Sparse stepped frequency radar range spectrum layered virtual estimation method, device and storage medium
Through the virtual estimation method of sparse stepping frequency radar distance spectrum hierarchical virtual estimation method, the false alarm and missing alarm problems caused by target distance coupling in multi-target scenarios are solved, and efficient HRRP detection performance improvement and power consumption saving are achieved.
Patent Information
- Application Number
- CN202211724965.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-30
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-12-30
AI Technical Summary
In multi-target scenarios, sparse step frequency radar has serious false alarm or missed alarm problems due to target distance coupling, and the existing compression perception solution method has large calculation volume and is prone to divergence, and there is significant virtual error in the virtual array method. , affects the accuracy of HRRP.
The virtual estimation method of sparse stepping frequency radar distance spectrum is used to detect the target scene by frame units, the Doppler layer is split and the high-resolution distance spectrum is calculated separately, combining differential arrays and four The order accumulation theory is used to solve the frequency point, and the frequency point is selected using the minimum redundant line array, and energy normalization is performed on each Doppler layer to alleviate the change in the callback energy.
In the undersampling state, reduce the mutual coupling between multiple targets, improve HRRP detection performance, save radar power consumption, and significantly reduce target false alarm and missing alarm rates.
Smart Images

Figure CN118276014B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of radar detection, and in particular to a sparse stepped frequency radar range spectrum hierarchical virtual estimation method, device and storage medium. Background Art
[0002] The stepped frequency continuous wave (SFCW) radar system provides a high resolution range profile (HRRP) by means of bandwidth synthesis. Specifically, SFCW selects N frequency points at equal intervals in the working frequency band, detects all the frequency points one by one, obtains a sampling point for each frequency point, and the sampling points obtained from all the frequency points form a frequency domain sampling sequence. The frequency domain sampling sequence is subjected to a fast Fourier transform (FFT) to obtain the HRRP. Because SFCW needs to switch the frequency points multiple times, each switch requires several microseconds to stabilize the phase-locked loop, which makes the SFCW detection efficiency low. The sparse stepped frequency radar (SSF) system can use the detection results of some frequency points to calculate the HRRP of the same resolution, which improves the detection efficiency and is very valuable in the market.
[0003] The commonly used compressed sensing solution method in the current SSF radar HRRP calculation method not only has a large amount of calculation and involves complex exponential and division operations, but is also very unfriendly to low-cost and low-power products that require pure hardware implementation. In addition, the convergence process is prone to divergence, resulting in the difficulty of achieving satisfactory detection performance. Another commonly used method is to use the virtual array method in array signal processing theory to virtualize the remaining frequency points, thereby forming a uniformly spaced frequency domain sampling sequence, and performing FFT on this sequence to obtain HRRP. However, this method has significant virtual errors when using fewer frequency points to virtualize more frequency points. These errors worsen the sidelobes in HRRP, thereby causing mutual coupling between targets at different distances, and ultimately resulting in serious target false alarms or missed alarms in the detection results, limiting the performance of the SSF radar system in multi-target scenarios. Summary of the invention
[0004] The purpose of the present invention is to solve the problem that since the sparse stepped frequency radar perceives the target distance by estimating the high-resolution range spectrum, the problem of multi-target coupling will interfere with the subsequent solution operation, resulting in serious target false alarm or missed alarm in the detection process.
[0005] In view of at least one of the above technical problems, the present invention provides a sparse stepped frequency radar range spectrum hierarchical virtual estimation method, comprising:
[0006] The sparse stepped frequency radar detects the target scene in frames, and each detection generates a unit frame;
[0007] In the unit frame, all frequency points are measured one by one and a set of measurement data is obtained, and a set of measurement data is regarded as a snapshot; the set of all detected snapshots constitutes a snapshot sequence;
[0008] Using a filter bank, splitting the snapshot sequence into Doppler layers;
[0009] Each layer of the Doppler layer is individually virtualized and a high-resolution range spectrum is calculated;
[0010] In the high-resolution range spectrum, the target is detected by range spectrum, or all high-resolution range spectra are combined into a Doppler range two-dimensional spectrum and then the target is detected by a two-dimensional detection method.
[0011] Furthermore, the step of measuring all the frequency points one by one includes: selecting a frequency point in the unit frame by using a minimum redundant line array in the unit frame, and measuring the selected frequency point.
[0012] Furthermore, the detection frame rate of the sparse stepped frequency radar is less than the maximum Doppler frequency in the target scene.
[0013] Furthermore, the step of individually virtualizing and calculating a high-resolution range spectrum for each layer of the Doppler layer includes: using a differential array and fourth-order accumulation theory to calculate a virtual array, performing a solution operation on the virtual array to obtain a high-resolution range spectrum; the solution operation includes selecting a continuous part of the virtual array for Fourier transform to obtain a result sequence, and using the phase of the result sequence to correct the target distance.
[0014] Furthermore, before performing the solving operation, energy normalization is performed on each snapshot of the Doppler layer. The energy normalization is to alleviate the change of the callback energy with distance.
[0015] Furthermore, the filter bank comprises a high-pass filter cascaded sub-band analysis filter bank; the cut-off frequency of the high-pass filter is less than the sub-band bandwidth of the sub-band analysis filter. The form of the high-pass filter includes Wherein α∈(0,1); the filtering operation of the subband analysis filter bank is implemented by sliding discrete Fourier transform.
[0016] Furthermore, the widths of all sub-bands of the sub-band analysis filter bank are positively correlated with the absolute values of the sub-band center frequencies.
[0017] Furthermore, in the step of detecting the target scene, the frame repetition period is unequally spaced.
[0018] On the other hand, the present invention further provides a stepped frequency radar device, characterized in that the device is used to execute any one of the sparse stepped frequency radar range spectrum layered virtual estimation methods described above.
[0019] On the other hand, the present invention also provides a computer-readable storage medium, which stores a program executable by a stepped frequency radar device, characterized in that the executable program is used to execute any one of the sparse stepped frequency radar range spectrum layered virtual estimation methods when executed by the stepped frequency radar device.
[0020] The beneficial effects of the present invention are as follows: the SSF radar to which the present invention is applicable mainly works in an undersampling state, and the typical frame rate in this state is reduced to about 100 times lower than the Nyquist sampling rate, so as to save the radar power-on duty cycle, thereby saving radar power consumption. The mutual coupling between multiple targets is reduced by Doppler stratification, thereby improving the detection performance of subsequent HRRP. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 A schematic diagram of the position of the detection target in an embodiment of the present invention;
[0022] Figure 2 It is a schematic diagram of the filter bank in an embodiment of the present invention for layering echoes of different rates. DETAILED DESCRIPTION
[0023] In order to describe the technical solution and advantages of the relational data table organization structure more completely and clearly, the present invention will be analyzed and explained in conjunction with examples and drawings. Obviously, this embodiment is only used to explain the present invention and is not used to limit the present invention.
[0024] The following are explanations of some abbreviated terms in the embodiments of the present invention:
[0025] SSF: sparse stepped frequency, sparse stepped frequency radar;
[0026] HRRP: high resolution range profile, high resolution range profile.
[0027] The commonly used compressed sensing solution method in the current SSF radar HRRP calculation method not only has a large amount of calculation and involves complex exponential and division operations, but is also very unfriendly to low-cost, low-power products that need to be implemented by pure hardware. In addition, the convergence process is prone to divergence, and the detection performance is difficult to achieve a satisfactory level. Another commonly used method is to use the virtual array method in array signal processing theory to virtualize the remaining frequency points, thereby forming a uniformly spaced frequency domain sampling sequence, and performing FFT on this sequence to obtain HRRP. However, this method has significant virtual errors when using fewer frequency points to virtualize more frequency points. These errors worsen the sidelobes in HRRP, thereby causing mutual coupling between targets at different distances, and ultimately leading to serious target false alarms or missed alarms in the detection results, limiting the performance of the SSF radar system in multi-target scenarios.
[0028] In the current Doppler radar, the operation of obtaining the velocity spectrum through Doppler filtering and the layered operation in the embodiment of the present invention seem to be very similar, but there are inherent essential differences. The purpose of the current Doppler radar performing Doppler filtering is to accurately measure the speeds of different echo components to obtain an accurate velocity spectrum, so the frame rate is required to be no lower than the Nyquist sampling rate, which is usually on the order of kilohertz. The SSF radar to which the embodiment of the present invention is applicable mainly works in an under-sampling state, and the typical frame rate in this state is reduced to about 100 times lower than the Nyquist sampling rate, thereby saving the radar power-on duty cycle and thus saving the radar power consumption.
[0029] The following is a detection process in an embodiment of the present invention:
[0030] After the SSF radar is turned on, it detects the environment in frames. Each unit frame contains the measurement of all L sparse frequency points, which is called a snapshot. The snapshot data obtained in the nth frame is recorded in the form of a column vector:
[0031]
[0032] In formula (1), d represents the element in the snapshot data, and L represents the total number of sparse frequency points.
[0033] Construct subband analysis filters, whose subband filter transfer functions are H (0) , H (1) …H (M-1) , where M represents the number of subbands, and H (0) +H (1) +…+H (M-1) =1 holds. Using the subband analysis filter to perform slow-time filtering on the snapshot data, layered processing can be achieved to obtain a total of M Doppler layers. At this time, the snapshot data of the mth Doppler layer obtained in the nth frame can be expressed as:
[0034]
[0035] In formula (2), F m {*} indicates that the input sequence is filtered using the mth subband filter in the subband analysis filter bank.
[0036] The virtual operation processing process used in the embodiment of the present invention includes:
[0037] First calculate The projection matrix can be calculated as:
[0038]
[0039] In formula (3), (*) H It means to find the Hermite transpose;
[0040] Then calculate the covariance matrix of the mth Doppler layer output of the nth frame, and its calculation method can be expressed as:
[0041]
[0042] In formula (4), LP{*} represents the slow time low-pass filtering operation;
[0043] Then The vectors obtained after vectorization are sorted and redundant operations are deleted to obtain a virtual frequency domain sampling sequence. Then, the FFT operation is performed on this sequence to obtain the HRRP of the mth Doppler layer of the nth frame. Next, the HRRP of the remaining Doppler layers in the nth frame is calculated in the same way, and then the target can be detected one by one in all M HRRPs.
[0044] SSF radar uses a frequency-by-frequency measurement method to synthesize a broadband waveform and then obtain HRRP. The RF circuit architecture is extremely simplified and the aliasing distance is long, but it has the disadvantages of low measurement efficiency and serious mutual coupling between targets. The embodiment of the present invention proposes a method of first performing Doppler dimension layered processing on the echo signal, and then virtualizing the frequency point layer by layer and estimating the HRRP. In order to describe the technical solution in the embodiment of the present invention in detail, the echo sampling expression under the transmission of the lth frequency point is first given as follows:
[0045]
[0046] In formula (5), E represents the number of targets in the current environment, and T represents the inter-frame interval, i.e., the inverse of the frame rate;
[0047] s e,l The echo expression of target e is as follows:
[0048]
[0049] In formula (6), C represents the speed of light, r e represents the distance of the e-th target, f l Indicates the frequency of the lth frequency point;
[0050] Assume that the target e moves in space at a speed v e Doing uniform linear motion, its position at time 0 is Figure 1 As shown,
[0051]
[0052] In formula (7), b e represents the vertical distance from the e-th target to the radar along the moving direction, r 0,e represents the initial distance from the e-th target to the radar, O(t) represents the acceleration and its higher-order terms, v r,e represents the instantaneous radial velocity at time 0;
[0053] Substituting the above formula into the echo sampling expression, we can get
[0054]
[0055] In equation (8), the second term on the right side of the equation represents the Doppler frequency, and the third term represents the Doppler modulation term, which means that the echo of target e can be regarded as a modulated signal with a center frequency equal to the Doppler frequency. Therefore, the filter bank has an effect on d l Sequence filtering can separate the echoes of different rates into layers, thus alleviating the pressure of multi-target mutual coupling on the post-processing. Figure 2 As shown. The layered echo sequence presents a narrowband passband characteristic, so it can be shifted to zero frequency and then extracted to reduce the calculation pressure of subsequent solution, where the frequency shift amount is equal to the sub-band center frequency. It is worth noting that the above expression shows that the echo sampling result of the static reflector is a constant, which means that the echo of the static reflector can be easily suppressed by a high-pass filter. The common high-pass filter form is:
[0056]
[0057] Where α∈(0,1).
[0058] If the frame rate is too low and the bandwidth of the Doppler modulation term is too large, d lThe sequence has serious aliasing. For example, when observing an object moving radially with a speed of 3 meters per second and an acceleration of 0.1 meters per square second at a frequency of 79 GHz and a frame rate of 20 Hz, the bandwidth of the selected sub-band filter is 1 Hz. At this time, although the echo energy of this target is dispersed in several sub-bands, there will still be some sub-bands that almost do not contain the echo energy of this target. The present invention can still achieve the purpose of alleviating the pressure of multi-target mutual coupling on the post-stage solution processing, but the effect of alleviating is reduced.
[0059] For each Doppler layer, the expected number of targets will be significantly reduced compared to before stratification. At this time, HRRP is calculated for each layer separately. The simplest method is to express the snapshot data of the mth Doppler layer of the nth frame as Then the projection matrix expression is:
[0060]
[0061] In formula (9), the denominator on the right is the energy normalization implemented to alleviate the change of echo energy with distance, and it is filtered to obtain the covariance matrix:
[0062]
[0063] In the embodiment of the present invention, the commonly used filter is Where α∈(0,1), then all elements of the covariance matrix are rearranged and merged according to the frequency values of the sparse frequency points. The most common averaging method can be used to merge. At this time, a virtual vector is obtained. The continuous part of this vector is intercepted for Fourier transform, and the phase of the transformed result sequence is multiplied by The result of is used as the correction term of the distance to correct the sequence, and the HRRP of the mth Doppler layer of the nth frame can be obtained. Finally, the HRRP is used for target detection. The minimum redundancy design is used for sparse frequency points to ensure that the length of the continuous part in the virtual vector is maximized. S represents the length of the continuous part in the virtual vector, and F represents the number of Fourier transform points.
[0064] in addition, or This is also possible and will not be further elaborated in the embodiments of the present invention, where Qcum(*) represents obtaining a 4th-order or any higher-order cumulative amount.
[0065] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A sparse stepped frequency radar range spectrum hierarchical virtual estimation method, characterized in that: include: The sparse stepped frequency radar detects the target scene in frames, and each detection generates a unit frame; In the unit frame, all frequency points are measured one by one and a set of measurement data is obtained, and a set of measurement data is regarded as a snapshot; the set of all detected snapshots constitutes a snapshot sequence; Using a filter bank, splitting the snapshot sequence into Doppler layers; Each layer of the Doppler layer is individually virtualized and a high-resolution range spectrum is calculated; In the high-resolution range spectrum, the target is detected by range spectrum, or all high-resolution range spectra are combined into a Doppler range two-dimensional spectrum and then the target is detected by a two-dimensional detection method.
2. The method for layered virtual estimation of sparse stepped frequency radar range spectrum according to claim 1, characterized in that: The step of measuring all frequency points one by one includes: selecting a frequency point in a unit frame through a minimum redundant line array, and measuring the selected frequency point.
3. The method for layered virtual estimation of sparse stepped frequency radar range spectrum according to claim 1, characterized in that: The detection frame rate of the sparse stepped frequency radar is less than the maximum Doppler frequency in the target scene.
4. The method for layered virtual estimation of sparse stepped frequency radar range spectrum according to claim 1, characterized in that: The step of individually virtualizing and calculating a high-resolution range spectrum for each layer of the Doppler layer includes: using a differential array and fourth-order accumulation theory to calculate a virtual array, performing a solution operation on the virtual array to obtain a high-resolution range spectrum; the solution operation includes selecting a continuous part of the virtual array for Fourier transform to obtain a result sequence, and using the phase of the result sequence to correct the target distance.
5. The method for layered virtual estimation of sparse stepped frequency radar range spectrum according to claim 4, characterized in that: Before performing the solving operation, energy normalization is performed on each snapshot of the Doppler layer.
6. The sparse stepped frequency radar range spectrum hierarchical virtual estimation method according to claim 1, characterized in that: The filter bank comprises a high-pass filter cascaded with a sub-band analysis filter bank; the cut-off frequency of the high-pass filter is smaller than the sub-band bandwidth of the sub-band analysis filter.
7. The method for layered virtual estimation of sparse stepped frequency radar range spectrum according to claim 6, characterized in that: The high-pass filter is of the form comprising Wherein α∈(0,1); the filtering operation of the subband analysis filter bank is implemented by sliding discrete Fourier transform.
8. The method for layered virtual estimation of sparse stepped frequency radar range spectrum according to claim 6, characterized in that: The widths of all sub-bands of the sub-band analysis filter bank are positively correlated with the absolute values of the sub-band center frequencies.
9. The method for layered virtual estimation of sparse stepped frequency radar range spectrum according to claim 1, characterized in that: In the step of detecting the target scene, the repetition period of the unit frame is unequally spaced.
10. A stepped frequency radar device, characterized in that: The device is used to execute the sparse stepped frequency radar range spectrum hierarchical virtual estimation method described in any one of claims 1-9.
11. A computer-readable storage medium storing a program executable by a stepped frequency radar device, characterized in that: The executable program is used to execute the sparse stepped frequency radar range spectrum layered virtual estimation method according to any one of claims 1 to 9 when executed by the stepped frequency radar device.
