3, 25.4 ) for speed components ( _x , _y b between _{x 1} , _{y 1} ) of the same time interval are connected to the same time interval- specific time output ( 79.0, 79.1 ), and that the multiplier stages ( 39.3 to 39.6 ) connected downstream of each measurement coefficient output of the coefficient calculator ( 50 ) correspond to the associated number of time intervals time interval-specific time output ( 79.0, 79.1 ) are connected. 14. Adaptive filter according to claim 13, characterized in that the time comparison circuit ( 79 ) is designed such that at the time interval specific time output ( 79.0, 79.1 ) as a time signal value (T ' _Mi , T' _{Mi 1} ) there is either a zero signal value when the measuring time (T _Mi ) is less than or equal to the maneuver time (T _{A 0} , T _{A 1} ) of the lower interval limit, or a difference signal value between the measuring time (T _Mi ) and maneuver time (T _{A 0} , T _{A 1} ) of the lower interval limit is present, if the measuring time (T _Mi ) is greater than the lower and less than or equal to the maneuvering time (T _{A 0} , T _{A 1} ) of the upper interval limit, or if there is a difference signal value between the upper and lower interval limit if the measuring time (T _Mi ) is greater than is the maneuver time (T _{A 1} ) of the upper interval limit. 15. Adaptive filter according to one of claims 13 or 14, characterized in that the maneuver time memory ( 71 ) via the control bus ( 78 ) is connected to the computing device ( 80 ) and that the computing device ( 80 ) is designed such that the degree of the matrices H and F can be limited proportionally to the number (j) of maneuver times (T _Aj ) that have been detected up to the current measuring time (T _Mi ). 16. Adaptive filter according to one of claims 13 to 15, characterized in that the maneuver time memory ( 71 ) with the control bus ( 78 ) connected counter ( 71.5 ) for determining the number (j) of detected maneuver times (T _Aj ) and from the counter ( 71.5 ) has addressable memory locations ( 71.0, 71.1 ) for the maneuver times (T _Aj ). 17. Adaptive filter according to one of claims 1 to 16, characterized in that the vector standard || Δp || in the comparison device ( 90 ) of the output vector (Δp) is compared with the predefinable threshold.