Orientation-pure multi-target trajectory tracking method in multi-interference environment

Abstract

The invention discloses an azimuth-pure multi-target trajectory tracking method in a multi-interference environment, which belongs to the technical field of underwater acoustic array signal processing, and comprises the following steps of: S1, processing original data received by a sensor array by using a traditional azimuth estimation method to obtain an estimation result of an azimuth angle; s2, performing linear filtering on the azimuth angle result to reduce interference in the azimuth angle estimation result; s3, carrying out multi-target pure azimuth target trajectory tracking; according to the invention, under the condition that the multi-interference azimuth estimation result is input, linear filtering is carried out by using the GM-PHD method, so that the interference of the azimuth trajectory can be reduced and the leak detection condition can be improved; the EK-PHD method is used for carrying out centralized fusion on multi-sensor receiving measurement and carrying out multi-target trajectory tracking, so that multi-target pure azimuth trajectory tracking under multiple interferences is realized.

Description

Technical Field

[0001] The invention belongs to the technical field of underwater acoustic array signal processing, and in particular relates to a bearing-only multi-target trajectory tracking method in a multi-interference environment. Background Art

[0002] In complex environments, factors such as the number of targets and interference are constantly changing. How to accurately determine the number and status of targets based on measurement data is a huge challenge facing underwater target tracking technology. Traditional pure bearing target tracking technology based on data association is difficult to break through the bottleneck of large computational workload, and is prone to mistracking or loss of tracking, and is unable to estimate the number of targets. The tracking method based on random finite sets is performed in a set sense, which greatly reduces the computational complexity and can accurately track targets when the number of targets is unknown or variable.

[0003] At present, target tracking technologies based on random finite sets are mostly implemented for single nodes. Pure bearings target tracking is unobservable by a single sensor, and in the case of multiple interferences, when only the EK-PHD method is used to track the target trajectory, problems such as inaccurate target number estimation and missed detection often occur. Therefore, a pure bearings multi-target trajectory tracking method in a multi-interference environment is needed to solve the above problems. Summary of the invention

[0004] The object of the present invention is to provide a bearing-only multi-target trajectory tracking method in a multi-interference environment to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: a bearing-only multi-target trajectory tracking method in a multi-interference environment, comprising the following steps: Step S1, using a traditional azimuth estimation method to process the raw data received by the sensor array to obtain an azimuth angle estimation result; Step S2, performing linear filtering on the azimuth result to reduce interference in the azimuth estimation result; Step S3: performing multi-target bearing-only target trajectory tracking.

[0006] This method uses the GM-PHD filter to perform linear filtering on the azimuth estimation results under multiple interferences, and obtains the azimuth angle estimation result with continuous azimuth trajectory and less interference, which is used as the measurement input of the pure azimuth target trajectory tracking problem; secondly, a nonlinear relationship between the target motion trajectory and the measured azimuth angle is established, and the target motion trajectory is tracked using the EK-PHD filter to obtain a continuous target motion trajectory.

[0007] As a preferred implementation, in step S1, the estimated result of the azimuth angle is called .

[0008] As a preferred implementation, step S1 specifically includes estimating the azimuth of the original data received by the array, estimating the azimuth angle using methods such as CBF, MVDR, and Music, and performing comparative analysis on different frequency bands according to the time-frequency analysis diagram to find a set of azimuth estimation results with reasonable errors. .

[0009] As a preferred implementation, in step S2, As the measurement input in step S2, a linear input-output relationship is established, and the GM-PHD filter is used to filter the discontinuous multi-interference Linear filtering is performed to obtain a less-interfered and continuous azimuth trajectory result, called .

[0010] As a preferred implementation, step S2 includes the following steps: Step S201, establishing a linear filtering model, including a target azimuth state model and a measurement model; Step S202: algorithm initialization; Step S203: prediction; Step S204: Update; Step S205, trimming and merging Gaussian components; Step S206, extracting the target state; Step S207: Estimate the number of targets.

[0011] As a preferred implementation, in step S3, using As a measurement.

[0012] As a preferred implementation, step S3 includes the following steps: Step S301, establishing a nonlinear filtering model, including a target state model and a measurement model; Step S302: algorithm initialization; Step S303: prediction; Step S304: Update; Step S305: trimming and merging Gaussian components; Step S306, extracting the target state; Step S307, estimating the target number; Step S308: OSPA tracking error analysis.

[0013] Compared with the prior art, the present invention has the following beneficial effects: The present invention, when the input is a multi-interference azimuth estimation result, uses the GM-PHD method to perform linear filtering, which can reduce the interference of the azimuth trajectory and improve the missed detection situation; The present invention uses the filtered azimuth result as a nonlinear filtering input, uses the EK-PHD method to centrally fuse multi-sensor receiving measurements and performs multi-target trajectory tracking, thereby realizing multi-target pure azimuth trajectory tracking under multiple interferences. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flowchart of the present invention; Figure 2 This is a diagram showing the implementation of the multi-target test in the Swellex96 sea trial data set processed by the present invention; Figure 3 It is a schematic diagram of the receiving system and target position of the present invention; Figure 4 This is a diagram of the traditional azimuth estimation result of the present invention; Figure 5 This is the orientation result diagram after GM-PHD filtering of the present invention; Figure 6 This is a schematic diagram of multi-target tracking results of the present invention; Figure 7 This is a comparison chart of the target number estimation results of the present invention; Figure 8 This is a comparison diagram of the OSPA error results of the present invention. DETAILED DESCRIPTION

[0015] The present invention will be further described below in conjunction with the embodiments.

[0016] The following examples are used to illustrate the present invention, but they cannot be used to limit the scope of protection of the present invention. The conditions in the examples can be further adjusted according to specific conditions. Simple improvements to the method of the present invention under the premise of the concept of the present invention belong to the scope of protection claimed in the present invention.

[0017] See also Figure 1-8 The present invention provides a bearing-only multi-target trajectory tracking method in a multi-interference environment, comprising the following steps: Step S1: Use the traditional azimuth estimation method to process the raw data received by the sensor array to obtain the azimuth estimation result. The azimuth estimation result is called ; Specifically, it includes azimuth estimation of the original data received by the array, azimuth estimation using CBF, MVDR, Music and other methods, comparative analysis of different frequency bands based on time-frequency analysis diagrams, and finding a set of azimuth estimation results with reasonable errors. ; Step S2, performing linear filtering on the azimuth result to reduce interference in the azimuth estimation result; Will As the measurement input in step S2, a linear input-output relationship is established, and the GM-PHD filter is used to filter the discontinuous multi-interference Linear filtering is performed to obtain a less-interfered and continuous azimuth trajectory result, called ; Step S2 includes the following steps: Step S201, establishing a linear filtering model, including a target azimuth state model and a measurement model; Considering that the motion process of underwater targets is a random process, assuming that the change of target azimuth satisfies the CV model, that is, the uniform motion model, let The time azimuth is , the azimuth angle change rate is , the target state vector is expressed as , then the kinematic model of the target orientation is expressed as: ; in, is the state transfer matrix, For the The state of the moment, represents zero mean covariance, and the variance matrix is Gaussian process noise; Using the target azimuth estimate with certain errors and interference obtained by the traditional azimuth method as the measurement, the Measurement of time It can be expressed as: ; in, represents the measurement matrix, , is the position estimation error, which is modeled as zero mean with variance Gaussian measurement noise; Assume that Always present targets, there are Measurement, then The target azimuth state set at time is expressed as , The target orientation set at all times is expressed as The measurement set is represented as , The measurement set at all times is expressed as ; Step S202: algorithm initialization; The recursive operation of the GM-PHD method transfers the intensity function. The multi-objective state intensity function can be written in the form of a Gaussian mixture. , which can be expressed as: ; In the formula, is the weight of each Gaussian component, , denote the mean and variance of the Gaussian components respectively.

[0018] Step S203: prediction; Assumptions The intensity function at time is ,and Satisfies the Gaussian mixture model: ; After the prediction step, the posterior strength function can be expressed as: , Part I is the posterior strength of the survival target, and the expression can be written as: , where the mean is expressed as , the variance is expressed as ; The second part is the posterior strength of the new target, then After the moment is predicted, its posterior probability expression can be written as: , the posterior intensity function can be written as: ; Step S204: Update; Based on the target motion model, the state equation and measurement equation are combined to obtain The intensity function at the moment is expressed as: , where: , , , , , ; When the target is missed, the weight can be expressed as: ,in, represents the detection probability; Step S205, trimming and merging Gaussian components; After accumulation of multiple time steps, the number of Gaussian components is often large. After trimming and merging the weights of each Gaussian component, a reasonable number of Gaussian mixture components is obtained to obtain the Gaussian mixture posterior strength: ; Step S206, extracting the target state; GM-PHD filter posterior strength The Gaussian mixture expression is easy to extract multi-target state estimation, and the weights are generally selected The Gaussian component greater than 0.5 is taken as the target orientation information at that moment; Step S207, estimating the target number; The sum of the weights of the Gaussian components remaining after step S206 is the estimated number of targets: ; Step S3: perform multi-target pure bearing target trajectory tracking using As a measurement; Step S3 includes the following steps: Step S301, establishing a nonlinear filtering model, including a target state model and a measurement model; Step S302: algorithm initialization; Step S303: prediction; Step S304: Update; Step S305: trimming and merging Gaussian components; Step S306, extracting the target state; Step S307, estimating the target number; Step S308, OSPA tracking error analysis; Steps S302-S303 are the same as steps S202-S203, and steps S305-S307 are the same as steps S205-S207, which are: based on the target motion model, combining the state equation and the measurement equation to obtain The intensity function at the moment is expressed as: , where: , , , , , ; Using OSPA distance to analyze the PHD algorithm tracking error can intuitively reflect the tracking accuracy of the two different algorithms.

[0019] The pure bearings target tracking method models the target state space and measurement space into a random finite set form, thereby converting it into a set-valued filtering problem, extending the Gaussian mixture probability density hypothesis method (GM-PHD) to the nonlinear field, combining the extended Kalman method (EKF) with the PHD filter, namely EK-PHD, uploading the measurement information obtained by each sensor to the fusion center for filtering to obtain a fusion estimate of the target state; this method preprocesses the bearing estimation results under multiple interferences, and reduces the number of interferences and missed detections through GM-PHD filtering; then the filtered bearing results are used as the filter input of the pure bearings target tracking problem, and the EK-PHD method is used to perform pure bearings target tracking.

[0020] The working principle and use process of the present invention: Taking a typical underwater target positioning scenario as an example, an implementation example of the present invention is given; the method of the present invention is used to process a multi-target test in the Swellex96 sea trial data set, and the specific technical implementation method is as follows; like Figure 3 As shown, the receiving system has two horizontal line arrays, the south array consists of 26 elements and the north array consists of 27 elements. The target moves near the north array, and the jammer approaches the north and south arrays from far to near. The frequency of the narrowband signal emitted by the target is between 49-400 Hz; like Figure 4 and Figure 5 It can be seen that the filtered result has less interference and more accurate estimation results; like Figure 6 It can be seen that this method can estimate the target motion trajectory more accurately under multiple interference conditions, where this experiment includes two known targets and interferences and one unknown interference; For known targets and disturbances in approximately uniform linear motion, the estimated trajectory is relatively accurate, which shows that this method has good estimation performance for uniform linear motion targets; like Figure 7 It can be seen that this method has good target number estimation performance and reasonable OSPA error; According to the implementation example, the pure bearing multi-target trajectory tracking method can make a more accurate estimate of the number of targets under multiple interference azimuth angle inputs and achieve more precise target trajectory tracking.

[0021] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A bearing-only multi-target trajectory tracking method in a multi-interference environment, characterized by: The following steps are involved: Step S1, using a traditional azimuth estimation method to process the raw data received by the sensor array to obtain an azimuth angle estimation result; Step S2, performing linear filtering on the azimuth result to reduce interference in the azimuth estimation result; Step S3: performing multi-target bearing-only target trajectory tracking.

2. The method for tracking multi-target trajectories in a multi-interference environment according to claim 1, characterized in that: In step S1, the estimated result of the azimuth is called .

3. The method for tracking multi-target trajectories in a multi-interference environment according to claim 1, characterized in that: Step S1 specifically includes estimating the azimuth of the original data received by the array, estimating the azimuth angle using methods such as CBF, MVDR, and Music, and performing comparative analysis on different frequency bands based on the time-frequency analysis diagram to find a set of azimuth estimation results with reasonable errors. .

4. The method for tracking multi-target trajectories in a multi-interference environment according to claim 1, characterized in that: In step S2, As the measurement input in step S2, a linear input-output relationship is established, and the GM-PHD filter is used to filter the discontinuous multi-interference Linear filtering is performed to obtain a less-interfered and continuous azimuth trajectory result, called .

5. The method for tracking multi-target trajectories in a multi-interference environment according to claim 1, characterized in that: Step S2 includes the following steps: Step S201, establishing a linear filtering model, including a target azimuth state model and a measurement model; Step S202: algorithm initialization; Step S203: prediction; Step S204: Update; Step S205, trimming and merging Gaussian components; Step S206, extracting the target state; Step S207: Estimate the number of targets.

6. The method for tracking multi-target trajectories in a multi-interference environment according to claim 1, characterized in that: In step S3, using As a measurement.

7. The method for tracking multi-target trajectories in a multi-interference environment according to claim 1, characterized in that: Step S3 includes the following steps: Step S301, establishing a nonlinear filtering model, including a target state model and a measurement model; Step S302: algorithm initialization; Step S303: prediction; Step S304: Update; Step S305: trimming and merging Gaussian components; Step S306, extracting the target state; Step S307, estimating the target number; Step S308: OSPA tracking error analysis.

Images