JPH01223377A - Radar tracking filter system - Google Patents

Abstract

PURPOSE:To decrease error of estimated speed remarkably during the turning of a target, by arranging a filter to estimate and forecast a speed of the target with actual small variations as intact without separation into X and Y components. CONSTITUTION:A speed estimating/forecasting filter 65 estimates and forecasts a speed of a target without being separated into X and Y components. An angle of advance estimating/forecasting filter 66 calculates an angle of advance of a target to estimate and forecast. Forecast values of a speed and an angle of advance obtained by the filter 65 and the filter 66 are used to forecast positions of the X and Y components at the subsequent time with predictors 63 and 64. Then, estimators 61 and 62 estimate positions of X and Y components at the present time based on forecast positions of the X and Y components predicted using forecast values of the speed and angle of advance at the preceding time.

Description

[Detailed description of the invention]

[Industrial Application Field] This document #4 relates to a radar tracking filter method, and in particular aims to improve the next scan position prediction tracking performance of a target in a radar tracking processing device for a target acquired by an electronic scanning antenna of a monopulse radar. This invention relates to a radar tracking filter method. [Prior art] Conventionally, predictive filtering such as a Kalman filter and radar two-tugging filters of this type use % polar coordinate range and θ (azimuth) to rectangular coordinates X and Y.
The position and velocity were estimated and predictive filtered using the X and Y components. The dAz diagram is a professional diagram of the conventional radar tracking filter, and the coordinate f? which converts polar coordinate data to rectangular coordinate data. #Vessel1. Estimator (X component) 67,
Estimator (Y component) 6B. Predictor (X component)69. Predictor (Y component) 70. It is configured to include a speed prediction/estimation filter 71 and a coordinate inverse transformer 4. An example of a target model for such a radar tracking filter will be described. First, the coordinate transformation formula, system state Manpuku formula, and observation equation are first-order 11) and t2) formulas, respectively. These are formulas (3) to (6) and formulas (7) to H. [1] Standard conversion formula % formula %) [2] System-like state method X (n+1)=X (river+T @Vx tnl
・・・・・・・・・・・・・・・・・・＜3)Y (
n+1)-Y (n)+T-My(n)...
・”-”・(4) Vx(n+1)=Vxln)+wxt
n) ・・=−−−−−・−・＜5)VY
(n + 1) = Vy (n) + Wy (D)
...Song...-16) [3] Observation method % formula % (71 Here, each symbol represents the data at time n, the next content of g. Rin): Range θ-): Azimuth (Azimuth with north as 0 radians) RMin): Range radar observation value θ'('') * Azimuth radar observation value X (n): X coordinate position Yin): Y coordinate position XM (11): X coordinate position Radar observation value YM(n)
: Y-coordinate position radar observed value Vx(n): X-component velocity vY(n): Y-component velocity Wx l”) @X-component system disturbance W7 (n)
: Y component system disturbance VX in) : Taking the model as an example, a conventional radar tracking filter can be expressed as follows. Regarding the estimator (X component) 67 and (Y component) 68. Intersection (・1・) = total (・I・−1) 10KX(n)・(xm
(n)-Jn(・1・-1))・・・・・・・・・・・・
・(15) 9(nln)=Q(nln-1)+Ky(n)-(Ym
(n)-Q(nln-1))・・・・・・・・・・・・
-(16) Predictor (X component) 69. (Y component) Fi for 70, ex (n l n ) = Qx (n In-1) + K
vx(n)-9y(n l n)=Qy(p l
n-1)+KYYln)-(YMln sound (・1・-1
)) ...(18) In addition, (13), (14),
(17), (18) Equations ri, speed prediction/estimation filter 7
It also applies to 1. The symbols used in the above-mentioned equations (11) to (18) are quartic, and the inverse transformation processing in the coordinate inverse transformer is obtained based on the following equations [4].・・・・・・・・・(19) #(n+t In)=sin ”(Q(n+t+n)/
Q(n+t In))・・・・・・・・・(20) Q(nln)=sin ”(Q(nllll)A(nln
) - (22) The speed output from the speed prediction/estimation filter 71 is calculated as follows. The predicted value of this ζ is △ θ(n+1ln): where n+ is the predicted value of the azimuth at time n/\θ(nln): the estimated value of azimuth at time n at time n. point in time. ,picture. . time, aK, and constant value, which gives the initial conditions for the following equations (24) to (27).
・1・)=(YMtt)−YM(0))/T ・・
...(27) [Problem to be solved by the invention] The conventional radar tracking filter method described above has 1
There are two drawbacks as follows. tl+ Data because the first constant velocity linear measurement is performed when predicting the target position for the next scan. The disadvantage is that a prediction error occurs when the vehicle turns. +2) In filtering the speed of Tari and G,
To run separately into components and X components. The disadvantage is that the error increases when the target turns. A more detailed explanation of the above two drawbacks is as follows. That is, for (1) %471
4. Even under the best conditions where observation disturbances and system disturbances are zero, when the data is turning as shown in Figure 3, it is predicted that the data will deviate from straight to the outside due to constant velocity direct-to-center measurements. Put it away. Regarding (2), pilots always tend to dislike changing their speed, and the target's speed does not change by +9 even when turning. Therefore, it is better to filter-link the speed without dividing it into X and X components, but in the conventional method, it is divided into X and
．． The estimation error becomes large. tb, Even though the speed is often constant when turning, if you look at each X component and X component. One component is acceleration and the other component is deceleration, which is clearly disadvantageous for estimation. It is better to choose a filtering variable that takes a constant value with little change. In the radar tracking filter method, which uses a prediction filter to estimate the tracking and calculate the predicted value for the target model converted to coordinates,
A speed estimation/prediction filter that estimates and predicts the target's speed without separating it into the X component and the X component, and a heading angle estimation/prediction filter that calculates the target's heading angle and estimates/predicts it! g1 A predictor (Predicted) that predicts the X component and the position of the X component at the next point in time using the predicted values of speed and heading angle by the speed estimation/prediction filter and the heading angle estimation/prediction filter; An estimator (GST) that estimates the current position of the X component and the X component based on the predicted position of the X component and
imater). [Example] Next, the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram of one embodiment of the present invention. A coordinate converter l that converts polar coordinate data to rectangular coordinates. 4. Coordinate inverse transformer that converts orthogonal coordinates to polar coordinates. Est make (X component) to estimate the position of the X component and the X component at the boundary point based on the predicted position of the X component and the X component predicted using the predicted values of speed and advance angle at the previous point in time.61. (X component) 621 Predictor (X component) 6 that predicts the X component and the position of the X component at the next point in time using the predicted values of speed and advancing angle
3, (X component)64. A speed estimation/prediction filter 65 that directly estimates/predicts the target speed without separating it into X and X components. It is configured to include a heading angle estimation 1 prediction filter 66 that calculates, estimates and predicts the heading angle of the target. Among these components, the coordinate converter 1, the coordinate inverse converter 4, the EST make (X component) 61 and the EST make (X component) 62ri have the same content as in FIG. 2, and the percentage and predictor (X component) 63゜Predictor (X component)
64Fi each share the speed XX component separator 3, and a speed estimation/prediction filter 65 and a heading angle estimation/prediction filter 6.
In case of 6. It is configured to share two delay devices 9 and a speed calculation/advance angle calculator 2, respectively. Est make (X component) 61 is the filter gain kC
filter gain multiplier 5° adders 49, 50 . Equipped with a delay device 9 and a delay device 9,
X component) 621'! , a filter gain multiplier 6. providing a filter gain Ky(n). The configuration includes adders 41 and 42 and a delay device 9. The predictor (X component) 63 shares the speed XX component separator 3 with the predictor (X component) 64. Observation time interval multiplier 10. Includes an adder 51, a predictor (X component) 64Fi, a predictor (X component) 6
3 and the velocity XX component separator 3, and the observation time interval multiplier 10. The configuration includes an adder 43. The speed estimation/prediction filter 65 includes a heading angle estimation/prediction filter 66, two delay devices 9, and a speed calculation/advancing angle calculator 2.
A filter gain multiplier 7 which shares the filter gain f(v(n-1)) and a delay unit 9 of adders 44 and 45, leading angle estimation-prediction filter 6Fi, speed estimation/prediction Filter 65, two delay devices 9 and speed calculation
The advancing angle calculator 2 is shared, and the filter gain is set to ψ(n-s
) filter gain multiplier 82 providing multiple multiplier 40
．． The configuration includes two delay devices 9 in addition to adders 46 to 48. Next, first, examples of the filter model of the embodiment shown in FIG. 81 are shown in the following (6), (7), and (8). [6] Coordinate transformation formula The above-mentioned good method 11) is shown by equation (2). X(n)=R(n) sin #(n)
・・・・・・・・・・・・・・・(1) Yln
)=R(n) cosθ(n) −・・
..." ...t2) [7] System state method Equations (3) and 14 above and the following (28) and (
29) It is shown by the formula. X(n+1)-αn)+T-Vx(n) ”°=
°(3) Y(n+t)=Y(n)+T-Vy(n)
“””(4)V(n+1)=V(n)+w(n)
＋＋＊＋＋＋m (28)ψ(n+1)=
4p(n)+w'(n) ””(29)
[8] Kansei Fang Yang style The above-mentioned formulas 17) to (10) and the following (30) and (3
1). It is represented by the formula (J2). X1a(n) =RM(n) sinθM(n)
−−(7)=X(n)+vx(n)・”−
(8) YM(n) =RM(n) cosθM(II
) ・−・・・(9)=Y(n)+vY(n)
・・・・・・(10) Vw(ns)=
Xm(n) −XM(n−direct))”+(YM(n)−
YM(11-t))1・・・・・・・・・(30) =V(n-*)+v(n-1) ・-・-・-(3
1) 9M(nt)=ψ(nt)+v'(nt)
-(32) From these observation equations, the following holds true. Equation (33) above is meaningful as a formal formula for calculating ψu(n), and ψu(n) can be obtained from θ~2π.
The cases are divided based on the magnitude relationship between Xm(n) and XM(n-1) and Yhi(n) and YM(n-s)' in order to unify the range. Here, the symbols (6), (7), and (s) above are as follows. V(n): Target speed w(n) at time n
: System disturbance ψ(') added to the speed of the target at time n *Advancing angle of the target at time '(north is O tesian) w'(n) : System disturbance VM added to the advancing angle of the target at time n ( n-t): Observed value of the target's velocity at time n-1 v(n-1): Observed disturbance of the target's velocity at time n-1 ψu(nt): Fi+ Observed value of the target's velocity at time 1 Observed value v'(n-1): observed disturbance of the advancing angle of the target at time n-1. Note that the other variables mentioned above are the same as those in the description of the conventional technique. The above-mentioned equations (30) and (33) indicate the processing contents of the speed calculation/advance angle calculator 2 shown in FIG. Taking the above model as an example, the filter of the present invention is an estimator, a gradictor, a speed estimation e-prediction filter, and a progression filter represented by the following equations (9), (10), (11), and [12], respectively. It is configured with the functions of an angle estimation/prediction filter.

[9] Estimator......(16) (VM(n-1)-#(n-11n-t))...
...(40) Gu (nln) = Gu (n-i l n) River・
・” (36) [12] Advance angle estimation/prediction filter △ ψ (n-11n) == ψ (n-t l n-s ) 1 ψ (n-t) - Δψ (n-t) = 29 (n-11n)-
ψ(n −2In-t)...(37) (12) mentioned above
, (9(n-tln)Δ1ψ(n-211-) in equation (37) represents the amount of change in the advancing angle at the previous point in time. The amount of change in the advancing angle at the current point in time is set to be equal to the amount of change in the advancing angle at the previous point in time.This can be expressed as the following equation: △ ψ(rlln)−ψ(n −tln)=ψ(n−tln)
-ψ(n-1!1n-1) The left side of the equal sign in the above equation is the amount of change in the advancing angle at the current moment. This is the amount of change in the advancing angle at the time before the right side Fi. Each symbol (10) to [12] described above indicates the following content.仝(.In) :V(.). . Time, place, time, value △ ψ(nln): Predicted small amount of ψ(fl) at time n x(nln): Predicted small amount of VX(n) at time n・(・1・): Vy(・) Plan 11 at the time of
111 value Kv(n-r): Filter gain of 9(n-tln) Kq(nt): Filter gain of ψ(nt-t In)Other variables are the same as in the description of the conventional technology. It is. In the embodiment of FIG. 41, the estimator (X component) 6
1. (Y component) 62 is as mentioned above.

Process the contents of [9],
Also, the predictor (X component) 63° (Y component) 64Fi
The contents of (10), speed estimation/pre-1 filter 65 are [1
The contents of [1] and the advancing angle estimation/prediction filter 66 are [12]
Process the contents of each. Now, in equation (37), when 9(nln)<00, 91(nln) is converted to 91(nln)
When n)+2r△ (nln))2π, ψ(nln) is changed to ψ(nln)
-2π and its possible range from O to 2π. Also, in equation (41), when 9m(nt)-g+(n-tln-t), )r, Δ
ψ(nt)==pm(n-1)-91(n-I In
-t )-2gψM(n-s)-9+(nt In-
L) (at -t -Δψ(n-1)m9 diagram (nt)-
ψ(n-tln-1)+2π In cases other than the above, Δψ(n-h)-=g+m(n-1)-? (n-1
111-1), 9M(n-1) and ψ(n-tln-
1) is also divided across the north, Δψ(n-5)
Make sure that the angle difference is correct. When △ △ ψ(n-tln) is ψ(n-th n)+2r△ ψ(n-rln))2π. △ △ g+(n-1ln) is unified to ψ(nt In)-2g within the range of 0 to 2. Next, the operation of the embodiment shown in Fig. m1 will be explained. Data at time n as range data. When the range observation data RM (nl 12 and azimuth observation data θy (n) 110) of
)14 is obtained. The processing content of the coordinate converter l is the observation equation (7) of [8].
) and (9). After passing this XM[n) 13 and the adder 49 of the delay device (DWLAY) 61 of the predictor (X component) 63, an appropriate filter gain Kx
(Multiply nl by the filter gain multiplier 5 to obtain the estimated value X(n In) l 5 of △ (15). The filter gain Kx(n) can be calculated using a method such as a Kalman filter. In the same way, the estimated value (・1・) 16 of equation (16) is obtained from the estimator (Y component) 62. Up to this point, it is the same as the conventional technology, but firstly, the maximum value of the present invention is We will explain the content of filtering without separating the target speed into iXY components, which is a feature of , and the content of filtering by calculating the advancing angle of the target. XM(n-1) passed through DWLAY9 at the input end of 2. Using the data of YMln) 14 and YM(n-5), the speed calculation/advance angle calculator 2 calculates the target speed III I13 data VM(n-t) 21 and the observation data ψM of the advance angle. (III-t)24 is obtained. The processing contents of the speed calculation/advance angle calculator 2 are as shown in equations (30) and (33). In speed estimation-prediction filter 5, VM(It-t)
21 and the output of DWLAY9 11Ja[V(n-i
In-5) is subtracted by the adder 44, and then multiplied by the filter gain Kv(n-1) by the filter gain multiplier 7.
40) Obtain the estimated value V(n-1In)22 of the formula Δ. On the other hand, similarly, in the advancing angle estimation/prediction filter 66, 9M(n-1)
/\24 and the estimated value ψ(n-tln-
Based on the estimated value ψ(n
-tln) 25 is obtained. Based on these estimated values, t! Single NK (36) Formula K j-) Te9 (nln)
231>i, advancement Δ As for the angle, ψ(nln)26 is obtained as a predicted value by equation (37). Note that the 1 multiple multiplier 40 is necessary for this speed prediction process. Regarding the speed, it rarely changes even when turning, so it is as shown in equation (36).As for the advancing angle, it changes when turning, so the change at the @ time is the current taste. Thus, the predicted negligible (nln)23,9)(nln)2
6 is obtained without going through the X and Y separation processing of the velocity components, the first order is determined by the velocity XY component separator 3 of the predictor (39
) can be obtained by the formula. Observation time interval multiplier l of predictor (X component) 63 & (Y component) 64 is applied to these values.
The Yang interval during the scan at O, that is, the observation 32, and the predicted value ゛Q
(n+11n)29,'(n+tIn)30 are (19), (20), and (21), respectively. It is obtained by equation (22). Next, let's discuss the operation using actual data examples. Let's take an example of when a target turns.1 Normally, a target often turns at almost the same turning angle during each turn without changing its speed. Therefore, if we consider the case where the system disturbance #observation disturbance is zero, 1g3
As can be seen from the example in the figure, in the conventional method, the predicted position is always predicted outside the trajectory, whereas in the method of the present invention, the prediction only deviates at the beginning and end of the turn. Other periods correspond exactly to the actual position. Furthermore, when the magnitude of the actual radar observation disturbance and the target system disturbance are added, the results of the combination simulation also show that the predicted value during turning is significantly improved compared to the conventional case. Also, because the accuracy of the predicted values is good, the estimated values are also better than before when turning. Furthermore, since the method of the present invention does not separate the estimated value of speed into X and Yd, it is possible to apply a deep filter, that is, reduce the filter gain to increase the degree of smoothness with respect to human power, and estimate It has also become possible to reduce errors. [Effects of the Invention] As explained above, the present invention has a filter that directly estimates and predicts the target speed, which has little actual variation, without separating it into the X component and the Y component. This has the effect of significantly reducing the error in estimating the speed when the target is turning. In addition, there is a filter that calculates the target's advancing angle and estimates/predicts it, a predictor that predicts the position of the X component and Y component at the next point in time using the predicted values of the speed and advancing angle, and a X component, Y predicted using the predicted value of the angle
By having an estimator that estimates the current position of the X component and Y component based on the predicted position of the component,
The error in the predicted position at the next point in time when the target is turning can be significantly reduced, and since the prediction degree is estimated based on a good predicted value, the error in the estimated position when the target is turning can be significantly reduced. There is an effect.

[Brief explanation of the drawing]

Fig. 1 is a schematic diagram of one embodiment of the radar tracking filter of the present invention, Fig. 2 is a 10-step diagram of a conventional radar tracking filter, and Fig. 3 shows the percentage of observed disturbance and system disturbance. FIG. 4 is a prediction characteristic diagram showing an example of prediction for a turning target by the conventional method and the method of the present invention in comparison when there is no turning target. l... Coordinate converter 22... Velocity calculation
Advance angle calculator, 3... Speed XY component separator, 4
...Coordinate inverse transformer, 5 to 8, 34.35...
... Filter gain multiplier, 9 ... Delay device,
10...Observation time interval multiplier, 11...
・Radar observation value of target range at time n, 12・
... Radar observation of the azimuth of the target at time n, m, 13... Dege at time n. Target's X-coordinate position radar observed value, 14. Old...Target's Y-coordinate position radar observed value at time n, 15...X
Estimated value of (n) at time n, 16...Y
Estimated value of (n) at time n, 17...X(
n-H) at time n 611f11.18...
...Predicted value of Y(n+t) at time n, 19...
...X in) n- is the predicted value at the time, 20
...Y(n)On- is the predicted value at the time, 2
1... Observed value of the speed of the target at time n-1, 22... Reset value of V(nt) at time n. 23...Predicted value of V(n) at time n, 2
4...n- is the measured value of the advancing angle of the target at the time point, 25......the estimated value of ψ(n-s) at the n time point, 26......ψ( 1m) at time n, 27...Predicted value of Vx(n) at time n, 28...Predicted value of Vτ1fl) at time n, 29... The predicted value of Rl (n+1) at time n, 30...the predicted value of θ(n+x) at time n, and the estimated value of 31...Rln) at time n. 32...θ(estimated value of nJ at time n, 3
3...Velocity synthesizer, 36...Vx(n
) estimated value at time n, 37...MY(n
) at time point n. 38...Predicted value of VX(11+1) at time n, 39...■Predicted value of y(n+1) at time n, 40...Multiple multiplier, 41- 53・
... Adder, 61 ... Estimator (
X component), 62... Estimator (Y component)
, 63... Predictor (X component), 64...
...Predictor (Y component), 65... Speed estimation/prediction filter, 66... Advance angle estimation/
Prediction filter, 67... Estimator (X component), 68... Estimator (Y component), 6
9... Predictor (X component), 70...
...Predictor (Y component). Agent Patent Attorney Susumu Uchihara

Claims (1)

[Claims] In the radar tracking filter method, which calculates tracking estimation and predicted values using a prediction filter for a target model converted from polar coordinates to rectangular coordinates in radar tracking processing, the speed of the target is estimated without separating it into X and Y components. - A speed estimation/prediction filter to predict, a heading angle estimation/prediction filter to calculate and estimate and predict the heading angle of the target, and speed and heading angle by the speed estimation/prediction filter and the heading angle estimation/prediction filter. A predictor that predicts the position of the X component and Y component at the next point in time using the predicted values of A radar tracking filter system characterized by comprising an estimator that estimates the current position of the X component and Y component based on the current position.