阅读说明：本技术 一种基于强跟踪滤波的雷达目标跟踪方法 (Radar target tracking method based on strong tracking filtering ) 是由 葛泉波 王梦梦 孙长银 于 2021-01-21 设计创作，主要内容包括：本发明提供一种基于强跟踪滤波的雷达目标跟踪方法,适用于目标运动状态的雷达目标跟踪过程,所述方法包括：根据目标的运动特征,通过引入次优渐消因子,构建用于对目标运动状态进行强跟踪滤波的强跟踪滤波器模型,以及基于所述强跟踪滤波器模型,重复执行目标的强跟踪滤波,获得目标的运动状态跟踪信息,从而可以实现目标跟踪过程中的模型参数进行实时估计,有效地提高了目标跟踪的效果和稳定性。(The invention provides a radar target tracking method based on strong tracking filtering, which is suitable for a radar target tracking process of a target motion state, and comprises the following steps: according to the motion characteristics of the target, by introducing a suboptimal fading factor, a strong tracking filter model for carrying out strong tracking filtering on the motion state of the target is constructed, and strong tracking filtering of the target is repeatedly executed based on the strong tracking filter model to obtain the motion state tracking information of the target, so that the model parameters in the target tracking process can be estimated in real time, and the target tracking effect and stability are effectively improved.)

1. A radar target tracking method based on strong tracking filtering is characterized in that the method is suitable for a radar target tracking process of a target motion state, and comprises the following steps:

according to the motion characteristics of the target, a strong tracking filter model for carrying out strong tracking filtering on the motion state of the target is constructed by introducing a suboptimal fading factor;

and repeatedly executing the strong tracking filtering of the target based on the strong tracking filter model to obtain the motion state tracking information of the target.

2. The radar target tracking method based on the strong tracking filtering as claimed in claim 1, wherein the constructing a strong tracking filter model for performing the strong tracking filtering on the motion state of the target by introducing a sub-optimal fading factor according to the motion characteristics of the target comprises:

determining a motion model of the target according to the motion characteristics of the target, and further determining a process noise covariance and an observation noise covariance of the target;

constructing a Kalman filter first model according to the process noise covariance and the observation noise covariance;

and introducing a suboptimal fading factor into the process noise covariance matrix, and transforming the first Kalman filter model into a second Kalman filter model.

3. The strong tracking filtering based radar target tracking method according to claim 2, wherein the introducing a sub-optimal cancellation factor into the process noise covariance matrix, transforming the Kalman filter first model into a Kalman filter second model comprises:

constructing a suboptimal fading factor, and introducing the suboptimal fading factor into the process noise covariance;

calculating the suboptimal fading factor by adopting an approximate suboptimal method;

and transforming the first Kalman filter model into a second Kalman filter model based on the solved suboptimum fading factor.

4. The strong tracking filtering based radar target tracking method according to claim 3, wherein said introducing a sub-optimal cancellation factor into said process noise covariance matrix, transforming said Kalman filter first model into a Kalman filter second model, further comprising:

and introducing a weakening factor beta for weakening the regulation effect of the suboptimal fading factor in the process of constructing the suboptimal fading factor.

5. The radar target tracking method based on the strong tracking filtering as claimed in claim 1, wherein the constructing a strong tracking filter model for performing the strong tracking filtering on the motion state of the target by introducing a sub-optimal fading factor according to the motion characteristics of the target comprises:

determining a motion model of the target according to the motion characteristics of the target, and further determining a process noise covariance and an observation noise covariance of the target;

constructing a Kalman filter first model according to the process noise covariance and the observation noise covariance;

and introducing various suboptimal fading factors into the process noise covariance matrix, and transforming the first Kalman filter model into a third Kalman filter model.

6. The radar target tracking method based on the strong tracking filtering according to claim 1, wherein the strong tracking filtering of the target comprises:

acquiring motion state information of a target at the current moment as an observed value of the motion state of the target at the current moment;

acquiring a strong tracking filtering result of a target at the previous moment; the strong tracking filtering result comprises an optimal estimation of the motion state at the previous moment and an estimation error covariance of the motion state at the previous moment;

obtaining a prediction error covariance of the motion state at the current moment according to the estimation error covariance at the previous moment, and determining the filtering gain at the current moment according to the prediction error covariance at the current moment so as to determine the strong tracking filter model at the current moment;

according to the optimal estimation of the previous moment, combining the observed value of the current moment and based on the strong tracking filter model of the current moment, obtaining the optimal estimation of the current moment; acquiring an estimation error covariance of the current moment according to the filter gain of the current moment and by combining an observed value of the current moment; and the estimation error covariance of the current moment and the optimal estimation of the current moment are the strong tracking filtering result of the current moment.

Technical Field

The invention belongs to the technical field of target tracking, and relates to a radar target tracking method based on strong tracking filtering.

Background

The target tracking of the radar is one of important directions in the application of radar technology, and the distance, the distance change rate (radial speed), the direction, the height and other motion state information from a target to an electromagnetic wave emission point is calculated and obtained by transmitting a radio wave signal and receiving the reflected radio wave signal, so that the accurate position of the target is obtained.

In the target tracking process of the radar, the acquired motion state information needs to be filtered, wherein strong tracking filtering is a common filtering method in radar target tracking, and can achieve a high-precision target tracking effect. However, in the current research, the adaptive estimation of the target tracking model parameters is very important in the target tracking process, and the lack of tracking of the model parameters may cause the analysis and adjustment of the model parameters by researchers to be inadequate, thereby increasing the tracking difficulty of the target motion state.

Disclosure of Invention

In view of the defects in the prior art, the present invention aims to provide a radar target tracking method based on strong tracking filtering, which is used for solving the problems that the model parameters are unknown due to the fact that the existing strong tracking filtering method cannot estimate the model parameters in real time, so that the blindness of model adjustment and the opacity in analysis are increased, and the like.

In order to achieve the above and other related objects, the present invention provides a radar target tracking method based on strong tracking filtering, which is suitable for a radar target tracking process of a target motion state, and the method includes: according to the motion characteristics of the target, a strong tracking filter model for carrying out strong tracking filtering on the motion state of the target is constructed by introducing a suboptimal fading factor; and repeatedly executing the strong tracking filtering of the target based on the strong tracking filter model to obtain the motion state tracking information of the target.

In an embodiment of the present invention, the constructing a strong tracking filter model for performing strong tracking filtering on the motion state of the target by introducing a suboptimal fading factor according to the motion characteristic of the target includes: determining a motion model of the target according to the motion characteristics of the target, and further determining a process noise covariance and an observation noise covariance of the target; constructing a Kalman filter first model according to the process noise covariance and the observation noise covariance; and introducing a suboptimal fading factor into the process noise covariance matrix, and transforming the first Kalman filter model into a second Kalman filter model.

In an embodiment of the present invention, said introducing a sub-optimal cancellation factor into said process noise covariance matrix, and transforming said Kalman filter first model into a Kalman filter second model, includes: constructing a suboptimal fading factor, and introducing the suboptimal fading factor into the process noise covariance; calculating the suboptimal fading factor by adopting an approximate suboptimal method; and transforming the first Kalman filter model into a second Kalman filter model based on the solved suboptimum fading factor.

In an embodiment of the present invention, the introducing a sub-optimal cancellation factor into the process noise covariance matrix to transform the Kalman filter first model into the Kalman filter second model further includes: and introducing a weakening factor beta for weakening the regulation effect of the suboptimal fading factor in the process of constructing the suboptimal fading factor.

In an embodiment of the present invention, the constructing a strong tracking filter model for performing strong tracking filtering on the motion state of the target by introducing a suboptimal fading factor according to the motion characteristic of the target includes: as described above, the present invention provides a device. Determining a motion model of the target according to the motion characteristics of the target, and further determining a process noise covariance and an observation noise covariance of the target; constructing a Kalman filter first model according to the process noise covariance and the observation noise covariance; and introducing various suboptimal fading factors into the process noise covariance matrix, and transforming the first Kalman filter model into a third Kalman filter model.

In an embodiment of the present invention, the strong tracking filtering of the target includes: acquiring motion state information of a target at the current moment as an observed value of the motion state of the target at the current moment; acquiring a strong tracking filtering result of a target at the previous moment; the strong tracking filtering result comprises an optimal estimation of the motion state at the previous moment and an estimation error covariance of the motion state at the previous moment; obtaining a prediction error covariance of the motion state at the current moment according to the estimation error covariance at the previous moment, and determining the filtering gain at the current moment according to the prediction error covariance at the current moment so as to determine the strong tracking filter model at the current moment; according to the optimal estimation of the previous moment, combining the observed value of the current moment and based on the strong tracking filter model of the current moment, obtaining the optimal estimation of the current moment; acquiring an estimation error covariance of the current moment according to the filter gain of the current moment and by combining an observed value of the current moment; and the estimation error covariance of the current moment and the optimal estimation of the current moment are the strong tracking filtering result of the current moment.

As described above, according to the radar target tracking method based on strong tracking filtering provided by the present invention, the Kalman filter model is improved by constructing the suboptimal fading factor into the process noise covariance, so that in the target tracking process, based on the observed value and the strong tracking filtering result at the previous moment, the real-time estimation and dynamic adjustment of the parameters in the filter model are realized, and the target tracking effect and stability can be effectively improved.

Drawings

Fig. 1 is a schematic flow chart of a strong tracking filtering-based radar target tracking method in an implementation 1 according to the present invention;

FIG. 2 is a flowchart illustrating the sub-steps included in step S100 according to an embodiment;

FIG. 3 is a flow diagram illustrating the implementation of the strong tracking filtering at time k in one embodiment;

FIG. 4 is a diagram illustrating an embodiment of tracking information obtained by different methods for a motion state of a target;

FIG. 5 is a diagram illustrating the root mean square error of tracking information of a moving state of an object obtained by different methods according to an embodiment;

description of the element reference numerals

S100 to S200

S101 to S104

S201 to S204

Detailed Description

The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict.

It should be noted that the drawings provided in the following embodiments are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number, shape and size of the components in actual implementation, and the type, quantity and proportion of the components in actual implementation may be changed freely, and the layout of the components may be more complicated.

Example 1

Referring to fig. 1, a schematic flow chart of the radar target tracking method based on strong tracking filtering according to embodiment 1 of the present invention is shown. In this embodiment, the target is in a uniform motion state.

As shown in fig. 1, the method comprises the following steps:

s100, according to the motion state characteristics of the target, constructing a strong tracking filter model by introducing a suboptimal fading factor, wherein the model is used for carrying out strong tracking filtering on the motion state of the target;

as shown in fig. 2, the step S100 specifically includes the following sub-steps:

s101, determining a motion model of the target according to the motion characteristics of the target;

specifically, according to the uniform motion characteristic of the target, determining a motion model of the target, which is as follows:

x_k＝F_k,k-1x_k-1+w_k,k-1 (1)

Z_k＝H_kx_k+u_k； (2)

wherein the content of the first and second substances,

in the formula, k represents the current time, namely the time of the radar observing the motion state of the target; k-1 represents the time immediately preceding the time k;

x_krepresenting a motion state vector of the target at the k moment; x '(k), y' (k) represent the position of the target at time k on the x and y coordinates, respectively; v. of_x(k)，v_y(k) Respectively representing the speed of the target in the x coordinate direction and the y coordinate direction at the k time;

F_k,k-1representing a motion state transition matrix of a target motion system from k-1 time to k time, wherein T represents an observation sampling period of the radar;

w_k,k-1is the process noise of Gaussian white noise from the time k-1 to the time k, and satisfies w_k,k-1∈RⁿI.e. included in the n-dimensional real number domain, and satisfies that the mean is zero and the covariance is:

Q_k,k-1∈R^n×n

wherein Q is_k,k-1Process noise w from time k-1 to time k_k,k-1The covariance of (a);

Z_kthe observation vector of the motion state vector at the k moment is obtained by radar observation and meets the following requirements:

Z_k∈R^p

wherein R is^pIs a p-dimensional real number domain;

H_kthe observation matrix representing time k is:

H_k＝[1 0 0] (5)

u_kthe observation noise of Gaussian white noise at the moment k and the measurement error of the radar satisfy u_k∈R^pAnd the mean is zero and the covariance is:

R_k∈R^p×p

wherein R is_kObserving noise v for time k_kThe covariance of (a);

the process noise and the observation noise are not correlated.

S102, observing the motion state of the target, and acquiring the motion state information of the target at the k moment; obtaining the observation noise covariance R based on the motion state information obtained by observation_k。

S103, constructing a first Kalman filter model of target tracking based on the process noise covariance and the observation noise covariance.

Specifically, the process noise covariance is randomly generated when the model solves each motion state vector, and the observation noise covariance is obtained in step S102.

The first Kalman filter model is an initial iteration equation of the Kalman filter and comprises the following steps:

wherein the content of the first and second substances,estimating the prior state of the motion state at the k moment based on the optimal estimation at the k-1 moment;

the method is the optimal estimation of the target motion state at the moment k, namely, the value obtained after the noise influence of the motion state observed value at the moment k is removed through strong tracking filtering;

the optimal estimation of the target motion state at the moment of k-1 is carried out;

K_kin order to filter the gain matrix of the filter,a transposed matrix which is a filter gain matrix at the time k;

a transposed matrix which is an observation matrix at the time k;

P_k|k-1the prediction error covariance at the k moment is obtained based on the motion state at the k-1 moment;

P_k|kthe covariance of the estimation error at time k;

and I is an identity matrix.

And S104, introducing a suboptimal fading factor into the process noise covariance matrix, and transforming the first Kalman filter model into a second Kalman filter model.

The method specifically comprises the following steps:

S104A, constructing a suboptimal fading factor lambda_kAnd applying said sub-optimal fading factor lambda_kIntroducing the process noise covariance matrix to indirectly adjust the process noise covariance matrix P_k|k-1；

To properly increase the tuning effect of the process noise covariance, a suboptimal cancellation factor λ is constructed_kComprises the following steps:

the prediction error covariance (equation 8) is transformed into:

S104B, solving the suboptimal fading factor lambda_k；

According to the principle of orthogonalityAnd formula (8) can be derived:

namely:

wherein, V_kThe covariance of the innovation at time k is approximated by:

wherein the content of the first and second substances,

in the above formula, the γ₁Tracking an innovation of the initial moment, gamma, for the target_kIs an innovation matrix for the time instant k,the new information matrix at the moment k is transposed, rho is a forgetting factor, and the following conditions are met:

0＜ρ≤1

then, formula (10) is substituted for formula (11) to obtain:

then:

for convenience of expression, let:

then

S104C, based on the calculated suboptimum fading factor, the first Kalman filter model is transformed into a second Kalman filter model.

Further, the step S104 further includes:

in the N_kThe attenuation factor beta is introduced to avoid overshoot.

When lambda is_kWhen too large, λ will be caused_kQ_k,k-1The excessive adjustment leads to the over-adjustment, so that the noise proportion is increased, the filtering difficulty is increased, and the result precision of the target motion state estimation is reduced; therefore, to attenuate λ_kSo that λ is_kThe adjustment of the noise term is constrained, and a weakening factor beta is introduced.

Specifically, in said N_kIntroducing said attenuation factor β, converting formula (19) to:

substituting formula (22) into formula (21) to obtain new suboptimal fading factor lambda_k。

And S200, repeatedly executing the strong tracking filtering of the target based on the strong tracking filter model, thereby obtaining the motion state tracking information of the target.

Specifically, the single strong tracking filtering includes acquiring motion state information of the target at a certain moment through a radar, and acquiring a strong tracking filtering result of the target at a previous moment of the certain moment; and obtaining a strong tracking filtering result corresponding to the previous moment based on the strong tracking filter model constructed in the step S100 according to the strong tracking filtering result of the previous moment and the motion state information of the previous moment, where the strong tracking filtering result includes the motion state tracking information of the target at the previous moment.

And repeatedly executing the strong tracking filtering process of the target until the target tracking process is finished, thereby obtaining the tracking information of the motion state of the target.

To explain the strong tracking filtering process in detail, please refer to fig. 3, which shows a schematic flow chart of performing the strong tracking filtering at time k, including:

s201, acquiring motion state information of a target at the time k as an observed value of the motion state of the target at the time k;

wherein the motion state information includes position information of the object, and velocity information.

S202, acquiring a strong tracking filtering result of a target at the moment k-1, wherein the strong tracking filtering result comprises optimal estimation at the moment k-1 and estimation error covariance at the moment k-1;

wherein the k-1 moment is the last moment of the k moment; the strong tracking filter result at the time k-1 is the strong tracking filter result obtained by performing the strong tracking filter process at the previous time, i.e., the optimal estimate at the time k-1 (obtained by equation 7) obtained by the Kalman filtering second model at the time k-1, and the estimated error covariance at the time k-1 (obtained by equation 10).

It should be noted that, at an initial time of target tracking, that is, when k is equal to 1, the optimal estimation at the previous time is a target motion state preset value obtained before target tracking is performed, that is:

when k is equal to 0, the reaction solution is,

wherein x is₀Is a preset value of the motion state of the target.

Similarly, at the initial time of target tracking, that is, when k is equal to 1, the estimation error covariance at the previous time is the estimation error covariance preset value of the target motion state obtained before target tracking is performed.

S203, acquiring a prediction error covariance at the k moment according to the estimation error covariance at the k-1 moment; determining the filtering gain at the k moment according to the prediction error covariance at the k moment, so as to determine a strong tracking filter model at the k moment;

that is, the prediction error covariance at the time k is obtained by equation 9 based on the obtained prediction error covariance at the time k-1, and the filter gain at the time k is determined based on equation 8, thereby determining the strong tracking filter model at the time k.

S204, according to the optimal estimation of the k-1 moment, combining an observed value of a target motion state at the k moment and based on the strong tracking filter model at the k moment determined in the step S203, obtaining the optimal estimation of the k moment; acquiring an estimation error covariance at the k moment according to the filter gain at the k moment determined in the step S203 and by combining an observed value of a target motion state at the k moment; and the covariance of the estimation error at the moment k and the optimal estimation at the moment k are taken as strong tracking filtering results at the moment k so as to iteratively execute the acquisition filtering process at the next moment.

In this embodiment, compared to the prior art in which the time-varying fading factor is directly used to adjust the covariance matrix of the prediction error, the present invention adjusts the covariance matrix of the process noise by using the suboptimal fading factor, and then indirectly adjusts the covariance matrix of the prediction error P_k|k-1The method can solve the problem that the Kalman filtering method in the prior art cannot estimate the model parameters in real time, so that the tracking accuracy of the radar target is reduced. Further, the sub-optimal fading factor is constrained by introducing a weakening factor into the sub-optimal fading factorThe influence on the filtering result due to the overlarge suboptimal fading factor can be avoided, and the precision of the target motion state estimation result can be effectively guaranteed.

Example 2

In this embodiment, the implementation steps of the radar target tracking method based on strong tracking filtering are the same as those in embodiment 1, except that:

in step S200, introducing a multiple fading matrix into the process noise covariance matrix to adjust a prediction error covariance matrix, and transforming the Kalman filter first model into a Kalman filter third model;

in particular, the matrix L will be multiply faded_kIntroducing the process noise covariance matrix, thereby adjusting the prediction error covariance matrix, and adjusting equation (9) as:

wherein the multiple vanishing matrix L_kExpressed as:

wherein the content of the first and second substances,respectively, representing different sub-optimal fading factors.

Equation (17) then translates to:

traces are found on both sides of equation (25):

the equation (26) is simplified as follows:

tr[L_kM_k]＝tr[N_k] (27)

substituting (24) into (27) to obtain:

thus, according to equation (28), it is possible to obtain:

wherein the content of the first and second substances,

representation matrix N_kThe element of the ith row and ith column,representation matrix M_kRow ith and column ith.

In this embodiment, in constructing the Kalman filtering model, by introducing multiple suboptimal fading factors, each data channel in the noise matrix can be adjusted more specifically than by introducing a single suboptimal fading factor.

In order to verify that the radar target tracking method based on strong tracking filtering provided by the invention is applied to a target tracking system with process noise errors, the tracking effect of the radar target tracking method is superior to that of a common strong tracking filtering target tracking method, and the following simulation is carried out, wherein an unmodified strong tracking filtering method is adopted by a comparison group.

Setting the target object to be tracked to be in uniform motion, setting the initial position to be (-120, -20) and the initial speed to be (15, 5). And obtaining the motion state information of the target by obtaining the distance from the target to the radar according to the radar, namely, tracking the target. The scan period of the radar is set to 0.8, and other parametric models of the simulation system are set as follows:

and when the target tracks the initial moment, setting the preset value of the motion state of the target and the preset value of the covariance of the estimation error as follows:

the experimental interval was 1s and 1000 monte carlo simulations were performed.

And finally, respectively acquiring target motion state tracking information obtained after the simulation of each method and acquiring the root mean square error of each target motion state tracking information as corresponding simulation results. The result of the target motion state tracking information obtained after the simulation of each method is shown in fig. 4, and the result of the root mean square error of the target motion state tracking information is shown in fig. 5. As is apparent from fig. 4 and 5, the method of the present invention has a good tracking effect on the motion trajectory of the target, the tracking trajectory of the filter output is still stable under the influence of the noise Q outside the model, and the accuracy is considerable, compared with the comparison group, the method is closer to the real trajectory, and the root mean square error is smaller than the root mean square error of the comparison group, and the total root mean square error is kept within 0 to 12, so that the method can meet the general industrial application requirements. The experimental result can be concluded that, in the target tracking problem of the radar, compared with the non-improved strong tracking filtering, the radar target tracking method based on the strong tracking filtering provided by the invention has better tracking precision and robustness.

In summary, compared with the existing target tracking method, the radar target tracking method based on the strong tracking filtering improves the Kalman filter model by constructing the suboptimal fading factor into the process noise covariance, so that in the target tracking process, the real-time estimation and dynamic adjustment of parameters in the filter model are realized based on the observed value and the strong tracking filtering result at the previous moment, and the target tracking effect and stability can be effectively improved.

The foregoing embodiments are merely illustrative of the principles and utilities of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.