CN114236524A - Doppler radar sequential smooth variable structure filtering method and device - Google Patents

Abstract

The application provides a Doppler radar sequential smooth variable structure filtering method and device, belongs to the technical field of radar data processing, and can simultaneously solve the problem of a nonlinear underdetermined observation model of a Doppler radar and the problem of robust tracking under the uncertain condition of a target motion model. The method adopts the strategies of target position measurement and radial velocity measurement decoupling processing and sequential estimation of the Doppler radar to solve the problem of a nonlinear underdetermined observation model: in a first measurement conversion module and a first state estimator, a first-stage target state estimation is obtained by utilizing nonlinear target position measurement and a generalized smooth variable structure filtering method; and in the second-stage measurement conversion module and the second state estimator, pseudo measurement is constructed by using radial velocity measurement, and the first-stage target state estimation is updated to obtain the final posterior state estimation of the current frame target. The generalized smooth variable structure filtering method ensures the robustness of the model under the uncertain condition. Therefore, the target tracking precision of the Doppler radar can be improved, and meanwhile, the robust estimation performance is kept under the condition that a target motion model is uncertain.

Description

Doppler radar sequential smooth variable structure filtering method and device

Technical Field

The embodiment of the application relates to the technical field of radar data processing, in particular to a Doppler radar sequential smooth variable structure filtering method and device.

Background

A radar target tracking device is a device for continuously and effectively estimating the number, position, speed, acceleration and other motion states of targets by utilizing the information of target distance, azimuth angle, Doppler frequency offset and the like contained in radar echoes, and is widely applied to the fields of automatic driving, air traffic control, meteorological monitoring, security protection, national defense and military and the like, wherein the Doppler radar can observe the radial speed information of target motion, so that the effective information dimension is increased in the radar target state estimation link, and the target tracking performance is greatly improved.

A target tracking algorithm or device of the doppler radar needs to establish a radar observation model and a target state transition model, but because the observation model of the doppler radar is a highly nonlinear observation model, and the highly nonlinear model can cause filters frequently used in common radar tracking methods, such as a Kalman Filter (KF) and the like, to generate a large nonlinear estimation error, and even cause filtering divergence; meanwhile, a target state transition model of the doppler radar has uncertainty, while the traditional bayesian filtering method including a Kalman Filter (KF) and the like all depend on accurate description of a target motion model, if the deviation between the target state transition model and the motion model of a real target is large, for example, the target generates strong maneuvering motion, a tracking error is increased sharply, and even filtering divergence loses the target.

In order to solve errors caused by a Doppler radar nonlinear observation model, a series of nonlinear filtering methods including Extended Kalman Filtering (EKF), insensitive Kalman filtering (UKF) and Particle Filtering (PF) and Sequential Kalman Filtering (SKF) methods with high estimation precision and low calculation complexity are applied when the Doppler radar is applied, but the traditional Bayesian methods including the sequential Kalman filtering and a large number of improved forms thereof can reduce the nonlinear errors but cannot solve the problem of uncertainty of a target state transition model.

In order to solve the problem of uncertainty of a target state transition model of the Doppler radar, robust Kalman filtering, H-infinite filtering and Smooth Variable Structure Filtering (SVSF) which is concerned in recent years are often adopted, and the SVSF is a robust filtering method under the condition of model uncertainty, so that the boundedness of state estimation errors can be ensured, and meanwhile, the calculation complexity is low; however, the standard smooth-varying structure filtering method requires a linear observation model and a full-rank observation matrix, thereby realizing bijective mapping from the target state space to the observation space. Since the observation model of the doppler radar is nonlinear, in order to solve the problem of the nonlinear observation model, the jacobian matrix is also used to substitute the nonlinear function into the standard smooth variable structure filter structure, but this introduces the difficult problem of inversion of a new underdetermined matrix, which results in a large estimation error of the target velocity, and thus, the smooth variable structure filtering method cannot be applied to the doppler radar device.

Therefore, the accuracy problem caused by the nonlinear underdetermined observation model of the doppler radar and the robust tracking problem under the uncertain condition of the motion model of the target still need to be solved urgently.

Disclosure of Invention

The embodiment of the application provides a Doppler radar sequential smooth variable structure filtering method, device, equipment and storage medium, and aims to improve the tracking precision of a Doppler radar observation target state and maintain robust estimation performance under the condition that a target motion model is uncertain.

In a first aspect, an embodiment of the present application provides a doppler radar sequential smooth varying structure filtering method, including the following steps:

the first measurement conversion module is used for converting the nonlinear target position measurement of the current frame Doppler radar into linear position measurement under a Cartesian coordinate system;

the first state estimator obtains a first-stage target state estimation by utilizing the generalized smooth variable structure filter and the linear position measurement in a Cartesian coordinate system;

the second measurement conversion module is used for constructing target distance-radial velocity product pseudo measurement by utilizing the target radial velocity measurement of the current frame Doppler radar;

and the second state estimator updates the first-stage target state estimation based on the target distance-radial velocity product pseudo measurement to obtain the final posterior state estimation of the current frame target.

The first measurement conversion module converts the nonlinear target position measurement vector under the three-dimensional spherical coordinate or the two-dimensional polar coordinate of the Doppler radar into the linear position measurement vector under a Cartesian coordinate system by adopting an unbiased measurement conversion method, and calculates a first conversion deviation and a first conversion covariance.

The first state estimator obtains a first-stage target state estimation by using the generalized smooth-varying structure filter and the linear position measurement in a Cartesian coordinate system, and comprises the following steps of:

calculating a target state prior estimation, a radar measurement prior estimation and a prior state estimation covariance matrix of a current frame;

calculating a priori position measurement error according to the linear position measurement of the current frame, the first conversion deviation, the first conversion covariance and the radar measurement priori estimate;

calculating a first innovation gain term according to the prior position measurement error;

and calculating a target state posterior estimation and a posterior state estimation covariance matrix of the current frame according to the target state prior estimation, the prior state estimation covariance matrix, the prior position measurement error and the first new gain term, and taking the target state posterior estimation and the posterior state estimation covariance matrix of the current frame as first-stage target state estimation.

The second measurement conversion module constructs the distance-radial velocity product pseudo measurement by using the target distance measurement and the target radial velocity measurement of the Doppler radar, and calculates a second conversion deviation, a second conversion variance and a second conversion covariance of the distance-radial velocity product pseudo measurement by adopting a deflection-free conversion method.

The second state estimator updates the first-stage target state estimation based on the target distance-radial velocity product pseudo measurement to obtain a final posterior state estimation of the current frame target, and comprises the following steps:

pre-whitening the distance-radial velocity product pseudo measurement to obtain a pre-whitened pseudo measurement, a deviation of the pre-whitened pseudo measurement and a variance of the pre-whitened pseudo measurement;

and taking the first-stage target state estimation, the pre-whitening pseudo measurement, the deviation of the pre-whitening pseudo measurement and the variance of the pre-whitening pseudo measurement as input, constructing a local approximate linear minimum mean square error estimator, and updating the first-stage target state estimation to obtain the final posterior state estimation of the current frame target.

In a second aspect, an embodiment of the present application provides a doppler radar sequential smooth-varying structure filtering apparatus, where the filtering apparatus includes a first measurement transformation module, a first state estimator, a second measurement transformation module, and a second state estimator, where:

the first measurement conversion module is used for converting the nonlinear target position measurement of the current frame Doppler radar into linear position measurement under a Cartesian coordinate system;

the first state estimator is used for obtaining a first-stage target state estimation by utilizing the generalized smooth variable structure filter and the linear position measurement in a Cartesian coordinate system;

the second measurement conversion module is used for constructing target distance-radial velocity product pseudo measurement by utilizing target radial velocity measurement of the current frame Doppler radar;

and the second state estimator is used for updating the first-stage target state estimation based on the target distance-radial velocity product pseudo measurement to obtain the final posterior state estimation of the current frame target.

The first measurement conversion module is used for converting the nonlinear target position measurement vector under a three-dimensional spherical coordinate or a two-dimensional polar coordinate of the Doppler radar into the linear position measurement vector under a Cartesian coordinate system by adopting an unbiased measurement conversion method, and calculating a first conversion deviation and a first conversion covariance.

The first state estimator comprises:

the first calculation unit is used for calculating a target state prior estimation, a radar measurement prior estimation and a prior state estimation covariance matrix of a current frame;

a second calculation unit, configured to calculate a priori position measurement error according to the linear position measurement of the current frame, the first conversion bias, the first conversion covariance, and the radar measurement priori estimate;

a third calculating unit, configured to calculate a first innovation gain term according to the priori position measurement error;

and the fourth calculation unit is used for calculating a target state posterior estimation and a posterior state estimation covariance matrix of the current frame according to the target state prior estimation, the prior state estimation covariance matrix, the prior position measurement error and the first new gain term, and taking the target state posterior estimation and the posterior state estimation covariance matrix of the current frame as the first-stage target state estimation.

The second measurement conversion module constructs the distance-radial velocity product pseudo measurement by using the target distance measurement and the target radial velocity measurement of the Doppler radar, and calculates a second conversion deviation, a second conversion variance and a second conversion covariance of the distance-radial velocity product pseudo measurement by adopting a deflection-free conversion method.

The second state estimator comprises:

a pre-whitening processing unit, configured to perform pre-whitening processing on the distance-radial velocity product pseudo measurement to obtain a pre-whitening pseudo measurement, a deviation of the pre-whitening pseudo measurement, and a variance of the pre-whitening pseudo measurement;

and the posterior state estimation unit is used for taking the first-stage target state estimation, the pre-whitening pseudo measurement, the deviation of the pre-whitening pseudo measurement and the variance of the pre-whitening pseudo measurement as input, constructing a local approximate linear minimum mean square error estimator, and updating the first-stage target state estimation to obtain the final posterior state estimation of the current frame target.

Has the advantages that:

the sequential smooth variable structure filtering method provided by the invention has the advantages of two aspects:

firstly, the method has the advantage of robustness under the uncertain condition of the model, and based on the sliding mode variable structure control theory, the method can ensure the bounded property of the tracking error under the unknown condition of the target motion model, and solves the problems that the tracking filtering error is increased sharply and even the filtering is diverged when the traditional nonlinear Bayes filter (such as a Kalman filter, a sequential Kalman filter and the like) faces the modeling error of the target state transition model.

Specifically, firstly, nonlinear target position measurement of the Doppler radar is converted into linear position measurement in a Cartesian coordinate system, a linear generalized smooth variable structure filter can be further constructed, robust estimation of a target state is achieved, and robust state estimation performance under the condition of model uncertainty is guaranteed by further updating and correcting output first-stage target state estimation.

Secondly, the method adopts a state space sequential estimation method to divide the observation vector of the Doppler radar into two parts, including target position measurement and target radial velocity measurement, and the two parts are used for respectively and sequentially updating state estimation, so that the underdetermined problems of a Doppler radar nonlinear model and a non-full rank observation matrix in the conventional smooth variable structure filtering method are solved, and the target state estimation precision of the smooth variable structure filter can be effectively improved by fully utilizing the radial velocity measurement information of the Doppler radar.

Specifically, the sequential smooth-varying structure filter solves the problem that a standard smooth-varying structure filter cannot utilize nonlinear Doppler measurement, and also solves the underdetermined problem of non-full-rank observation matrix inversion faced when the improved smooth-varying structure filter of the existing literature adopts Jacobian matrix local linearization operation. Therefore, compared with the existing smooth variable structure filtering method, the method of the invention more effectively utilizes Doppler velocity information, thereby obviously improving the target state tracking precision.

In conclusion, the method can obviously improve the accuracy of the Doppler radar target state estimation, simultaneously ensures the filtering robustness under the uncertain condition of the model, and has good application value.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings used in the description of the embodiments of the present application will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the application, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.

FIG. 1 is a flow chart illustrating the steps of a filtering method according to an embodiment of the present application;

FIG. 2 is a diagram of a target maneuver trajectory in a simulation experiment according to an embodiment of the present application;

FIG. 3 is a diagram illustrating coordinate error versus tracking time for simulation results according to an embodiment of the present application;

FIG. 4 is a graphical illustration of velocity error versus tracking time for simulation results presented in one embodiment of the present application;

fig. 5 is a functional block diagram of a filtering apparatus according to an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. It is to be understood that the embodiments described are only a few embodiments of the present application and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

Referring to fig. 1, a flowchart of steps of a doppler radar sequential smooth-varying structure filtering method according to an embodiment of the present invention is shown, where the method is applied to a doppler radar sequential smooth-varying structure filtering apparatus, where the filtering apparatus includes a first measurement conversion module, a first state estimator, a second measurement conversion module, and a second state estimator, and the method specifically includes the following steps:

s101: the first measurement conversion module converts the nonlinear target position measurement of the current frame Doppler radar into linear position measurement under a Cartesian coordinate system.

Specifically, let the scanning period of the doppler radar be T, and the vector of the measured data of the current frame k of the doppler radar to the target state be z_m(k) The measured data includes the distance r_m(k) Azimuth angle theta_m(k) Angle of pitch

And radial velocity

According to the structure of the measurement data of the Doppler radar, the vector of the measurement data of the current frame k to the target state is taken as z_m(k) The method is divided into a target position measurement and a target radial velocity measurement, and the expression form is as follows:

in the formula (I), the compound is shown in the specification,

is a target position measurement including distance, azimuth and pitch angles,

is the target radial velocity measurement, k represents the number of frames of the doppler radar, k is 1, 2, 3 … ….

According to the type of the Doppler radar, the error standard deviation of each measured variable can be checked, wherein sigma_rThe standard deviation of the observed error representing the distance,

view of representing radial velocityThe standard deviation of the error is measured and measured,

standard deviation of observation error, σ, representing pitch angle_θAnd expressing the standard deviation of the observation error of the azimuth angle, wherein rho is the noise-related coefficient of the distance measurement and the radial velocity measurement.

The target position measurement obtained by the doppler radar is nonlinear, and the first measurement conversion module needs to convert the nonlinear target position measurement into a linear position measurement in a cartesian coordinate system.

Specifically, the non-linear target position under the doppler radar spherical coordinate system or polar coordinate system is measured by using an unbiased measurement conversion method (UCM)

Linear position measurement converted into Cartesian coordinate system

First conversion deviation mu^p＝[μ^x,μ^y,μ^z]^TAnd a first covariance R^p(k)；

Wherein r is_mRepresentative of distance, θ_mWhich represents the azimuth angle, is,

representing the pitch angle, and x, y and z are three coordinate axes in a cartesian coordinate system.

It should be noted that the present embodiment takes a three-dimensional doppler radar as an example, and in other embodiments, for example, the effective dimension may be directly truncated for a two-dimensional scene.

S102: and the first state estimator obtains the first-stage target state estimation by utilizing the generalized smooth variable structure filter and the linear position measurement in a Cartesian coordinate system.

The smooth variable structure filter is a robust filtering method under the condition of model uncertainty, can ensure the boundedness of state estimation errors, and has low computational complexity, but the standard smooth variable structure filtering method requires a linear observation model and a full-rank observation matrix, so that the bijection mapping from a target state space to an observation space is realized, which cannot be used for the nonlinear measurement of the Doppler radar. The method converts nonlinear target position measurement into linear position measurement, so that data processing can be performed by using a generalized smooth variable structure filter, and the robust state estimation performance under the condition of model uncertainty is ensured.

The first state estimator obtains a first stage target state estimate, and specifically comprises the following steps:

s1021: and calculating a target state prior estimation, a radar measurement prior estimation and a prior state estimation covariance matrix of the current frame k.

In particular, the target state is estimated a priori

Radar measurement prior estimation

Sum prior state estimate covariance matrix P^p(k | k-1), determined according to the following formula:

in the formula (I), the compound is shown in the specification,

is a variable of the state estimation that is,

is in a preset shapeThe state of the state transition matrix is,

is a preset control input matrix, u is a known control input variable,

is an observation matrix, m and n represent the dimensions of the state variables and the observation variables, respectively, Q and Γ are process noise covariance and its coefficient matrices,

and

is different from the target state transition models F and G which are required to be accurately known by the traditional Bayesian filter,

and

norm-bounded model errors are allowed to exist compared to true values.

In the method, a decorrelation unbiased measurement conversion method is applied to convert nonlinear target position measurement into linear position measurement, so that a matrix is observed

Is linear time invariant and is superior to conventional smooth variable structure filters that directly use nonlinear position measurements.

S1022: calculating a priori position measurement error according to the linear position measurement of the current frame, the first conversion deviation, the first conversion covariance and the radar measurement priori estimate.

In particular, the a priori position measurement error e^p(k | k-1), determined according to the following formula:

in the formula (I), the compound is shown in the specification,

for linear position measurement, mu^p(^K) Is a first transition deviation;

is a radar measurement prior estimate.

S1023: and calculating a first innovation gain term according to the prior position measurement error.

Specifically, the first innovation gain term may be determined by:

first, the blending error term is calculated:

in the formula, e_z(k-1| k-1) is the posterior measurement error of k-1 frame, | · calculation of calculation_ABSMeans taking the absolute value element by element for the vector,

and

is an attenuation factor with a value in the interval (0, 1);

is to the preset state transition matrix

Introducing a transformation state transition matrix after the transformation matrix T, and transforming the state transition matrix

A diagonal matrix representing a main diagonal element as an arbitrary vector a;

then, a first innovation gain term k (k) is calculated:

wherein sat (.) represents a saturation function;

and

respectively are preset smooth layer parameter vectors; h₁＝I_3×3Is the observation matrix

The full rank blocking submatrix of (1), the full rank blocking submatrix H₁＝I_3×3Is linear time invariant.

S1024: and calculating a target state posterior estimation and a posterior state estimation covariance matrix of the current frame according to the target state prior estimation, the prior state estimation covariance matrix, the prior position measurement error and the first new gain term, and taking the target state posterior estimation and the posterior state estimation covariance matrix of the current frame as first-stage target state estimation.

Posterior estimation of the target state at the current frame kMeter

State covariance matrix P with a posteriori state^p(k | k), determined according to the following formula:

wherein the content of the first and second substances,

is a priori estimate of the state of said object, P^p(k | k-1) is the prior state estimate covariance matrix, e^p(k | k-1) is the a priori position measurement error, and k (k) is the first innovation gain term;

is a unit matrix, R^p(k) Is the first covariance of the first set of co-variances,

is the observation matrix.

Will calculate the posterior estimate

State covariance matrix P with a posteriori state^p(k | k) as a result of the first stage target state estimation.

S103: and the second measurement conversion module is used for constructing target distance-radial velocity product pseudo measurement by utilizing the target radial velocity measurement of the current frame Doppler radar.

Specifically, the second measurement conversion module uses the target distance measurement and the target radial velocity measurement of the doppler radar to construct the distance-radial velocity product pseudo measurement, and calculates the distance-radial velocity by using a non-deflection conversion methodSecond conversion deviation mu of false product measurement^ηThe second conversion variance R^ηAnd a second conversion covariance R^pη。

The distance-radial velocity product pseudo metric is defined according to the following formula:

wherein r is_m(k) In order to be the distance between the two,

the target radial velocity measurement is made.

Then, respectively calculating a second transformation deviation mu of the pseudo measurement of the distance-radial velocity product^ηSecond conversion variance R^ηAnd a second conversion covariance R^pη：

Where ρ represents a correlation coefficient of a distance observation error and a radial velocity observation error,

σ denotes the standard deviation of observation error for each observation dimension, σ_rThe standard deviation of the observed error representing the distance,

the standard deviation of the observed error representing the radial velocity,

standard deviation of observation error, σ, representing pitch angle_θIndicating the standard deviation of observation error for azimuth.

S104: and the second state estimator updates the first-stage target state estimation based on the target distance-radial velocity product pseudo measurement to obtain the final posterior state estimation of the current frame target.

The method specifically comprises the following substeps:

s1041: pre-whitening the distance-radial velocity product artifact measurement to obtain a pre-whitened artifact measurement, a deviation of the pre-whitened artifact measurement and a variance of the pre-whitened artifact measurement;

specifically, the process of performing pre-whitening processing on the distance-radial velocity product pseudo measurement includes:

first, the pre-whitening coefficient matrix l (k) is calculated:

L(k)＝-R^ηp(k)(R^P(k))^-1＝[L₁(k) L₂(k) L₃(k)]

in the formula, R^ηp(k) For the second conversion covariance R^pη(k) Transpose of R^p(k) Is the first covariance.

Then, according to the pre-whitening coefficient matrix, the pre-whitening pseudo measurement epsilon is calculated_c(k) The conversion deviation mu of the pre-whitening artifact measurement^ε(k) And the variance R of the pre-whitening artifact measure^ε(k)：

Wherein, the vector

Scalar η for said linear position measurement_c(k) Pseudo-measurement of the target distance-radial velocity product;

calculating a conversion deviation mu of a pre-whitening artifact measurement^ε(k)：

μ^ε(k)＝L(k)μ^p(k)+μ^η(k)

Wherein the vector mu^p(k) For said first conversion offset, a scalar μ^η(k) Is the second conversion offset.

Calculating the variance R of the pre-whitening artifact measure^ε(k)：

R^ε(k)＝R^η(k)-(R^pη(k))^T(R^P(k))^-1R^pη(k)

Wherein R is^p(k) Is the first covariance, R^η(k) Is the second conversion variance, R^pη(k) Is the second conversion covariance.

S1042: and taking the first-stage target state estimation, the pre-whitening pseudo measurement, the deviation of the pre-whitening pseudo measurement and the variance of the pre-whitening pseudo measurement as input, constructing a local approximate linear minimum mean square error estimator, and updating the first-stage target state estimation to obtain the final posterior state estimation of the current frame target.

The method specifically comprises the following steps:

K^ε(k)＝P^p(k|k)H^ε(k)^T[H^ε(k)P^p(k|k)H^ε(k)^T+R^ε(k)]^-1

P(k|k)＝[I-K^ε(k)H^ε(k)]P^p(k|k)[I-K^ε(k)H^ε(k)]^T+K^ε(k)R^ε(k)K^ε(k)^T

wherein the non-linear function

The definition is as follows:

matrix H^ε(k) As a non-linear function

The Jacobian matrix of (1) is defined as:

estimating the corrected posterior state

And the modified state covariance matrix P (k | k) is used as the final posterior state estimation of the target of the current frame k.

According to the method, firstly, Doppler radar nonlinear target position measurement is converted into linear position measurement in a Cartesian coordinate system, a linear generalized smooth variable structure filter can be constructed, robust estimation of a target state is achieved, and output first-stage target state estimation is further updated and corrected, so that robust state estimation performance under the condition of model uncertainty is guaranteed.

Secondly, the method adopts a state space sequential estimation method to divide the observation vector of the Doppler radar into two parts including target position measurement and target radial velocity measurement, and the two parts are used for respectively and sequentially updating state estimation, so that the underdetermined problems of a Doppler radar nonlinear model and a non-full rank observation matrix in the conventional smooth variable structure filtering method are solved, and further the target state estimation precision of a smooth variable structure filter can be effectively improved by fully utilizing the radial velocity measurement information of the Doppler radar.

The embodiment also provides a simulation scenario for comparing simulation experiments of the sequential smooth-varying structure filtering method (SSVSF), the extended kalman filtering method (EKF), the smooth-varying structure filtering method (PO-SVSF) using only position measurement, and the smooth-varying structure filtering method (SVSF) using jacobian matrix local linearization proposed by the present invention.

The maneuvering target tracking is a typical scene of the uncertain problem of the model, so that a simulation scene of tracking and filtering of the 2-D Doppler millimeter wave radar on the single maneuvering target in an automatic driving scene is made, and the radar is established at a coordinate origin.

Referring to fig. 2, fig. 2 is a diagram of a target maneuvering trajectory in a simulation environment, an abscissa x and an ordinate y in fig. 2 form a two-dimensional plane, a real point in the diagram is a doppler radar in the simulation environment, and a dotted line in the diagram is a target maneuvering trajectory.

Simulated radar target simulation data is generated using the simulation parameters of table 1 below, with a target state simulation scalar defined as position, velocity in the X-Y coordinate axis, i.e., X ═ X yv_x v_y]^T(ii) a Radar measurement variable definition

Estimating the target state by adopting a uniform motion model (CV), namely:

the measurement matrix of the filter of the first stage of the sequential smooth variable structure filtering method is as follows:

TABLE 1 Radar target tracking scene parameters

Simulation experiments are performed according to the sequential smooth varying structure filtering method (SSVSF), the extended kalman filtering method (EKF), the smooth varying structure filtering method using only position measurement (PO-SVSF), and the smooth varying structure filtering method using jacobian matrix local linearization (SVSF) provided in this embodiment, and an average result of 2000 monte carlo experiments is used as comparison data; the Root Mean Square Error (RMSE) for 2000 monte carlo experiments is given in table 2 below.

TABLE 2 State estimation root mean square error of uncertain targets in motion model

Obviously, as can be seen from table 2, the root mean square error of the target state estimation is the smallest in the sequential smooth and variable structure filtering method (SSVSF) proposed in this embodiment under the uncertain condition of the motion model.

Fig. 3 shows the variation of the coordinate error of each method target state estimation with the tracking time, and fig. 4 shows the variation of the speed error of each method target state estimation with the tracking time.

As can be seen from fig. 3 and 4, the Sequential Smooth Variable Structure Filtering (SSVSF) method proposed by the present invention achieves the best tracking accuracy, and compared with the Extended Kalman Filtering (EKF), the SSVSF method is more robust in tracking performance during target maneuvering, for example, the peak coordinate error is reduced by more than half at 10-15s and 23-28 s; compared with PO-SVSF measured only by using the position, the SSVSF can effectively use Doppler velocity information, and greatly improves the target state estimation precision; however, the locally linearized SVSF faces an underdetermined problem caused by a non-full rank observation matrix, and the target state estimation performance, especially the speed estimation performance, is the worst.

Therefore, through simulation experiments, the method solves the problem of a nonlinear underdetermined observation model of the Doppler radar, and meanwhile, under the condition that a target motion model is uncertain, Doppler velocity measurement information can be fully utilized, and more robust and accurate radar target state estimation is realized.

The embodiment of the present application provides a doppler radar sequential smooth transition structure filtering apparatus, the filtering apparatus includes a first measurement transformation module 100, a first state estimator 200, a second measurement transformation module 300, and a second state estimator 400, wherein:

the first measurement conversion module 100 is configured to convert the nonlinear target position measurement of the current frame doppler radar into a linear position measurement in a cartesian coordinate system;

the first state estimator 200 is configured to obtain a first-stage target state estimate by using a generalized smooth-varying structure filter and the linear position measurement in a cartesian coordinate system;

the second measurement conversion module 300 is configured to construct a target distance-radial velocity product pseudo measurement by using a target radial velocity measurement of the current frame doppler radar;

the second state estimator 400 is configured to update the first-stage target state estimation based on the target distance-radial velocity product pseudo measurement to obtain a final posterior state estimation of the current frame target.

The first measurement conversion module is used for converting the nonlinear target position measurement vector under a three-dimensional spherical coordinate or a two-dimensional polar coordinate of the Doppler radar into the linear position measurement vector under a Cartesian coordinate system by adopting an unbiased measurement conversion method, and calculating a first conversion deviation and a first conversion covariance.

The first state estimator comprises:

the first calculation unit is used for calculating a target state prior estimation, a radar measurement prior estimation and a prior state estimation covariance matrix of a current frame;

a second calculation unit, configured to calculate a priori position measurement error according to the linear position measurement of the current frame, the first conversion bias, the first conversion covariance, and the radar measurement priori estimate;

a third calculating unit, configured to calculate a first innovation gain term according to the priori position measurement error;

and the fourth calculation unit is used for calculating a target state posterior estimation and a posterior state estimation covariance matrix of the current frame according to the target state prior estimation, the prior state estimation covariance matrix, the prior position measurement error and the first new gain term, and taking the target state posterior estimation and the posterior state estimation covariance matrix of the current frame as the first-stage target state estimation.

The second measurement conversion module is used for constructing the distance-radial velocity product pseudo measurement by utilizing the target distance measurement and the target radial velocity measurement of the Doppler radar, and calculating a second conversion deviation, a second conversion variance and a second conversion covariance of the distance-radial velocity product pseudo measurement by adopting a deflection-free conversion method.

The second state estimator comprises:

a pre-whitening processing unit, configured to perform pre-whitening processing on the distance-radial velocity product pseudo measurement to obtain a pre-whitening pseudo measurement, a deviation of the pre-whitening pseudo measurement, and a variance of the pre-whitening pseudo measurement;

and the posterior state estimation unit is used for taking the first-stage target state estimation, the pre-whitening pseudo measurement, the deviation of the pre-whitening pseudo measurement and the variance of the pre-whitening pseudo measurement as input, constructing a local approximate linear minimum mean square error estimator, and updating the first-stage target state estimation to obtain the final posterior state estimation of the current frame target.

The embodiments in the present specification are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.

As will be appreciated by one of skill in the art, embodiments of the present application may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (apparatus), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While preferred embodiments of the present application have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including the preferred embodiment and all such alterations and modifications as fall within the true scope of the embodiments of the application.

Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or terminal that comprises the element.

The principle and the implementation of the present application are explained herein by applying specific examples, and the above description of the embodiments is only used to help understand the method and the core idea of the present application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.

Claims (10)

1. A Doppler radar sequential smooth variable structure filtering method is characterized by comprising the following steps:

the first measurement conversion module is used for converting the nonlinear target position measurement of the current frame Doppler radar into linear position measurement under a Cartesian coordinate system;

the first state estimator obtains a first-stage target state estimation by utilizing the generalized smooth variable structure filter and the linear position measurement in a Cartesian coordinate system;

the second measurement conversion module is used for constructing target distance-radial velocity product pseudo measurement by utilizing the target radial velocity measurement of the current frame Doppler radar;

and the second state estimator updates the first-stage target state estimation based on the target distance-radial velocity product pseudo measurement to obtain the final posterior state estimation of the current frame target.

2. The filtering method as claimed in claim 1, wherein the first measurement transforming module transforms the nonlinear target position measurement vector in three-dimensional spherical coordinates or two-dimensional polar coordinates of the doppler radar into the linear position measurement vector in cartesian coordinates by using an unbiased measurement transformation method, and calculates a first transformation bias and a first transformation covariance.

3. The filtering method according to claim 2, wherein said first state estimator obtains a first stage target state estimate using a generalized smoothed structure filter and said linear position measurement in cartesian coordinates, comprising the steps of:

calculating a target state prior estimation, a radar measurement prior estimation and a prior state estimation covariance matrix of a current frame;

calculating a priori position measurement error according to the linear position measurement of the current frame, the first conversion deviation, the first conversion covariance and the radar measurement priori estimate;

calculating a first innovation gain term according to the prior position measurement error;

and calculating a target state posterior estimation and a posterior state estimation covariance matrix of the current frame according to the target state prior estimation, the prior state estimation covariance matrix, the prior position measurement error and the first new gain term, and taking the target state posterior estimation and the posterior state estimation covariance matrix of the current frame as first-stage target state estimation.

4. The filtering method as claimed in any one of claims 1 to 3, wherein said second measurement transformation module uses a target range measurement and a target radial velocity measurement of Doppler radar to construct said range-radial velocity product pseudo-measurement, and uses a non-deflection transformation method to calculate a second transformation bias, a second transformation variance and a second transformation covariance of said range-radial velocity product pseudo-measurement.

5. The filtering method according to claim 4, wherein the second state estimator updates the first stage target state estimation based on the target distance-radial velocity product pseudo measurement to obtain a final a posteriori state estimation of the target of the current frame, comprising the steps of:

pre-whitening the distance-radial velocity product pseudo measurement to obtain a pre-whitened pseudo measurement, a deviation of the pre-whitened pseudo measurement and a variance of the pre-whitened pseudo measurement;

and taking the first-stage target state estimation, the pre-whitening pseudo measurement, the deviation of the pre-whitening pseudo measurement and the variance of the pre-whitening pseudo measurement as input, constructing a local approximate linear minimum mean square error estimator, and updating the first-stage target state estimation to obtain the final posterior state estimation of the current frame target.

6. A Doppler radar sequential smooth variable structure filtering device is characterized in that the filtering device comprises a first measurement conversion module, a first state estimator, a second measurement conversion module and a second state estimator, wherein:

the first measurement conversion module is used for converting the nonlinear target position measurement of the current frame Doppler radar into linear position measurement under a Cartesian coordinate system;

the first state estimator is used for obtaining a first-stage target state estimation by utilizing the generalized smooth variable structure filter and the linear position measurement in a Cartesian coordinate system;

the second measurement conversion module is used for constructing target distance-radial velocity product pseudo measurement by utilizing target radial velocity measurement of the current frame Doppler radar;

and the second state estimator is used for updating the first-stage target state estimation based on the target distance-radial velocity product pseudo measurement to obtain the final posterior state estimation of the current frame target.

7. The filtering apparatus as claimed in claim 6, wherein the first measurement transforming module transforms the nonlinear target position measurement vector in three-dimensional spherical coordinates or two-dimensional polar coordinates of the doppler radar into the linear position measurement vector in cartesian coordinates by using an unbiased measurement transformation method, and calculates a first transformation bias and a first transformation covariance.

8. The filtering apparatus as claimed in claim 7, wherein said first state estimator comprises:

the first calculation unit is used for calculating a target state prior estimation, a radar measurement prior estimation and a prior state estimation covariance matrix of a current frame;

a second calculation unit, configured to calculate a priori position measurement error according to the linear position measurement of the current frame, the first conversion bias, the first conversion covariance, and the radar measurement priori estimate;

a third calculating unit, configured to calculate a first innovation gain term according to the priori position measurement error;

and the fourth calculation unit is used for calculating a target state posterior estimation and a posterior state estimation covariance matrix of the current frame according to the target state prior estimation, the prior state estimation covariance matrix, the prior position measurement error and the first new gain term, and taking the target state posterior estimation and the posterior state estimation covariance matrix of the current frame as the first-stage target state estimation.

9. The filtering apparatus as claimed in claims 6-8, wherein the second measurement transforming module is configured to construct the pseudo-range-radial-velocity measurement by using a target range measurement and a target radial velocity measurement of a doppler radar, and calculate a second transformation bias, a second transformation variance and a second transformation covariance of the pseudo-range-radial-velocity measurement by using a non-deflection transformation method.

10. The filtering apparatus as claimed in claim 9, wherein the second state estimator comprises:

a pre-whitening processing unit, configured to perform pre-whitening processing on the distance-radial velocity product pseudo measurement to obtain a pre-whitening pseudo measurement, a deviation of the pre-whitening pseudo measurement, and a variance of the pre-whitening pseudo measurement;

and the posterior state estimation unit is used for taking the first-stage target state estimation, the pre-whitening pseudo measurement, the deviation of the pre-whitening pseudo measurement and the variance of the pre-whitening pseudo measurement as input, constructing a local approximate linear minimum mean square error estimator, and updating the first-stage target state estimation to obtain the final posterior state estimation of the current frame target.

Images