JPH09133750A - Target motion analysis method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To facilitate the convergence of a solution at the time of analysing a target motion by solving a nonlinear equation. SOLUTION: A received signal corresponding to the radiation sound of a target is inputted to a bearing information calculation part 22 and a frequency information calculation part 23, thereby computing an observation azimuth θm(t) as well as an observation frequency component νkm(t) for the delivery thereof to an initial value calculation part 24 and a maneuver detection part 25. In addition, the maneuver detection part 25 sends the estimated value of VXdo* of an inner state variable and an estimated information matrix Σvxdo*<-1> to the initial value calculation part 24, with a component related to the velocity of the estimated value VXo* of the inner state variable set at zero upon the occurrence of a maneuver, respectively to the initial value calculation part 24. This initial value calculation part 24 solves a pseudo-linear equation formed out of information and observation variable for the information, using the estimated value VXdo* and the matrix Σ vxdo* as foreseen information, and finds an analysing initial value VXoo* for the delivery thereof to an inner state variable estimation part 26. This inner state variable estimation part 26 analyzes the motion of the target on the basis of an evaluation function governed by the initial value VXoo*.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention receives a sound radiated from a target by a receiver sensor array attached to a moving body such as a movable ship, and from an observation amount disturbed by noise,
The present invention relates to a target motion analysis method for estimating internal state quantities related to the position and speed of a target that moves relative to each other.

[0002]

2. Description of the Related Art Conventionally, techniques in such a field include:
For example, there was one shown in the following document. Reference 1; ACASSP, 1989 IEEE, JMPassrieux, D.Pillion,
P. Blanc-Benon and C. Jauffret “Target Motion Anal
ysis with Bearing and Frequencies Measurements via
Instrumental Variable Estimator ”P.2645-2648 Reference 2; GEC JOURNAL OF RESEARCH, 3 [3] (198)
5) MS Woolfson “An Evaluation of Manoeuvre D
etector Algorithms "P.181-190 Fig. 2 is a geometrical explanatory view showing a conventional target motion analysis method, showing an observation system and a motion system. (X, Y) in Fig. 2 is the origin o. Is a fixed coordinate system of the vehicle, and the vehicle 1 and the target 2 are shown in the fixed coordinate system.
₁ (t) is the position vector of the vehicle 1 at time t, Vx ₂
(T) is the position vector of the target at time t, and r (t) is the distance between the vehicle 1 and the target 2 at time t ‖Vx ₂ (t) −Vx
₁ (t) ‖ and θ (t) respectively indicate the azimuth angle with respect to the Y axis of the target 2 as viewed from the vehicle 1 at time t.
Here, “‖‖” represents the norm of the vector. The target motion analysis assumes that the target 2 is performing a uniform linear motion,
The position and speed of the target 2 are identified from the observation time series of the azimuth angle of the sound source disturbed by noise with respect to the target 2 and the frequency of the received sound. Before explaining the conventional technique, the operation principle of the target motion analysis method will be described. The observation value including the observation value of the azimuth angle θ (t) at the time t and the observation noise of the Doppler shift frequency ν _k (t) of the k-th natural frequency component f _k of the sound source are expressed by the following equation (1). Represent.

[0003]

(Equation 1) Here, the relative position vector Vx (t) seen from the observer on the moving body 1 with respect to the target 2 is represented by the following equation (2), and each time t =
The internal state quantity VX (t) at t ₁ , t ₂ , ..., T _n is defined as in equation (3). However, T represents transposition,
"'" Represents time derivative. That is, the internal state quantity VX (t) is
It includes the position and velocity of the target 2 and the sound source natural frequency component of the target 2. _{Vx (t) = Vx 2 (} t) -Vx 1 (t) = [r x (t) r y (t)] T ··· (2) VX (t) = [Vx (t) T Vx '( when you put the internal state the amount of VX ₀ in _{^{t) T 1 / f 1 ...}} 1 / f p] T ··· (3) time t = t ₀ and _{VX (t 0), VX (} t) and the relationship of VX ₀ Is given by the state transition formula shown in (4) below.

[0004]

(Equation 2) The internal state quantity VX (t) is estimated at times t = t ₁ , t ₂ , ...,
This is performed by obtaining VX ₀ that minimizes the evaluation function L (VX ₀ ) of the following expression (5) composed of a set of observed values at t _n .

[0005]

(Equation 3) That is, it is assumed that the motions of the vehicle 1 and the target 2 are much smaller than the sound velocity c. Also, "E []" is
An ensemble average is shown.

In minimizing L (VX ₀ ), the following (7)
It is necessary to solve the following non-linear equation using the iterative method, and it is necessary to determine the initial value VX ₀₀ for that. Therefore, assuming that the observation error is minute, the equation (8) is derived. ∂L (VX ₀ ) / ∂VX ₀ = 0 (7) Vb ＝ MB VX ₀ ＋ MW (VX (t) VN ・ ・ ・ (8)

(Equation 4)

(Equation 5) In Reference 1, by using the approximate weight matrix MW ^{* in} which the unknown elements of the weight matrix MW (VX (t)) of this equation are replaced with appropriate setting values, the target motion analysis is performed by solving the pseudo linear equation for VX _0. Analysis initial value of VX ₀₀ ^* VX ₀₀ ^* = (Σ ^-1/2 MW ^{* -1} MB) ⁺ Σ ^-1/2 MW ^{* -1} Vb ... (10) is determined. Where “+” represents the generalized inverse matrix,
Σ ^-1/2 is the inverse of the upper triangular matrix determined by the Cholesky factorization of Σ. In the subsequent target motion analysis, the evaluation function L of the equation (5) is set with VX ₀₀ ^* given by the equation (10) as an initial value.
The optimal internal state quantity VX ₀ ^* that minimizes (VX ₀ ) is
It is obtained by solving the non-linear equation (7). Further, the estimated information matrix Σ _{VX0 *} ⁻¹ of the obtained estimated value VX ₀ ^* of the internal state quantity is defined by the following equation (11), and the estimated information matrix Σ _{VX0 *} ^{−1 at} time t is (4 (12) is obtained from the relationship of ().

[0007]

(Equation 6) According to Reference 2, when the target is maneuvered such as changing needles or shifting, the assumption conditions for target motion analysis are not satisfied. Therefore, the statistical distribution of the estimated residual Vε in Eq. Detect. Vε = VZ _m −VZ (VX ₀ ^* ) (13) If foreseeing information such as some value or its accuracy is obtained for the target state quantity VX ₀ , these foreseeing state quantities and their state quantities The information matrices of VX _d0 ^* and Σ _{VX0 *} ^-1 , respectively, are used to minimize the evaluation function containing the foresight information in the following equation (14), and the internal state quantity VX utilizing the foresight information is obtained.
Estimate (t). H (VX ₀ ) = (VZ _m −VZ (VX ₀ )) ^T Σ ⁻¹ (VZ _m −VZ (VX ₀ )) + (VX ₀ −VX _d0 ^* ) Σ _{VX0 *} ⁻¹ (VX ₀ −VX _d0 ^* ) (14) VX _d0 ^* is used as the initial value when the equation (14) is minimized, but an appropriate value is assumed for the component whose foreseeing information is unknown. For example, maneuver at time t = t _d is assumed that the detected, estimated information matrix of the estimated state quantity relating to the relative position and the natural frequency of the target 2 at time t _d and its state amount after the time t _d motion analysis Can be considered as foresight information in. Therefore, assuming that the components related to the estimated state quantity and the velocity of the estimated information matrix are zero are set as VX _d0 ^* and Σ _{VX0 *} ^-1 , respectively, and motion analysis is performed using observation values after time t _d. initialize. That is, the maneuver detection time t _d is set again to t = t _0, and VX _d0 ^*
With V as the initial value, VX ₀ that minimizes equation (14) is found,
Then, the internal state quantity VX (t) is estimated.

FIG. 3 is a functional block diagram of a desired motion analysis device used in a conventional desired motion analysis method. The target motion analysis device in this functional block diagram includes an input terminal 1
1, an azimuth information calculation unit 12 and a frequency information calculation unit 13 for inputting a reception signal from a receiver array (not shown). The azimuth information calculation unit 12 calculates the arrival direction of the radiated sound of the target 2 from the received signal, and the frequency information calculation unit 13 calculates the frequency component of the radiated sound of the target 2. The output sides of the azimuth information calculation unit 12 and the frequency information calculation unit 13 are both connected to the initial value calculation unit 14 and the maneuver detection unit 15. The initial value calculation unit 14 has a function of calculating the initial state amount from the azimuth information and the frequency information, and the maneuver detection unit 15 includes the azimuth information calculation unit 12 and the frequency information calculation unit 13 which provide the azimuth information and the frequency information. Therefore, it has a function of detecting the maneuver of the target 2. An internal state quantity estimating unit 16 is connected to the initial value calculating unit 14 and the maneuver detecting unit 15. The internal state quantity estimating unit 16 analyzes the target motion, and the analysis result is output via the output terminal 17. Next, the operation of the apparatus of FIG. 3 when the position information of the target 2 at the time of detecting the maneuver is used as the foreseeing information The sound emitted from the target 2 is received by the receiver sensor array,
The received signal is input from the input terminal 11. The received signal is input to the azimuth information calculation unit 12 and the frequency information calculation unit 13. The azimuth information calculation unit 12 calculates the target 2 from the received signal.
The observed azimuth angle θ _m (t) is calculated, and the frequency information calculation unit 13 calculates the Topra shift observed frequency component ν _km (t) (k =
1, 2, ..., P) are calculated, and the calculation results are sent to the initial value calculation unit 14 and the maneuver detection unit 15. Each time t =
n sets of observation azimuth angles θ at t ₁ , t ₂ , ..., T _n
m (t) and Doppler shift observation frequency component ν _km (t) (k = 1,
2, ..., P) is input to the initial value calculation unit 14, the initial value calculation unit 14 uses (10) to calculate the initial value VX of the motion analysis.
₀₀ ^* is obtained, and the initial value VX ₀₀ ^* is used as the internal state quantity estimation unit 16
Send to The internal state quantity estimation unit 16 determines the estimated value VX ₀ ^* for the internal state quantity by solving the nonlinear least squares problem of the equation (14) with VX ₀₀ ^* as the initial value of the motion analysis. Then, the internal state amount estimation unit 16 determines the internal information amount Vx of the target motion.
₂ estimate Vx ₂ ^* to calculate a (t) of (t), together and outputs the result from an output terminal 17, the estimated value for the internal state quantity
VX ₀ ^* is sent to the maneuver detector 15.

In the next step, at each time t = t _{n + 1} ,
Observation azimuth θ _m (t) and Doppler shift observed frequency component ν _km (t) of (mn) pairs of t _{n + 2} , ..., T _m (> t _n ).
When (k = 1,2, ..., p) is input to the initial value calculation unit 14 and the maneuver detection unit 15, the maneuver detection unit 15 calculates the statistical distribution of the estimated arithmetic error defined by the equation (13). Find out. As a result, the occurrence of maneuvers, for example, is detected.
When the maneuver detection unit 15 detects a maneuver, the maneuver detection unit 15 calculates the estimated information matrix at the maneuver detection time point (t = t _d ) using the equation (12),
The estimated value of the internal state quantity and the estimated information matrix in which the velocity-related component is zero are VX _do ^* and Σ _{VX0 *} , respectively _.
^-1 and send it to the internal state quantity estimating unit 16. The internal state quantity estimation unit 16 uses the estimated value VX _do ^* and the estimated information matrix Σ _{VX0 *} ⁻¹ sent from the maneuver detection unit 15 as the foreseeing information and _uses the estimated value VX _do ^* as the initial value of the motion analysis. ,
Observation azimuth θ _m (t) and Doppler shift observation frequency component ν of (m−d) sets at time t = t _{d + 1} , t _{d + 2} , ..., T _{m.
For km} (t) (k = 1,2, ..., P), target motion analysis that minimizes the evaluation function of equation (14) is performed. Also, if the maneuver is not detected by the maneuver detection unit 15, the time _{t = t 1, t 2,} ..., estimated value of the internal state of the VX ₀ ^* as an initial value of the motion analysis obtained in t _n, t = T _{n + 1} , t
_{n + 2, ..., t m} (> t n) in the (m-n) sets of observation azimuth θ _m (t) and the Doppler shift observed frequency component [nu _miles
For (t) (k = 1, 2, ..., P), the target motion analysis that minimizes the evaluation function of equation (5) is performed.

[0010]

However, the conventional target motion analysis method has the following problems. Except when the foreknowledge information about the internal state quantity is obtained for all components, the state quantity VX _d0 ^* assuming an appropriate value for the component for which the foreknowledge information is unknown is set to the minimum of the evaluation function of equation (14). There is a problem that it takes a long time to converge the solution due to the influence of this assumed component because it is unavoidable to use it as the initial value when obtaining the internal state quantity VX ₀ to be converted. It is an object of the present invention to solve the problem that it is necessary to assume an appropriate value for a state quantity component whose foreseeing information is unknown, and thus it takes time to converge the solution.

[0011]

In order to solve the above-mentioned problems, a first invention receives a sound radiated from a target by a receiver sensor array attached to a movable navigation body, From the azimuth observation amount time series which is the azimuth measurement result with respect to, the initial value of the motion analysis of the target is determined on the assumption that the target performs a uniform linear motion, and the state quantity related to the position and velocity of the target is determined. In the target motion analysis method for estimating, the following method is taken. That is, in the target motion analysis method of the first invention, when the foreseeing information including the approximate value and the accuracy of the state quantity of the target is obtained, the foreseeing information, the azimuth observation amount, and the azimuth observation amount. From the angle measurement accuracy, the velocity component of the vehicle and the time series of the position of the vehicle, a pseudo-linear equation in which the observation noise component is considered to be minute is constructed, and the pseudo-linear equation is solved to An analysis initial value including the contents of the foresight information is set as an initial value of the motion analysis, and a nonlinear least square method is solved by repeatedly performing the calculation to estimate the state quantity related to the target position and speed.

A second aspect of the present invention receives the sound radiated from a target by a receiver sensor array attached to a moving body capable of moving, and obtains a time series of azimuth observation amount which is the azimuth measurement result with respect to the target and From the frequency observation amount time series that is the Doppler shift frequency measurement result of the radiated sound, the target performs a constant-velocity linear motion with respect to the state amount related to the position and velocity of the target and the state amount related to the natural frequency of the sound source of the target. In the target motion analysis method for estimating and estimating the target motion analysis on the assumption that the target motion analysis is performed, the following method is taken. That is, when the foreseeing information including the approximate value and the accuracy of the target state quantity is obtained, the foreseeing information, the azimuth observation amount, the angle measurement accuracy of the azimuth observation amount, and the velocity component of the vehicle are obtained. And a time series of the position of the spacecraft and a time series of the frequency observation amount and the measurement accuracy of the frequency observation amount to form a pseudo linear equation in which the observation noise component is minute, and the pseudo linear equation By setting the analysis initial value including the contents of the foresight information as the initial value of the kinematic analysis, and solving the nonlinear least squares method by iteratively calculating the state quantity related to the target position and velocity and the target sound source. The state quantity related to the natural frequency is estimated.

In a third aspect of the present invention, the foreseeing information in the first aspect of the present invention when a maneuver such as a change in the target needle and gear shift is detected, the target information at the time of the maneuver detection when the component related to the speed is zero. It is composed of an estimated value of the state quantity and an estimated information matrix of the state quantity. Then, an observation noise component from the configured foresight information, the azimuth observation amount after the maneuver detection, the angle measurement accuracy of the azimuth observation amount, the velocity component of the navigation vehicle, and the time series of the position of the navigation object. Is set to be a minute, and the analysis initial value including the contents of the foreseeing information is determined as the initial value of the motion analysis by solving the pseudo linear equation, and the nonlinear least squares method is performed by repeating the calculation. By solving the above, the state quantity relating to the state quantity relating to the target position and speed is estimated.

According to a fourth aspect of the present invention, the foreseeing information in the second aspect of the present invention when a maneuver such as a needle change or gear shift of the target is detected, the target information at the time of the maneuver detection when the component relating to the speed is zero. It is composed of an estimated value of the state quantity and an estimated information matrix of the state quantity. Then, the configured foresight information, the azimuth observation amount, the angle measurement accuracy of the azimuth observation amount, the time component of the velocity component of the navigation vehicle, the position of the navigation vehicle, the frequency observation amount, and the frequency observation. A quasi-linear equation that the observation noise component is minute is constructed from the time series of the measurement accuracy of the quantity, and the analysis initial value including the contents of the foreseeing information is solved by solving the quasi-linear equation to obtain the initial value of the motion analysis. Then, the state quantity related to the target position and velocity and the state quantity related to the target sound source natural frequency are estimated by performing the iterative calculation and solving the nonlinear least squares method.

According to the first to fourth aspects of the invention, since the target motion analysis method is configured as described above, the sound radiated by the target is received by the receiver sensor array, and the observed amount time series of the reception result is used. , The target state quantity is estimated by setting the initial value of the target motion analysis on the assumption that the target performs constant velocity linear motion. Here, when the foreseeing information consisting of the approximate value and the accuracy of the target state quantity is obtained, the foreseeing information, the azimuth observation amount, the angle measurement accuracy of the azimuth observation amount, and the velocity component of the vehicle are obtained. A quasi-linear equation that the observed noise component is minute is constructed from the time series of the position of the vehicle. By solving the quasi-linear equation, the analysis initial value including the contents of the foresight information is determined as the initial value of the motion analysis. The target state quantity is estimated by solving the nonlinear least squares method by iterative calculation. Since the analysis initial value at this time includes the contents of the foresight information,
It is suitable as an initial value for solving the nonlinear least squares method. Therefore, the above problem can be solved.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION A feature of this embodiment is that a pseudo linear equation is solved from a time series of foresight information and an observation amount, and an optimal initial value is set for the time series of the observation amount including the foresight information, Even if a maneuver occurs, excellent target motion analysis can be performed. First, the principle of the desired motion analysis method of this embodiment will be described. Weighted sequence MW of equation (8)
Using the approximate weight matrix MW ^* that approximates the unknown element of (VX (t)) with an appropriate estimated value, the equation (8) is transformed into (1
Equation 5) is obtained. (MW ^* ) ⁻¹ (Vb−MBVX ₀ ) = VN (15) By multiplying both sides of equation (15) by (Σ ^−1/2 ) ^T , equation (16) can be obtained. Here, Σ ^−1/2 is an upper triangular matrix determined by the Cholesky factorization of Σ ⁻¹ and satisfies the following expression (17). (Σ ^-1/2 ) ^T (MW ^* ) ^-1 (Vb-MBVX ₀ ) = (Σ ^-1/2 ) ^T VN (16) Σ = (Σ ^1/2 ) ^T Σ ^1/2・.. (17) If (16) is VN ₀ , E [VN ₀ VN ₀ ^T ] = (Σ ^−1/2 ) ^T E [VN VN ^T ] (Σ ^−1/2 ) = (Σ ^{−1 / 2) T ΣΣ -1/2} = become ^{(Σ -1/2) T (Σ 1} /2) T Σ 1/2 Σ -1/2 = I ··· (18).

Therefore, the equation (8) is VX ₀ using the matrix MW ^*.
Solving the pseudo-linear equation with respect to is equivalent to finding a linear least squares solution that minimizes the evaluation function I (VX ₀ ) defined by the following equation (19). I (VX ₀ ) = ‖ (Σ ^-1/2 ) ^T (MW ^* ) ^-1 (vb-MBVX ₀ ) ‖ ² ... (19) When the formula (19) is transformed, the following formula (20) is obtained. To be caught. I (VX ₀ ) = (Vb−MBVX ₀ ) ^T (MW ^* ΣMW ^{* T} ) ⁻¹ (Vb−MBVX ₀ ) ... (20) Therefore, the estimated state quantity VX _d0 of the internal state quantity is used as the foreseeing information.
^{* And} the estimated information matrix Σ _{VXd0 *} ^{−1 of the} estimated state quantity are known, G (VX ₀ ) = (Vb−MBVX ₀ ) ^T (MW ^* ΣMW ^{* T} ) ⁻¹ (Vb−MBVX ₀ ) + (VX ₀ −VX _d0 ^* ) Σ _{VXd0 *} ⁻¹ (VX ₀ −VX _d0 ^* ) ・ ・ ・ (21) By finding the solution that minimizes the evaluation function, An initial value suitable for a time series can be obtained. If the equation (21) is differentiated by VX ₀ , the following (22)
The linear equation of is obtained, and the initial value of the target motion analysis is VX _00.
Becomes equation (23). {MB ^T (MW ^* ΣMW ^{* T} ) ^-1 MB + Σ _{VXd0 *} ^-1 } VX ₀ = MB ^T (MW ^* ΣMW ^{* T} ) ^-1 Vb + Σ _{VXd0 *} ^-1 VX _d0 ^*・ ・ ・ (22) VX ₀₀ ^{* = {MB T (MW * ΣMW} * T) -1 MB + Σ VXd0 * -1} -1 × {MB T (MWΣMW T) -1 Vb + Σ VXd0 * -1 VX d0 *} ··· (23) now, the time t = when maneuvers it is assumed that detected at t _d, the estimated state quantity and the condition of the estimated information matrix at time t = t _d, what put the ingredients on the velocity zero VX _d0 ^* and matrix sigma _{VXd0 *} ^-1 and these are foresight information. Then, at each time t = t _{d + 1} , t
Equation (23) is derived from the observation direction θ _m (t) of the (m−d) set of _{d + 2} , ..., T _m and the Doppler shift observation frequency or ν _km (k = 1,2, ..., p). Is used to calculate the initial analysis value VX ₀₀ ^* of the target motion analysis after maneuvering. In the calculation of the initial value VX ₀₀ ^* , since the velocity-related component of the estimated information matrix Σ _{VXd0 *} ^-1 is set to zero, it is possible to _remove the influence of the velocity component assumed to be zero in the estimated state quantity VX _d0 ^*. . Also,
The initial value VX ₀₀ ^* contains foresight information regarding the position at the time of maneuver detection, and is also a solution of a pseudo-linear equation for the observation time series after maneuver detection. Therefore,
Convergence can be accelerated when obtaining a least-squares solution that minimizes the evaluation function of Expression (14) as an initial value that makes use of both the prior information and the observation data.

FIG. 1 is a functional block diagram of a target motion analysis device used in the target motion analysis method according to the embodiment of the present invention. The target motion analysis device includes an azimuth information calculation unit 22 and a frequency information calculation unit 23 that input a reception signal from a wave receiver sensor array (not shown) via an input terminal 21. The azimuth information calculation unit 22 calculates the arrival direction of the radiated sound of the target 2 from the received signal, and the frequency information calculation unit 23 calculates the frequency component of the radiated sound of the target. The output sides of the azimuth information calculation unit 22 and the frequency information calculation unit 33 are both connected to the initial value calculation unit 24 and the maneuver detection unit 25. The maneuver detector 25
The maneuver of the target 2 is detected from the azimuth information and the frequency information given by the azimuth information calculation unit 22 and the frequency information calculation unit 23, and the estimated state quantity VX _d0 ^* and the estimated information matrix Σ _{VXd0 *} ^-1 are output. It has a function and passes them to the initial value calculation unit 24. The initial value calculation unit 24
It has a function of calculating the initial state quantity from the azimuth information, the frequency information, the estimated state quantity VX _d0 ^*, and the estimated information matrix Σ _{VXd0 *} ⁻¹ . An internal state quantity estimating unit 26 is connected to the initial value calculating unit 24 and the maneuver detecting unit 25. The internal state quantity estimating unit 26 analyzes the target motion, and the analysis result is output via the output terminal 27.

Next, the operation of FIG. 1 will be described. Here, the case where the position information of the target 2 at the time of detecting the maneuver is used as the foreseeing information will be described. The sound radiated by the target 2 is received by a receiver sensor array (not shown), and the received signal is sent to the azimuth information calculation unit 22 and the frequency information calculation unit 23 via the input terminal 21. The azimuth information calculation unit 22 calculates the observation azimuth angle θ _m (t) from the received signal, and the frequency information calculation unit 23 calculates the Doppler shift observation frequency component ν _km (t) (k = 1,2, ..., p). Then, those results are sent to the initial value calculation unit 24.
N sets of observation azimuths θ _m (t) and Doppler shift observation frequency components ν at each time t = t ₁ , t ₂ , ..., T _{n
When km} (t) (k = 1,2, ..., p) is input to the initial value calculation unit 24, the initial value calculation unit 24 uses the equation (10) to analyze the initial state value VX _00. ^* Is obtained and the analysis initial value VX ₀₀ ^* is sent to the internal state quantity estimating unit 26. The internal state quantity estimating unit 26 determines the estimated value VX ₀ ^* of the internal state quantity by performing the target movement analysis that minimizes the evaluation function of (5) by using the analysis initial value VX ₀₀ ^* as the initial value of the movement analysis. The internal state quantity estimation unit 26 sends the estimated value VX ₀ ^* to the maneuver detection unit 25, and
Using the equations (4) and (2), the internal state quantity Vx of the target 2
₂ (t) estimated value VX ₂ ^* (t) through an output terminal 27 of output.

In the next step, at each time t = t _{n + 1} ,
The observed azimuth angle θ _m (t) and the Doppler shift observed frequency component ν _km (t) (k =) of (mn) sets of t _{n + 2} , ..., T _m (> t _n ).
1, 2, ..., P) is input to the initial value calculation unit 24 and the maneuver detection unit 25, the maneuver detection unit 25 outputs (1
The statistical distribution of the estimated arithmetic error defined by equation (3) is investigated, and the maneuver is detected by the investigation result. When the maneuver detector 25 detects a maneuver, the maneuver detector 25 uses the equation (12) to detect the maneuver t.
= T _d, the estimated information matrix is calculated, and the estimated value of the internal state quantity with the velocity-related component set to zero and the estimated information matrix are set to VX _d0 ^* and Σ _{VXd0 *} ^-1 , respectively, and these are _sent to the initial value calculation unit 24. send. The initial value calculation unit 24 uses the estimated value VX _d0 ^* of the internal state quantity and the estimated information matrix Σ _{VXd0 *} ⁻¹ as the foreseeing information, and at each time t = t _{d + 1} , t _{d + 2} , ..., T _m . (Md)
Using the equation (23) for the observation azimuth θ _m (t) and the Doppler shift observed frequency component ν _km (t) (k = 1,2,…, p), the initial value of the analysis of motion analysis VX ₀₀ ^* And the analysis initial value VX ₀₀ ^* is sent to the internal state quantity estimating unit 26. The internal state quantity estimation unit 26 sets the analysis initial value VX ₀₀ ^* as the initial value of the motion analysis,
A target motion analysis that minimizes the evaluation function of equation (14) is performed. Then, the internal state quantity estimation unit 26 uses the equations (4) and (2) to calculate the position vector Vx ₂ (t) for the target 2.
The estimated value VX ₂ ^* (t) of the internal state quantity corresponding to is calculated and output via the output terminal 27.

If the maneuver detector 25 does not detect a maneuver, the estimated value VX of the internal state quantity is calculated.
_{With 0} ^* as the initial value of the motion analysis, each time t = t ₁ ,
Evaluation of equation (5) for m sets of observation azimuths θ _m (t) and Doppler shift observed frequency components ν _km (t) (k = 1,2,…, p) of t ₂ , ..., t _m. The target motion analysis that minimizes the function is performed, and the estimated value VX ₂ ^* (t) of the internal state quantity corresponding to the position vector Vx ₂ (t) for target 2 is calculated and output terminal 2
7 for output. As described above, according to the present embodiment, from the foreseeing information regarding the state quantity of the target 2, the observed quantity and the measurement accuracy of the observed quantity, the pseudo linear equation is constructed assuming that the observed noise component is minute, and the pseudo linear equation is calculated. By solving the linear equation, the analysis initial value VX ₀₀ ^* of the motion analysis including the foresight information regarding the state quantity of the target 2 is determined. Therefore, it becomes possible to make use of both the prior information and the observation data, and when solving the non-linear equation (14), the solution converges quickly and the conventional problems can be solved.

The present invention is not limited to the above embodiment, but can be variously modified. For example, in the above-described embodiment, the internal state quantity to be obtained is calculated by setting the position and velocity of the target 2
However, the present invention can be applied to the case where only the position and velocity of the target 2 are obtained. In this case,
The sound source natural frequency and the Doppler shift frequency may be removed from the equations (3), (4), (5) and (9). Also,
In the target motion analysis device of FIG. 1, the azimuth information calculation unit 22, the frequency information calculation unit 23, the initial value calculation unit 24, the maneuver detection unit 25, and the internal state amount estimation unit 26 are described as individual units. A configuration similar to that of the above-described embodiment may be adopted in which online processing is performed on a received signal from the receiver sensor array.

[0023]

As described in detail above, according to the first to fourth inventions, foreseeing information regarding the target state quantity,
Based on the observed amount and the measurement accuracy of the observed amount, a pseudo-linear equation is constructed assuming that the observation noise component is small, and by solving the pseudo-linear equation, the initial analysis of the motion analysis including the foresight information about the target state quantity. The value is set. Therefore, it becomes possible to make use of both the a priori information and the observation data, and if the nonlinear equation is solved by using the analysis initial value, the solution converges quickly and the target state quantity estimate can be obtained quickly.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a target motion analysis device used in a target motion analysis method according to an embodiment of the present invention.

FIG. 2 is a geometrical explanatory view showing a conventional desired motion analysis method.

FIG. 3 is a functional block diagram of a desired motion analysis device used in a conventional desired motion analysis method.

[Explanation of symbols]

 22 azimuth information calculation unit 23 frequency information calculation unit 24 initial position calculation unit 25 maneuver detection unit 26 internal state quantity estimation unit

Claims (4)

[Claims] 1. A target sensor array which receives sound emitted from a target by a receiver sensor array attached to a movable body, and the target is equal from the azimuth observation time series which is the azimuth measurement result for the target. In the target motion analysis method of determining the initial value of the motion analysis of the target on the assumption that the fast linear motion is performed, and estimating the state quantity related to the position and speed of the target, a rough value related to the state quantity of the target and When the foresight information consisting of the accuracy is obtained, the foresight information, the azimuth observation amount, the angle measurement accuracy of the azimuth observation amount, the velocity component of the navigation vehicle, and the time series of the position of the navigation vehicle. From the above, a pseudo-linear equation in which the observation noise component is considered to be minute is formed, and by solving the pseudo-linear equation, an analysis initial value including the contents of the foresight information is determined as an initial value of the motion analysis, and repeated calculation is performed. Nonlinear least squares Desired motion analysis method and estimates the state amount related to the position and velocity of the target using the or initial value solving online processing. 2. A sound sensor emitted from a target is received by a wave receiver sensor array attached to a movable body, and a time series of azimuth observation amount, which is a measurement result of azimuth with respect to the target, and the radiated sound. From the frequency observation amount time series which is the Doppler shift frequency measurement result of, it is assumed that the state amount relating to the position and velocity of the target and the state amount relating to the sound source natural frequency of the target are assumed to perform constant velocity linear motion. In the target motion analysis method for determining and estimating the initial value of the motion analysis of the target, when the foreseeing information including the approximate value and the accuracy of the state quantity of the target is obtained, the foreseeing information and the azimuth are obtained. Observation noise from the observation amount, the angle measurement accuracy of the azimuth observation amount, the time series of the velocity component of the moving body, the position of the moving body, and the time series of the frequency observation amount and the measurement accuracy of the frequency observation amount. The component is minute To form a pseudo-linear equation and solve the pseudo-linear equation to define an analysis initial value including the contents of the foresight information as the initial value of the motion analysis, and iteratively calculate the nonlinear least squares method or A target motion analysis method, wherein an initial value is used for online processing to estimate a state quantity related to the target position and velocity and a state quantity related to the target sound source natural frequency. 3. The estimated value and the state quantity of the target state quantity at the time of the maneuver detection when the maneuver such as a change of the target needle and a gear shift is detected are zero when the component related to the speed is zero. Of the estimated information matrix, the configured foresight information, the azimuth observation amount after the maneuver detection, the angle measurement accuracy of the azimuth observation amount, the velocity component of the vehicle and the position of the vehicle. From the time series, construct a pseudo-linear equation that the observation noise component is minute, and solve the pseudo-linear equation to set the analysis initial value including the contents of the foresight information as the initial value of the motion analysis, and repeat. 2. A state quantity relating to a state quantity relating to the position and velocity of the target is estimated by performing a calculation to solve the nonlinear least squares method or by using the initial value for online processing. The desired motion analysis method. 4. The estimated value of the target state quantity at the time of detecting the maneuver and the state quantity thereof when the maneuver such as a change of the target needle and a gear shift is detected is zero. The estimated information matrix of, the configured foresight information, the azimuth observation amount, the angle measurement accuracy of the azimuth observation amount, the speed component of the vehicle and the time series of the position of the vehicle, and An analysis initial value that includes the contents of the foresight information by constructing a pseudo-linear equation in which the observed noise component is minute from the frequency observation amount and the time series of the measurement accuracy of the frequency observation amount, and solving the pseudo-linear equation As the initial value of the motion analysis, and the nonlinear least squares method is solved by performing an iterative calculation, or the initial value is used for online processing to determine the state quantity related to the target position and velocity and the target sound source natural frequency. Desired motion analysis method according to claim 2, wherein the estimating the state amount.