A practical long-time multi-frame TBD detection algorithm for radar targets

Abstract

The application relates to the technical field of radar target detection, in particular to a practical long-time multi-frame TBD detection algorithm for radar targets, which comprises the following steps: pre-processing and plot extraction are performed on target echo data; a track characteristic signal function is constructed; data is screened and the track characteristic signal function is recursively updated; the screened and optimized track characteristic signal function is subjected to characteristic signal processing; and the results after processing are analyzed and output. The practical long-time multi-frame TBD detection algorithm for radar targets has innovatively introduced a new characteristic representation method, that is, a track characteristic signal function, which comprehensively reflects target motion characteristics through phase, frequency and frequency modulation slope, is different from a traditional energy accumulation method, improves the detection accuracy and robustness, has good applicability in low SCR, target fluctuation and simultaneous multi-target application environments, has relatively low calculation complexity, and is beneficial to engineering implementation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar target detection, and particularly relates to a practical long-time multi-frame TBD detection algorithm for radar targets. BACKGROUND

[0002] With the development of stealth technology and the application of various unmanned platforms, the observability of radar targets is continuously reduced, and the MF-TBD faces more severe low signal-to-clutter ratio (SCR) conditions, which requires longer accumulation detection time. This leads to the following problems: on the one hand, the classic radar MF-TBD algorithm cannot efficiently complete the search of the potential motion trajectory of the target, and there is a prominent contradiction between the accumulation gain and the algorithm running efficiency; on the other hand, the classic MF-TBD algorithm mainly relies on energy accumulation, and under the condition of extremely low SCR, the so-called "SNR threshold" phenomenon occurs, and the accumulation gain is close to 0. In addition, the existing classic radar MF-TBD algorithm has the problem of poor target environment adaptability to varying degrees, and there is still a lack of universal engineering solutions for various application backgrounds such as target maneuvering, target fluctuation and multiple targets. For example, the HT-TBD algorithm is difficult to adapt to non-linear motion targets; the DP-TBD algorithm and the PF-TBD algorithm have great improvement in maneuvering target detection and tracking, but the calculation amount is significantly increased in the presence of multiple targets, and the performance is seriously deteriorated; the RFS-TBD algorithm can adapt to the needs of maneuvering multi-target detection and tracking, but the algorithm calculation amount is huge, and the engineering implementation is still difficult. Therefore, we propose a practical long-time multi-frame TBD detection algorithm for radar targets. SUMMARY

[0003] The main purpose of the present application is to provide a practical long-time multi-frame TBD detection algorithm for radar targets and a preparation method, which can effectively solve the problems in the background art.

[0004] To achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0005] A practical long-time multi-frame TBD detection algorithm for radar targets comprises the following steps:

[0006] S1: pre-processing and plot extraction of target echo data;

[0007] S2: constructing a track feature signal function;

[0008] S3: screening and recursively updating the track feature signal function;

[0009] S4: feature signal processing of the screened and optimized track feature signal function;

[0010] S5: analyzing and outputting the results after processing.

[0011] Preferably, in the S1, the preprocessing method comprises missing value processing, outlier detection and processing, duplicate data processing and data consistency check.

[0012] Preferably, in the S2, the step of constructing the track feature signal function is:

[0013] S201: defining feature representation, corresponding to the relative position, motion speed and acceleration of the historical associated point track by phase, frequency and frequency modulation slope respectively;

[0014] S202: ignoring interference factors, simplifying the model and focusing on describing the idealized target motion characteristics;

[0015] S203: mapping point track data, mapping point track data into TFS function, converting complex point track time series information into signal processing problem.

[0016] Preferably, in the S2, the function definition of the track feature signal function is:

[0017]

[0018] Wherein:

[0019]

[0020]

[0021] In the formula: is the track feature signal function, and is the coordinate position function, is the allowable state transition point track serial number set of the point track in the lth frame, ω i l is the phase of the lth frame, is the phase correction factor, and x and y are point track coordinates.

[0022] Preferably, in the S3, the step of screening the data is:

[0023] S301: setting the screening threshold of the track feature signal function of the suspected point track as γ1=η1N W , wherein 0<η1<1, N W is the total number of frames of the current track feature signal function screening window;

[0024] S302: after the basic condition that the current frame number n is greater than the starting frame number n0 is met, it is judged whether the total number of non-zero values in the track feature signal function is less than γ1;

[0025] S303: If the total number of non-zero values is less than γ1, the track feature signal function is considered as a false track of clutter and is eliminated, otherwise the function is reserved for subsequent recursive and updating calculation.

[0026] Preferably, in the S3, the recursive updating formula of the track feature signal function is:

[0027]

[0028]

[0029] Preferably, in the S4, the steps of the feature signal processing include:

[0030] S401: performing FRFT transformation on the X and Y component track feature signals of each suspected point track after updating and screening;

[0031] S402: performing target detection in the FRFT transformation space with a fixed threshold and extracting peak value parameters;

[0032] S403: directly performing zero processing on the spectral values lower than the threshold in the FRFT domain to realize filtering, and then applying inverse transformation to realize track recovery.

[0033] Preferably, in the S401, the formula of the FRFT transformation is:

[0034]

[0035] In the formula, α is the transformation order, K α (t, u) is the transformation kernel, x(t) is the function to be transformed, and X α (u) is the transformed function.

[0036] Preferably, in the S402, the construction formula of the fixed threshold is

[0037]

[0038] In the formula, γ2 is a target confirmation threshold set according to the false alarm probability, which can be set as γ2 = η2N, wherein η2 can be set to 0.5-0.8.

[0039] Preferably, in the S403, the formula of the FRFT inverse transformation is:

[0040]

[0041] Compared with the prior art, the present application has the following beneficial effects:

[0042] 1. The application innovatively introduces a new feature representation method, i.e., a track feature signal function, which is used to describe the global credibility of a suspected point track and comprehensively reflects the target motion characteristics through phase, frequency and frequency modulation slope, thereby improving the accuracy and robustness of detection.

[0043] 2. The application proposes a track feature signal function recursive updating mechanism based on historical information, which dynamically optimizes the track feature signal calculation of the current frame by using the detection results of previous frames, and effectively eliminates false point tracks caused by non-targets by combining with the maximum expected speed constraint, thereby enhancing the adaptability and anti-interference ability of the algorithm in a dynamically changing environment.

[0044] 3. The application uses FRFT to transform the track feature signal, realizes effective accumulation of signal energy and clutter suppression. As an advanced signal processing tool, FRFT has superiorities in processing non-stationary signals and signals with specific modulation characteristics compared with traditional Fourier transform, thereby improving the weak target detection performance of the algorithm in a complex background. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 It is a principle block diagram of a practical radar target long-time multi-frame TBD detection algorithm of the application.

[0046] Figure 2 It is a target and clutter point track spatial position distribution diagram in a single simulation of the embodiment;

[0047] Figure 3 It is the actual total number of clutter point tracks of each frame in the single simulation of the embodiment;

[0048] Figure 4 It is a target TFS function under X coordinate in the single simulation of the embodiment;

[0049] Figure 5 It is a target TFS function under Y coordinate in the single simulation of the embodiment;

[0050] Figure 6 It is an energy accumulation diagram of the target TFS signal in the FRFT domain in the single simulation of the embodiment. DETAILED DESCRIPTION

[0051] In order to make the technical means, creative features, purposes and effects realized by the application easy to understand, the application is further described below in combination with specific embodiments.

[0052] EMBODIMENT

[0053] As Figure 1As shown, a practical radar target long-time multi-frame TBD detection algorithm is provided. First, the target echo data is pre-detected. In the pre-detection process, a predetermined target discovery probability p d0 A lower detection threshold γ2 is set. The signals detected through the threshold are subjected to condensation processing by a plot extractor and are sent to a subsequent multi-frame TBD detector for clutter plot suppression and target track recovery processing. The specific process is described as follows:

[0054] S1: pre-processing and plot extraction of target echo data;

[0055] The pre-processing method includes missing value processing, abnormal value detection and processing, repeated data processing and data consistency check.

[0056] S2: constructing a track feature signal function;

[0057] The step of constructing the track feature signal function is as follows:

[0058] S201: defining a feature representation, corresponding to the relative position, motion speed and acceleration of the historical associated plot by phase, frequency and frequency modulation slope;

[0059] S202: ignoring interference factors, simplifying the model and focusing on describing the idealized target motion characteristics;

[0060] S203: mapping plot data, mapping the plot data into the TFS function, and converting the complex plot time sequence information into a signal processing problem.

[0061] The function definition of the track feature signal function is as follows:

[0062]

[0063] Wherein:

[0064]

[0065]

[0066] In the formula: is the track feature signal function, and are coordinate position functions, is the set of allowable state transition plot numbers of the plot in the lth frame, ω i l is the phase of the lth frame, is the phase correction factor, and x and y are plot coordinates.

[0067] S3: filtering the data and recursively updating the track feature signal function;

[0068] The step of screening the data comprises:

[0069] S301: setting a screening threshold of a track feature signal function of a suspected track as γ1=η1N W wherein 0<η1<1, N W is the total number of frames of a screening window of the current track feature signal function;

[0070] S302: after the basic condition that the current frame number n is greater than the starting frame number n0 is met, it is judged whether the total number of non-zero values in the track feature signal function is less than γ1;

[0071] S303: if the total number of non-zero values is less than γ1, the track feature signal function is regarded as a false track of clutter and is removed, otherwise the function is reserved for subsequent recursive and updating calculation.

[0072] The recursive updating formula of the track feature signal function is:

[0073]

[0074]

[0075] S4: performing feature signal processing on the screened and optimized track feature signal function;

[0076] The step of the feature signal processing comprises:

[0077] S401: performing FRFT transformation on the X and Y component track feature signals of each suspected track after updating and screening;

[0078] Further, the formula of the FRFT transformation is:

[0079]

[0080] wherein α is the transformation order, K α (t,u) is the transformation kernel, x(t) is the function to be transformed, X α (u) is the transformed function.

[0081] S402: performing target detection in the FRFT transformation space with a fixed threshold, and extracting peak value parameters;

[0082] Further, the construction formula of the fixed threshold is

[0083]

[0084] In the formula, y2 is a target confirmation threshold set according to a false alarm probability, and y2=N / η2 can be set, where η2 is set to be between 0.5 and 0.8.

[0085] S403: The spectrum values below the threshold are directly processed by zeroing in the FRFT domain to achieve filtering, and then inverse transformation is applied to achieve track recovery.

[0086] Further, the formula of the FRFT inverse transformation is:

[0087]

[0088] S5: The results after processing are analyzed and output.

[0089] Simulation verification results

[0090] The target number is set to be 10, the target maximum motion speed is 40 m / s, the maximum motion acceleration is 1 m / s 2 . The distance resolution of the radar in the X and Y coordinates is 30 m, the antenna scanning period is 2 s, the standard deviation of the measurement error is 0.1 m, the observation range of the X and Y axes is 0-50 km, the average density of the single-frame clutter point track is 800, the clutter point track position is randomly and uniformly distributed in the observation range, and the accumulation frame number is 32.

[0091] The main algorithm parameters are set as follows: analysis distance, maximum allowable target continuous missing point frame number, TFS screening threshold factor, and screening observation window length.

[0092] It can be known from Figure 4 and Figure 5 that, after iterative updating, the target TFS signal is retained and the clutter point TFS signal is completely filtered out; before the TFS screening window works (i.e., the first 10 frames), the target TFS function is seriously deformed due to the influence of the clutter point track, and after the TFS function elimination processing based on the recursive maximum expected speed constraint, the TFS function presents obvious LFM signal characteristics, which meets the simulation target motion characteristic assumption.

[0093] It can be known from Figure 6 that the target echo has good energy aggregation in the FRFT domain, which can be smoothly detected by a general detection algorithm.

[0094] The application innovatively introduces a new feature representation method, i.e., a track feature signal function, which is used for describing the global credibility of a suspected point track, and comprehensively reflects the target motion characteristics through phase, frequency and frequency modulation slope, which is different from the traditional energy accumulation method, and improves the accuracy and robustness of detection; the application proposes a track feature signal function recursive updating mechanism based on historical information, uses the detection results of previous frames to dynamically optimize the track feature signal calculation of the current frame, and effectively eliminates false point tracks caused by non-targets by combining with the maximum expected speed constraint, thereby enhancing the adaptability and anti-interference ability of the algorithm in a dynamically changing environment; the application uses FRFT to transform the track feature signal, thereby realizing effective accumulation of signal energy and clutter suppression. As an advanced signal processing tool, FRFT has superiority in processing non-stationary signals and signals with specific modulation characteristics compared with the traditional Fourier transform, and improves the weak target detection performance of the algorithm in a complex background.

[0095] The basic principles and main features of the application and the advantages of the application are shown and described above. It should be understood by those skilled in the art that the application is not limited by the above examples, and the above examples and descriptions in the specification are only to illustrate the principles of the application, and various changes and improvements can be made to the application without departing from the spirit and scope of the application, and these changes and improvements all fall within the scope of the claimed application. The scope of protection of the application is defined by the appended claims and their equivalents.

Claims

1. A practical radar target long-time multi-frame TBD detection algorithm, characterized in that: It comprises: S1: pre-processing and plot extraction of target echo data; S2: constructing a track feature signal function; S3: screening data and recursively updating the track feature signal function; S4: processing the screened and optimized track feature signal function; S5: analyzing and outputting the results after processing; In the S2, the function definition of the track feature signal function is: Wherein: wherein: is a track feature signal function, is a coordinate position function, is a coordinate position function, is a set of allowable state transition point track indices of the point track in the lth frame, ω i l is a phase of the lth frame, is a phase correction factor, x and y are point track coordinates.

2. The utility radar target long-time multi-frame TBD detection algorithm according to claim 1, characterized in that: In the S1, the pre-processing method comprises missing value processing, outlier detection and processing, repeated data processing and data consistency check.

3. The utility radar target long-time multi-frame TBD detection algorithm according to claim 1, characterized in that: In the S2, the step of constructing the track feature signal function is: S201: defining feature representation, corresponding to the relative position, motion speed and acceleration of the historical associated plot by phase, frequency and frequency modulation slope respectively; S202: ignoring interference factors, simplifying the model and focusing on describing the idealized target motion characteristics; S203: mapping plot data, mapping plot data into TFS function, and converting complex plot time series information into signal processing problems.

4. The utility radar target long-time multi-frame TBD detection algorithm according to claim 1, characterized in that: In the S3, the step of screening data is: S301: set the screening threshold of the track feature signal function of the suspected point trace as γ1=η1N W where 0<η1<1, N W is the total frame number of the screening window of the current track feature signal function; S302: after the basic condition that the current frame number n is greater than the starting frame number n0 is met, it is judged whether the total number of non-zero values in the track feature signal function is less than γ1; S303: if the total number of non-zero values is less than γ1, the track feature signal function is regarded as a false clutter track and is removed, otherwise the function is reserved for subsequent recursive and update calculation.

5. The utility radar target long-time multi-frame TBD detection algorithm according to claim 1, characterized in that: In the S3, the recursive update formula of the track feature signal function is:

6. The utility radar target long-time multi-frame TBD detection algorithm according to claim 1, characterized in that: In the S4, the step of feature signal processing comprises: S401: performing FRFT transform on the X and Y component track feature signals of each suspected plot after updating and screening respectively; S402: target detection is performed in the FRFT transform space with a fixed threshold, and peak parameters are extracted; S403: in the FRFT domain, the spectral values below the threshold are directly processed by zero, to realize filtering, and then inverse transform is applied to realize track recovery.

7. The utility radar target long-time multi-frame TBD detection algorithm according to claim 6, characterized in that: In the S401, the formula of the FRFT transform is: In the formula, a is the transform order, K α (t, u) is the transform kernel, x(t) is the function to be transformed, and X α (u) is the function after transformation.

8. The utility radar target long-time multi-frame TBD detection algorithm according to claim 6, characterized in that: In the S402, the construction formula of the fixed threshold is Wherein, γ2 is a target confirmation threshold set according to the false alarm probability, which can be set as γ2=η2N, wherein η2 can be set to 0.5-0.

8.

9. The utility radar target long-time multi-frame TBD detection algorithm according to claim 6, characterized in that: In the S403, the formula of the FRFT inverse transform is:

Images