UA56074A - Adaptive device for tracking maneuvering targets - Google Patents

Abstract

The proposed adaptive device for tracking maneuvering targets contains 2-power and 3-power Kalman filters, a switch, a unit for detecting target maneuvering, and a unit for determining intensity of target maneuvering. The unit for detecting target maneuvering contains a unit for forming signal sample, a unit for rank calculation, a unit for calculating rank statistics, and a threshold unit. The unit for determining intensity of target maneuvering contains a signal sample reading unit, a memory unit, a unit for determining errors, and a unit for determining target acceleration. The proposed device provides for the possibility of timely nonparametric detection of a target at a specified false alarm probability, decreasing dynamic errors in signal filtration, and enhancing stability of the target tracking system.

Description

The proposed invention relates to the field of radar and can be used to estimate the trajectory parameters of a maneuvering target.

A well-known device for monitoring and estimating the parameters of the target trajectory (1Й, in which the estimation is carried out according to the recurrent scheme of the linear Kalman filter. The structural diagram of this device consists of t Kalman filters connected in parallel, each of them tuned to a certain value of the exciting parameter (acceleration) in the range of possible accelerations maneuvering targets The resulting estimate of the filtering parameters is formed as a weighted sum of the conditional estimates at the filter outputs.

The disadvantage of the known device is the complexity and bulkiness of the scheme, as well as the fact that the obtained estimates have insufficient accuracy due to the influence of filters that are detuned relative to the value of the target's acceleration.

The closest to the proposed technical solution, the prototype chosen, is a device for tracking maneuvering targets, I2I|, which includes Kalman filters of the second and third orders with a switching scheme at their output, as well as t filters connected in parallel, tuned for different times the appearance of the target maneuver, the block for choosing the maximum estimate and the block for comparison with the threshold.

The operation of the device is as follows. The quadratic component of the target's motion equation due to its maneuver is analyzed. The criterion for detecting a maneuver is the parameter size exceeding some limit, exciting level. When the threshold is exceeded, the Kalman filter of the second order is switched to the Kalman filter of the third order.

The disadvantage of this device is that when detecting a target maneuver, the possibility of changing the law of the distribution of tracking errors is not taken into account due to the influence of various factors - measurement errors, passive and active obstacles, while the specified probability levels of correct detection of the maneuver and false alarms may not be ensured.

In addition, detection of the target's maneuver is done with a very long delay, which leads to the appearance of significant dynamic tracking errors. When evaluating the intensity of the maneuver, a cumbersome scheme with no filters is used.

The invention is based on the task of creating such an adaptive device for tracking maneuvering targets, in which the introduction of new blocks, such as a non-parametric target maneuver detection block and a maneuver intensity assessment block, will allow for timely detection of the maneuver under different laws of the distribution of the exciting parameter and, in addition, will allow to evaluate the characteristics maneuver Thus, the specified probabilities of correct detection and false alarm are ensured and dynamic tracking errors are reduced.

To solve the problem, the device, which contains serially connected blocks of Kalman filters of the second and third order, and a switch block, additionally includes a maneuver detection block and a maneuver intensity assessment block, while the maneuver detection block consists of the following serially connected: formation block samples, a rank calculation block, a rank statistics calculation block, and a threshold block, and the maneuver intensity estimation block consists of the following serially connected: sampling block, memory device, error accumulation block, and acceleration estimation block, while the input of the device is the filter input second-order Kalman filter, the second input of the second-order Kalman filter is connected to the input of the sampling block and to the second input of the rank calculation block, the first output of the second-order Kalman filter is additionally connected to the second input of the switch, the threshold block input is connected to the third the input of the switch and the second input of the sampling unit, the second output of the sampling unit and is connected to the first input of the de-sampling unit, the output of the acceleration estimation unit is connected to the second input of the third-order Kalman filter, and the output of the device is the output of the switch.

The entered blocks allow to detect the maneuver of the target and to estimate its intensity by the values of the sample of tracking errors. The use of a non-parametric detector is more effective in conditions where the error distribution law is unknown.

Based on the same sample, the value of the acceleration is estimated, which is used to correct the filter parameters. At the same time, the dynamic filtering errors are significantly reduced, and the tracking stability is increased.

Figure 1 shows the structural diagram of the proposed device.

Figure 2 shows the dependences of the size of the dynamic error Ap of tracking the trajectory of a maneuvering target without taking into account the acceleration at the output of the Kalman filter of the second order (curve 1), at the output of the Kalman filter of the third order, with the acceleration taken into account, with a delay in the detection of the maneuver of the target (curve 2), with inaccuracy in estimating the intensity of the maneuver (curve 3) and with minimal delay in detecting the maneuver and accounting for the acceleration of the target (curve 4).

Here it, yu, ik are the time of the start of the maneuver, the detection of the maneuver, and the end of the maneuver of the target, respectively.

Graphical dependences show that the timely detection of the target's maneuver and the calculation of the acceleration estimate, implemented in the proposed device, reduce the value of dynamic filtering errors and, thereby, increase the stability of tracking the target.

The proposed device contains serially connected: second-order Kalman filter unit 1, filter unit

Kalman of the third order 2, a switch 3, as well as a maneuver detection unit 4 and a maneuver intensity assessment unit 9, while the maneuver detection unit 4 consists of the serially connected: sampling unit 5, rank calculation unit 6, rank statistics calculation unit 7 and threshold block 8. Maneuver intensity assessment block 9 consists of serially connected: sampling block 10, storage device 11, error accumulation block 12 and acceleration estimation block 13, while the input of the device is the input of the filter block

Second-order Kalman filter 1, the second output of the second-order Kalman filter block 1 is connected to the input of the sampling block 5 and to the second input of the rank calculation block 6, the first output of the second-order Kalman filter block 1 is additionally connected to the second input of the switch 3, the output threshold unit 8 is connected to the third input of the switch C and the second input of the sampling unit 10, the second output of the sampling unit 5 is connected to the first input of the sampling unit 10, the output of the acceleration evaluation unit 13 is connected to the second input of the Kalman filter unit of the third order 2, the output of the device is the output of the switch 3.

The operation of the proposed device is as follows.

The input of the device in every n-th radar inspection receives the values of the parameters of the operational points of the control unit.

A Kalman filter of the second order works on the deterministic section of the target's trajectory, which forms and outputs through the commutator a vector of estimates of the parameters of the linear trajectory of the target Hu and the correlation matrix of measurement errors Chi.

In this case, the ratios of the linear Kalman filter are used. For example, when estimating the distance at the moment of time ip:

Ha - Pa Ya Ad(t pz), /) and t to her! ! ! иш, where п, пз, п - according to the measurement, the extrapolated value of the range and its estimate;

Ap - the smoothing coefficient determined by the values of extrapolation errors and parameter measurement errors;

Ha - Pa - A! ! ! , popa n - inconsistency between the measured and extrapolated values (inelastic).

In the absence of a maneuver, the parameter estimation error is white Gaussian noise with zero mean and variance: ra - НН В, (d) where Chpz is the correlation matrix of extrapolation errors;

H is a linear operator of correspondence between the estimated parameters and the measured coordinates;

A - correlation matrix of measurement errors.

The maneuver of the target occurring at the moment of time it causes a jump in acceleration d, which leads to the appearance of a tracking error component, the average value of which will change over time:

Midh|- (t) a, (3) where d is the value of acceleration;

E(,t) is the transition matrix of disturbances of the system.

A satisfactory approximation of the quantity K(i|,t) can be obtained if one restricts oneself to the acceleration component: : 1 i.e.

Kot) 5Ott) te, (3 where |, t - correspond to the current moment of time ii and the moment of the beginning of the maneuver of the entire yit;

T - information recovery period (radar inspection period).

Then the value of the exciting parameter in this case - the extrapolation error of the parameter at /|- step will be equal to: . 1. 2-2. de Rftua-ost t, iht (5)

The quadratic form of the follow-up error C) is subject to non-central y? distribution with the non-centrality parameter (11: 2

MUAP) ba JSC (6) p

Detection of the target's maneuver is done in block 4 by sampling from | - 1, 2,...M values of tracking errors 7; in the "sliding window" size M. . . . oooh And

When a maneuver is detected, the commutator is switched and the output is received, the estimation of the parameters "formed in the Kalman filter of the third order is received.

When evaluating the parameters, the acceleration values formed in block 9 from sample 7) on which the maneuver was detected are used.

Let us consider in more detail the operation of the maneuver detection unit 4 and the maneuver intensity assessment unit 9.

Target maneuver detection is carried out by analyzing sample 2, where | - 1, 2,..., M by comparing it with the reference sample 2(|-t).

The reference sample 24(|-t) precedes the analyzed sample 2; and is pre-recorded in the sampling block 5.

A non-parametric rank detector implemented in block 4 is used to detect the maneuver.

In block 6, the ranks of values of the analyzed sample 7 are determined; in comparison with the error values of the reference sample 4(|-t): Ви, Н2,...,Вт. The analyzed sample is 7; is also recorded in the register of the sampling unit 5.

Next, in block 7, the value of the rank statistic 5 is calculated.

Taking into account the nature of the change in tracking errors during the maneuver of the target, when a shift (trend) is observed in the dimensions of the discrepancy Di, the most effective rank correlation (31:

M

5-X B) (7

1st

The value 5 is then equal to the threshold level of Discord, and at 5 » Discord, a decision is made about the presence of a target maneuver.

The size of the threshold is selected from the given probability of false alarms, based on the analysis of the distribution of errors (2).

On the impulse of exceeding the threshold, the switch З is switched and also the unit 10 is controlled for removing the sample 7; by which the maneuver of the target was detected, and the recording of the given sample in the memory device, 11.

The assessment of the intensity of maneuver 4 in block 9 is carried out according to the expression obtained from the ratios (4) 1 (5):

M

7 Khro, and M av (8)

M 2; that), and where the size of Pm means the energy of the disturbance signal.

To obtain an estimate of the acceleration in the error accumulation block 12, the size 2m is calculated by accumulating errors 72; with weights

M

KO:am A HRO, iy

In block 13, an estimate of the acceleration of I is formed by normalizing the accumulated amount in accordance with (8).

The obtained value of a ; What characterizes the intensity of the maneuver of the target, enters the block of Kalman filter of the third order 2 and is taken into account when smoothing the trajectory of the maneuvering target, thereby significantly reducing the dynamic tracking errors.

Since the maneuver is detected with a certain delay of its "itp x Im", then in the Kalman filter of the third order this is taken into account by correcting the extrapolated values of the parameters.

In the technical implementation of the proposed device, the following typical nodes and blocks can be used.

Kalman filters of the second and third orders (blocks 1 and 2), the algorithm of which is described in I), are used as optimal smoothing filters of the deterministic trajectory and the trajectory of the target in the maneuver area, respectively.

The sampling unit 5 consists of two shift registers that sequentially store two adjacent samples of tracking errors 24) and 2(|-M) from the "sliding window" and can be performed on the basis of typical integrated circuits |41.

Rank calculation blocks 6, rank statistics calculation 7 and boundary block 8 can be performed as in the scheme of the rank detector |Z, fig. 8.2| with the use of M comparison circuits, an inversion counter, a memory device, a multiplier, a storage adder and a digital comparator based on typical integrated circuits (41.

The sample removal unit 10 is a set of M matching schemes, through which, on the pulse of exceeding the sampling threshold of values 2, are recorded in the memory device 11.

The storage device 11, the error accumulation unit 12 and acceleration estimates 13 ensure the execution of arithmetic operations of summation, multiplication and division and can be performed on integrated circuits (41.

The technical result that can be obtained in the implementation of the invention is as follows.

The use of a non-parametric rank correlation test for target maneuver detection is more effective compared to parametric methods with a changing or unknown a priori error distribution law, because it provides a stable probability of false alarms (ZI. At the same time, the nature of the shift of the non-centrality parameter op(b) is taken into account, which increases the probability of detecting a maneuver.

Changing and accounting for the intensity of the maneuver (acceleration of the target) makes it possible to reduce the dynamic errors of the escort, which can be estimated by the ratio (5).

SOURCES

1. Kuzmin S.3. The basis! designing systems for digital processing of radar information. - M:

Radio and communication, 1986. - p. 181. 2. Grishin Yu.P., Kazarynov YuM. Dynamic systems, resistance to failures. - M.: Radio and communication, 1985. - p. 163. 3. Detection of radio signals / Ed. A.A. Kolosova, - Moscow: Radio and communication, 1989. - p. 165. 4. Digital and analog integrated microcircuits!. Directory / Ed. S.V. Yakubovsky - M: Radio and communication, 1989.

CT Blocks of maneuver intensity 9.

AND

' 10 11 12 LE) ' b.| Block diagram Memorized - block of block estimates | 8! - sampling and gathering and accumulation of acceleration and device errors

FROM

Sho moish not non nt nova t t nia sho shi nn nm - - «(8 «С «С - o 9 o Z ZO - - («С OZ - Z « ( (| R! 1 o -

Filter block Filter block E A

Uv Kalman Kalman He, order of order

OT Maneuver block detection. 4!

AND.

With b 7 Ya

Block Block B Block 5 | Threshold |! the formation of the calculation of the calculation of block h 'rank'' of samples of the rank of statistics 1

Fig. 1

Ac

AND

- o Shk -43 them - el dn cha y A Y 0 -ya--u Pr sh-y li nn vysh ta. patina mm hundred battle S i

Fig. 2

Claims (1)

An adaptive tracking device for maneuvering targets, containing serially connected blocks of a second-order Kalman filter, a third-order Kalman filter and a switch, which is distinguished by the fact that a maneuver detection unit and 1 maneuver intensity assessment unit are additionally introduced, and the maneuver detection unit consists of a series of connected sample formation block, rank calculation block, rank statistics calculation block 1 threshold block, and the maneuver intensity assessment block consists of the serially connected sample removal block, memory device, error accumulation block 1 acceleration estimation block, while the input of the device is the input of the second-order Kalman filter block, the second output of the second-order Kalman filter block is connected to the input of the sampling block and to the second input of the rank calculation block, the first output of the second-order Kalman filter block is additionally connected to the second input of the switch, the output of the threshold block with connected to the third input of the switch unit and the second input of the de-sampling unit, the second output of the sampling unit is connected to the first input of the de-sampling unit, the output of the acceleration evaluation unit is connected to the second input of the third-order Kalman filter unit, and the output of the device is the output of the switch.