Method and System for Dynamic Separation and Anti-Jamming of Multiple Targets in Frequency-Modulated Continuous-Wave Radar Based on Synchronous Extraction Transformation

Abstract

The present invention discloses a method and system for multi-target dynamic separation and anti-interference of a frequency-modulated continuous-wave radar based on synchronous extraction transformation, including: S1. Generating a linearly frequency-modulated continuous wave and transmitting a high-frequency signal through a radio frequency front end; S2. Receiving the target reflection signal, mixing it with the transmitted signal through a mixer, extracting the intermediate-frequency beat signal and denoising it; S3. Performing time-frequency decomposition on the signal using synchronous extraction transformation to generate a high-resolution time-frequency distribution matrix; S4. Separating overlapping targets based on the time-frequency matrix using a density clustering algorithm, combining a dynamic decoupling compensation algorithm to eliminate range-velocity coupling errors, and outputting the time-frequency trajectories of independent targets; constructing a dynamic time-frequency domain mask filter to filter out noise, optimizing the detection sensitivity through an adaptive threshold detection mechanism and performing false alarm control; The present invention accurately separates dense targets through high-resolution time-frequency analysis, and combines a dynamic Doppler decoupling algorithm to improve the anti-interference ability and the dynamic target tracking accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of radar signal processing, and more specifically, to a method and system for dynamic separation and anti-interference of multiple targets of a frequency-modulated continuous-wave radar based on synchronous extraction transformation. Background Art

[0002] A frequency-modulated continuous-wave (FMCW) radar realizes target ranging and velocity measurement by transmitting a linearly frequency-modulated continuous-wave signal and analyzing the echo beat frequency, and has advantages such as low power consumption and high resolution.

[0003] Currently, in the existing FMCW radar technology, the spectrum analysis method based on the fast Fourier transform (FFT) is limited by the spectrum resolution and the Rayleigh criterion, and spectral aliasing is likely to occur in a dense multi-target scenario, resulting in difficulty in removing false targets. At the same time, the traditional fixed-threshold detection method has insufficient anti-interference ability in a complex electromagnetic environment and is difficult to distinguish dynamic targets from noise and clutter.

[0004] In addition, the Doppler frequency shift of a moving target will introduce a range-velocity coupling error. The existing decoupling methods rely on multiple scans or complex waveform designs, have poor real-time performance, and high hardware costs. For the tracking of high-speed dynamic targets, conventional static spectrum analysis is difficult to capture time-varying Doppler characteristics, resulting in broken target trajectories or missed detections.

[0005] The above problems seriously restrict the reliability and accuracy of FMCW radars in scenarios such as intelligent driving and security monitoring.

[0006] Therefore, how to solve the problems of spectral aliasing, insufficient anti-interference ability, and low tracking accuracy of dynamic targets of existing FMCW radars in a dense multi-target scenario is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0007] In view of this, the present invention provides a method and system for dynamic separation and anti-interference of multiple targets of a frequency-modulated continuous-wave radar based on synchronous extraction transformation to solve some of the technical problems mentioned in the background art.

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] A method for dynamic separation and anti-interference of multiple targets of a frequency-modulated continuous-wave radar based on synchronous extraction transformation, comprising the following steps:

[0010] S1. Signal transmission: Generate a linearly frequency-modulated continuous wave and transmit a high-frequency signal through a radio frequency front end;

[0011] S2. Signal reception and preprocessing: Receive the target reflection signal, mix it with the transmitted signal through a mixer, extract the intermediate-frequency beat signal and perform denoising processing;

[0012] S3. Synchronous extraction transform time-frequency analysis: The signal is subjected to time-frequency decomposition using the synchronous extraction transform to generate a high-resolution instantaneous frequency-time distribution matrix;

[0013] S4. Dynamic target separation and decoupling: Based on the time-frequency matrix, the density clustering algorithm is used to separate overlapping targets, and the dynamic decoupling compensation algorithm is combined to eliminate the range-velocity coupling error, and the time-frequency trajectories of independent targets are output;

[0014] S5. Anti-interference processing: A dynamic time-frequency domain mask filter is constructed to filter out noise, and the detection sensitivity is optimized through an adaptive threshold detection mechanism and false alarm control is performed.

[0015] Preferably, in step S1, the parameters of the transmitted signal are dynamically adjusted according to the scenario to ensure coverage of the detection range.

[0016] Preferably, in step S2, the denoising process is to perform wavelet threshold filtering preprocessing and signal normalization on the intermediate frequency beat signal.

[0017] Preferably, the specific content of step S3 is:

[0018] S31. The intermediate frequency beat signal after denoising is subjected to short-time Fourier transform STFT to generate a time-frequency matrix;

[0019] S32. Based on the SET transform, the instantaneous frequency ridge line is extracted, and multi-candidate ridge line screening and optimization are performed through ridge line continuity detection;

[0020] S33. A SET matrix is constructed to obtain a high-resolution instantaneous frequency-time distribution matrix.

[0021] Preferably, in step S32, the method for multi-candidate ridge line screening and optimization through ridge line continuity detection is:

[0022]

[0023] Among them, F candidate is the candidate frequency, that is, the potential significant frequency component in the detected signal, f is the frequency, |STFT(t,f)| is the amplitude of the short-time Fourier transform at time t and frequency f, ∂|STFT(t,f)| is the partial derivative of the amplitude spectrum with respect to frequency f, describing the rate of change of the amplitude spectrum in the frequency direction, used to screen out the points where the derivative of the amplitude spectrum on the frequency axis is zero;

[0024] The path continuity constraint is:

[0025]

[0026]

[0027]

[0028] Among them, f inst (t) is the extracted instantaneous frequency, Δt is the time resolution, Δf is the frequency resolution, and N FFT is the number of FFT points.

[0029] Preferably, the specific content of step S4 is as follows:

[0030] S41. Based on the time-frequency energy distribution of the time-frequency matrix generated by the synchronous extraction transformation, an improved density clustering algorithm is used to form independent target clusters by screening time-frequency points that meet the conditions through clustering parameters. The Mahalanobis distance criterion is introduced for inter-cluster conflict processing of overlapping clusters to separate the time-frequency energy ridge line and identify the time-frequency ridge line characteristics of independent targets;

[0031] S42. Perform multi-target cross-judgment. If there is multi-target crossing, perform dynamic decoupling and update the SET matrix;

[0032] S43. For the range-velocity coupling error caused by moving targets, use the Doppler frequency shift model real-time compensation algorithm to correct the range-velocity coupling error by calculating the target instantaneous frequency change rate in real time;

[0033] S44. Output the time-frequency trajectory of independent targets.

[0034] Preferably, in step S41, the Mahalanobis distance criterion introduced for overlapping clusters is:

[0035]

[0036] Among them, X i =[t i , f i T is the time-frequency point coordinate, µ j , Σ j are the mean and covariance matrix of the jth cluster respectively;

[0037] In step S42, the Doppler frequency shift model is:

[0038]

[0039] λ = c / f0

[0040] Among them, c = 3×10 8 m / s, f0 is the starting frequency, Δf d (t) is the target instantaneous frequency change rate, and v(t) is the target moving speed.

[0041] Preferably, the specific content of step S5 is as follows:

[0042] ​S51. Based on a preset interference feature library and real-time time-frequency energy distribution, construct a dynamic time-frequency mask filter to filter out noise;

[0043] S52. For the energy distribution characteristics of the time-frequency matrix, adopt statistical-driven dynamic threshold adjustment to calculate the detection threshold and optimize energy aggregation.

[0044] Preferably, in step S51, the dynamic time-frequency mask filter is specifically:

[0045]

[0046] Among them, R k is the interference area, which is identified through multi-level matching, including fixed clutter suppression and multipath interference detection;

[0047] In step S52, the calculated detection threshold is:

[0048] T(t)=µ(t)+k·σ(t)

[0049] Among them, µ(t) is the mean time-frequency energy within the sliding window, that is, the window length, σ(t) is the standard deviation, and k is the frequency modulation slope, which is dynamically adjusted according to the signal-to-noise ratio;

[0050] The energy aggregation optimization is to integrate the energy of weak target signals across multiple time-frequency units through morphological dilation operation:

[0051]

[0052] Among them, se is the structural element, se = 3×3, E enhanced (t,f) is the processed time-frequency energy, and E(t,f) is the original time-frequency energy or signal.

[0053] A frequency-modulated continuous wave radar multi-target dynamic separation and anti-interference system based on synchronous extraction transformation, based on the described frequency-modulated continuous wave radar multi-target dynamic separation and anti-interference method, includes: a signal transmission module, a signal reception and preprocessing module, a synchronous extraction transformation time-frequency analysis module, a dynamic target separation and decoupling module, and an anti-interference processing module;

[0054] The signal transmission module is used to generate a linearly frequency-modulated continuous wave and transmit a high-frequency signal through the RF front end;

[0055] The signal reception and preprocessing module is used to receive the target reflected signal, mix it with the transmitted signal through a mixer, extract the intermediate frequency beat signal and perform denoising processing;

[0056] The synchronous extraction transformation time-frequency analysis module is used to perform time-frequency decomposition on the signal by using synchronous extraction transformation to generate a high-resolution instantaneous frequency-time distribution matrix;

[0057] A dynamic target separation and decoupling module, which is used to separate overlapping targets based on the time-frequency matrix using a density clustering algorithm, combine a dynamic decoupling compensation algorithm to eliminate range-velocity coupling errors, and output the time-frequency trajectories of independent targets.

[0058] An anti-interference processing module, which is used to construct a dynamic time-frequency domain mask filter to filter out noise, optimize the detection sensitivity through an adaptive threshold detection mechanism, and perform false alarm control.

[0059] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a frequency-modulated continuous wave radar signal processing method and system based on synchronous extraction transformation, which are mainly applied to the fields of intelligent driving, UAV navigation, security monitoring, and industrial automation, and solve the problems of spectrum aliasing, insufficient anti-interference ability, and low dynamic target tracking accuracy of existing FMCW radars in dense multi-target scenarios.

[0060] Through high-resolution time-frequency analysis, the present invention accurately separates dense targets such as pedestrians and vehicles around the vehicle, and suppresses noise and clutter interference in a complex electromagnetic environment; combines a dynamic Doppler decoupling algorithm to eliminate range-velocity coupling errors of moving targets, and realizes real-time and accurate ranging for high-speed UAV obstacle avoidance or close-range operation of a robotic arm; at the same time, reduces the dependence on high-linearity hardware devices, adapts to low-cost radar modules, and significantly improves the environmental perception reliability and system economy in scenarios such as autonomous driving and industrial robots. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0062] Figure 1 It is a schematic diagram of a multi-target dynamic separation and anti-interference method for a frequency-modulated continuous wave radar based on synchronous extraction transformation provided by the present invention;

[0063] Figure 2 It is a flow chart of synchronous extraction transformation time-frequency analysis and dynamic target separation and decoupling provided by the present invention;

[0064] Figure 3 It is a schematic diagram of the effect of synchronous extraction transformation for processing FMCW signals provided by the present invention;

[0065] Figure 4 It is a schematic diagram of the effect of time-frequency signal clustering analysis provided by the present invention;

[0066] Figure 5 Schematic diagram of the FPGA hardware real-time processing framework provided by the present invention. Specific embodiments

[0067] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0068] An embodiment of the present invention discloses a method for multi-target dynamic separation and anti-interference of a frequency-modulated continuous-wave radar based on synchronous extraction transformation, as Figure 1 follows:

[0069] S1. Signal transmission: Generate a linearly frequency-modulated continuous wave and transmit a high-frequency signal through the RF front end;

[0070] S2. Signal reception and preprocessing: Receive the target reflected signal, mix it with the transmitted signal through a mixer, extract the intermediate-frequency beat signal and perform denoising processing;

[0071] S3. Synchronous extraction transformation time-frequency analysis: Perform time-frequency decomposition on the signal using synchronous extraction transformation to generate a high-resolution instantaneous frequency-time distribution matrix;

[0072] S4. Dynamic target separation and decoupling: Based on the time-frequency matrix, use the density clustering algorithm to separate overlapping targets, and combine the dynamic decoupling compensation algorithm to eliminate the range-velocity coupling error, and output the time-frequency trajectories of independent targets;

[0073] S5. Anti-interference processing: Construct a dynamic time-frequency domain mask filter to filter out noise, optimize the detection sensitivity through an adaptive threshold detection mechanism and perform false alarm control.

[0074] To further implement the above technical solution, in step S1, the parameters of the transmitted signal are dynamically adjusted according to the scenario to ensure coverage of the detection range.

[0075] The FMCW transmitted signal is:

[0076]

[0077] where A is the signal amplitude (typical value: A = 1V, determined by the power amplifier gain), f0 is the starting frequency (commonly used f0 = 77GHz for vehicle-mounted radars), B is the frequency modulation bandwidth (range: B = 1 - 4GHz, determining the range resolution ), and T is the frequency modulation period (typical value: T = 50µs, determining the maximum detection range ), where t is the time variable (t ∈ [0, T]).

[0078] The received signal is:

[0079]

[0080] Among them, A r is the received signal amplitude (related to the target RCS, typical value: A r = 10 mV), is the round-trip time delay of the signal (R is the target distance, c = 3×10 8 m / s), and n(t) is the additive noise (including thermal noise, multipath interference, etc., typical SNR value: -5~10 dB).

[0081] To further implement the above technical solution, in step S2, the denoising process is to perform wavelet threshold filtering preprocessing and signal normalization on the intermediate-frequency beat signal.

[0082] The intermediate-frequency beat signal is expressed as:

[0083]

[0084] The beat frequency is:

[0085]

[0086] Among them, v is the target radial velocity; the parameter range of the short-time Fourier transform is: the frequency f corresponding to the distance b,R ∈ [0.1 MHz, 5 MHz] (corresponding to R = 0.1~150 m); the frequency f corresponding to the velocity b,v ∈ [-100 MHz, 100 MHz] (corresponding to v = -200~200 m);

[0087] The wavelet threshold filtering can be expressed as:

[0088]

[0089] Among them, the threshold calculation is:

[0090]

[0091] Among them, σ is the noise standard deviation (estimated through the leading silent segment of the signal), N is the number of signal sampling points (typical value: N = 1024), and the wavelet basis Symlets 8 is selected, taking into account both time-frequency locality and computational efficiency;

[0092] The amplitude normalization formula:

[0093]

[0094] Amplitude normalization can eliminate the influence of signal amplitude fluctuations on subsequent time-frequency analysis.

[0095] To further implement the above technical solution, as Figure 2 , the specific content of step S3 is:

[0096] S31. Perform short-time Fourier transform (STFT) on the intermediate frequency beat signal after denoising to generate a time-frequency matrix;

[0097] Short-time Fourier transform (STFT):

[0098]

[0099] Among them, the window function w(t) uses a Gaussian window w(t)=e -π(t / σ)^2 , σ = 16 samples;

[0100] In this embodiment, the number of FFT points: N FFT = 512 (frequency resolution , is 24.4 kHz), and the overlap rate is 75% (time resolution );

[0101] Instantaneous frequency extraction:

[0102] f inst (t)=argmax f |STFT(t,f)|

[0103] S32. Extract the instantaneous frequency ridge line based on the SET transform, and perform multi-candidate ridge line screening and optimization through ridge line continuity detection;

[0104] S33. Construct a SET matrix to obtain a high-resolution instantaneous frequency-time distribution matrix;

[0105] SET(t,f)=STFT(t,f)·δ(f - f inst (t)).

[0106] In this embodiment, the resolution improvement multiple is:

[0107]

[0108] Taking the FMCW simulation signal as an example, as Figure 3 shown, it can be seen from the figure that SET effectively suppresses the energy diffusion phenomenon of STFT, and the time-frequency resolution of the two frequency modulation trajectories is significantly improved, verifying the advantage of synchronous extraction transform time-frequency analysis in target separation.

[0109] To further implement the above technical solution, the method for multi-candidate ridge line screening and optimization through ridge line continuity detection in step S32 is:

[0110]

[0111] Among them, F candidate is the candidate frequency, i.e., the potential significant frequency component in the detection signal, f is the frequency, |STFT(t, f)| is the amplitude of the short-time Fourier transform at time t and frequency f, and ∂|STFT(t, f)| is the partial derivative of the amplitude spectrum with respect to the frequency f, describing the rate of change of the amplitude spectrum in the frequency direction. It is used to screen out the points where the derivative of the amplitude spectrum with respect to the frequency axis is zero.

[0112] The path continuity constraint is:

[0113]

[0114]

[0115]

[0116] Among them, f inst (t) is the extracted instantaneous frequency, Δt is the time resolution, Δf is the frequency resolution, and N FFT is the number of FFT points.

[0117] To further implement the above technical solution, as Figure 2 , the specific content of step S4 is:

[0118] S41. Based on the time-frequency energy distribution of the time-frequency matrix generated by the synchronous extraction transform, an improved density clustering algorithm is adopted. Through the clustering parameters, the time-frequency points that meet the conditions are screened to form independent target clusters. For overlapping clusters, the Mahalanobis distance criterion is introduced to handle the inter-cluster conflict, so as to separate the time-frequency energy ridge line and identify the time-frequency ridge line characteristics of independent targets.

[0119] In this embodiment, the clustering parameters are specifically the neighborhood radius ϵ = 0.1f max and the minimum number of samples min_samples = 5.

[0120] S42. Perform multi-target cross-judgment. If there is multi-target crossing, perform dynamic decoupling and update the SET matrix.

[0121] S43. For the range-velocity coupling error caused by moving targets, use the Doppler frequency shift model real-time compensation algorithm. By calculating the real-time change rate of the target instantaneous frequency, correct the range-velocity coupling error.

[0122] S44. Output the time-frequency trajectory of the independent target.

[0123] To further implement the above technical solution, in step S41, the Mahalanobis distance criterion introduced for the overlapping clusters is as follows:

[0124]

[0125] where X i =[t i ,f i T is the time-frequency point coordinate, and μ j , Σ j are respectively the mean and covariance matrix of the j-th cluster;

[0126] In step S42, the Doppler frequency shift model is as follows:

[0127]

[0128] λ = c / f0

[0129] where c = 3×10 8 m / s, f0 is the starting frequency, Δf d (t) is the target instantaneous frequency change rate, and v(t) is the target motion speed.

[0130] In this embodiment, the clustering analysis result is as Figure 4 shown. It can be seen that the density clustering algorithm DBSCAN effectively separates the time-frequency points of the crossing trajectories (red and blue clusters), and at the same time correctly identifies the noise points (black). This result verifies the effectiveness of the density clustering algorithm in multi-target separation.

[0131] To further implement the above technical solution, the specific content of step S5 is as follows:

[0132] S51. Based on the preset interference feature library and the real-time time-frequency energy distribution, construct a dynamic time-frequency mask filter to filter out noise;

[0133] S52. For the time-frequency matrix energy distribution characteristics, adopt statistical-driven dynamic threshold adjustment to calculate the detection threshold and optimize the energy aggregation.

[0134] To further implement the above technical solution, in step S51, the dynamic time-frequency mask filter is specifically as follows:

[0135]

[0136] where R k is the interference area, which is identified through multi-level matching, including fixed clutter suppression and multipath interference detection;

[0137] ​In this embodiment, fixed clutter suppression specifically refers to filtering out low-frequency static reflection components with |f| < 10 Hz; multipath interference detection is to identify abnormal energy clusters with |τ - 2R direct / c| > 1 μs in the time-delay Doppler domain;

[0138] In step S52, the detection threshold is calculated as:

[0139] T(t) = μ(t) + k · σ(t)

[0140] where μ(t) is the mean time-frequency energy within the sliding window, i.e., the window length, σ(t) is the standard deviation, k is the frequency modulation slope, which is dynamically adjusted according to the signal-to-noise ratio, and k ∈ [2, 5];

[0141] Energy aggregation optimization is to integrate the energy of weak target signals across multiple time-frequency units through morphological dilation operations:

[0142]

[0143] where se is the structuring element, se = 3 × 3, E enhanced (t, f) is the processed time-frequency energy, and E(t, f) is the original time-frequency energy or signal;

[0144] In this embodiment, performance verification was carried out. The actual measurement shows that when the anti-interference processing is at a signal-to-noise ratio SNR = 0 dB, the false alarm rate is reduced to 10 -4 / frame, while ensuring that more than 95% of the weak targets (with energy 6 dB below the noise floor) can be effectively detected, and the interference suppression ratio (ISR) reaches 28 dB. When implemented in hardware, the energy statistic is calculated in parallel by FPGA, and the single-frame processing delay is controlled within 15 μs.

[0145] A frequency-modulated continuous-wave radar multi-target dynamic separation and anti-interference system based on synchronous extraction transform, based on a frequency-modulated continuous-wave radar multi-target dynamic separation and anti-interference method, includes: a signal transmission module, a signal reception and preprocessing module, a synchronous extraction transform time-frequency analysis module, a dynamic target separation and decoupling module, and an anti-interference processing module;

[0146] The signal transmission module is used to generate a linearly frequency-modulated continuous wave and transmit a high-frequency signal through the RF front-end;

[0147] The signal reception and preprocessing module is used to receive the target reflection signal, mix it with the transmitted signal through a mixer, extract the intermediate-frequency beat signal, and perform denoising processing;

[0148] The synchronous extraction transform time-frequency analysis module is used to perform time-frequency decomposition on the signal by using synchronous extraction transform to generate a high-resolution instantaneous frequency-time distribution matrix;

[0149] A dynamic target separation and decoupling module, which is used to separate overlapping targets by using a density clustering algorithm based on a time-frequency matrix, and combine a dynamic decoupling compensation algorithm to eliminate range-velocity coupling errors, and output the time-frequency trajectories of independent targets;

[0150] An anti-interference processing module, which is used to construct a dynamic time-frequency domain mask filter to filter out noise, and optimize the detection sensitivity and perform false alarm control through an adaptive threshold detection mechanism.

[0151] In this embodiment, the signal transmitting module includes a signal generation circuit, specifically including a direct digital synthesizer DDS and a voltage-controlled oscillator VCO; the number of bits N of the phase accumulator of the direct digital synthesizer is 32, and the output frequency accuracy ; the linearity error of the voltage-controlled oscillator is <0.1%, ensuring the phase continuity of the frequency-modulated signal.

[0152] The signal receiving and preprocessing module includes an analog-to-digital converter ADC and an anti-aliasing filter; the sampling rate f of the analog-to-digital converter s ≥2×(f b,R +f b,v ) = 10.2MHz (the actual configured f s = 12.5MHz), the resolution is 12 bits (dynamic range 72dB), the cut-off frequency f of the anti-aliasing filter c = 6MHz, and the roll-off coefficient is 0.2.

[0153] The synchronous extraction transform time-frequency analysis module has an FPGA parallel computing architecture, specifically including 8 groups of parallel complex multiplier arrays, which are used to calculate the STFT, and the FFT core is configured as a Radix-4 FFT IP core (single calculation delay 5.1µs).

[0154] A frequency-modulated continuous wave radar multi-target dynamic separation and anti-interference system based on synchronous extraction transform uses an FPGA+ARM heterogeneous architecture to implement a real-time processing pipeline to achieve hardware acceleration and result output, such as Figure 5 ;

[0155] When deployed on the FPGA side, computationally intensive modules such as time-frequency analysis, target separation, and anti-interference processing are deployed, and the throughput is improved through parallel design;

[0156] Among them, the time-frequency matrix calculation unit is 8 parallel complex multipliers (operating frequency f text = 200MHz), and the SET calculation is decomposed into 12 levels of pipelines to optimize the pipeline depth (single-cycle delay T stage = 5ns);

[0157] The ARM side runs the Linux system, is responsible for updating the interference rule library and visualizing the results, and interacts with the FPGA through the AXI bus (transmission bandwidth 4GB / s).

[0158] For the strict delay requirements (Δt total <50 ms) in the vehicle-mounted scenario, multi-dimensional optimization is implemented, including:

[0159] Data stream compression: Differential coding compression is adopted for the target list, and the compression rate can reach 65%, reducing the bus transmission pressure;

[0160] ΔR = R i - R i-1

[0161] Δv = v i - v pred

[0162] Dynamic resource allocation: Dynamically adjust the FPGA computing resources according to the number of targets N text ;

[0163] N core = min(8, [N target / 3])

[0164] Finally, a standardized target information frame is output, including data fields:

[0165] Frame = [Header, (R1, v1, SNR1), …, (R N , v N , SNR N ), CRC]

[0166] Among them, the distance R accuracy is 0.1 m, the speed ν accuracy is 0.1 m / s, and SNR is the signal-to-noise ratio (8-bit quantization).

[0167] In this specification, each embodiment is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For the relevant parts, please refer to the description of the method part.

[0168] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-target dynamic separation and anti-interference method for frequency-modulated continuous-wave radar based on synchronous extraction transformation, characterized in that, It includes the following steps: S1. Signal transmission: Generate a chirp continuous wave and transmit a high-frequency signal through the RF front-end; S2. Signal reception and preprocessing: Receive the target reflected signal, mix it with the transmitted signal through a mixer, extract the intermediate-frequency beat signal and perform denoising processing; S3. Synchronous extraction transform time-frequency analysis: Perform time-frequency decomposition on the signal using the synchronous extraction transform to generate a high-resolution instantaneous frequency-time distribution matrix; S4. Dynamic target separation and decoupling: Based on the time-frequency matrix, use the density clustering algorithm to separate overlapping targets, and combine the dynamic decoupling compensation algorithm to eliminate the range-velocity coupling error, and output the time-frequency trajectory of independent targets; S5. Anti-interference processing: Construct a dynamic time-frequency domain mask filter to filter out noise, optimize the detection sensitivity through an adaptive threshold detection mechanism and perform false alarm control; The specific content of step S3 is: S31. Perform short-time Fourier transform (STFT) on the intermediate-frequency beat signal after denoising processing to generate a time-frequency matrix; S32. Extract the instantaneous frequency ridge line based on the SET transform, and perform multi-candidate ridge line screening and optimization through ridge line continuity detection; S33. Construct a SET matrix to obtain a high-resolution instantaneous frequency-time distribution matrix; In step S32, the method for multi-candidate ridge line screening and optimization through ridge line continuity detection is: Among them, F candidate is the candidate frequency, that is, the potential significant frequency component in the detection signal, f is the frequency, and |STFT(t,f)| is the amplitude of the short-time Fourier transform at time t and frequency f. is the partial derivative of the amplitude spectrum with respect to the frequency f, describing the rate of change of the amplitude spectrum in the frequency direction. is used to screen out the points where the derivative of the amplitude spectrum with respect to the frequency axis is zero. The path continuity constraint is: |f inst (t)-f inst (t - Δt)| < 2Δf where f inst (t) is the extracted instantaneous frequency, Δt is the time resolution, Δf is the frequency resolution, N FFT is the number of FFT points, and f s is the sampling rate; The specific content of step S4 is: S41. Based on the time-frequency energy distribution of the time-frequency matrix generated by the synchronous extraction transform, use an improved density clustering algorithm to form independent target clusters by screening time-frequency points that meet the conditions through clustering parameters, introduce the Mahalanobis distance criterion for overlapping clusters to handle inter-cluster conflicts, and identify the time-frequency ridge line features of independent targets by separating the time-frequency energy ridge line; S42. Perform multi-target cross judgment. If there are multi-target crosses, perform dynamic decoupling and update the SET matrix; S43. For the range-velocity coupling error caused by moving targets, use the Doppler frequency shift model real-time compensation algorithm to correct the range-velocity coupling error by calculating the instantaneous frequency change rate of the target in real time; S44. Output the time-frequency trajectory of independent targets.

2. A method for dynamic separation and anti-interference of multiple targets of a frequency-modulated continuous-wave radar based on synchronous extraction transformation according to claim 1, characterized in that In step S1, the parameters of the transmitted signal are dynamically adjusted according to the scenario to ensure coverage of the detection range.

3. A method for dynamic separation and anti-interference of multiple targets of a frequency-modulated continuous-wave radar based on synchronous extraction transformation according to claim 1, characterized in that In step S2, the denoising processing is to perform wavelet threshold filtering preprocessing and signal normalization on the intermediate-frequency beat signal.

4. A method for dynamic separation and anti-interference of multiple targets of a frequency-modulated continuous-wave radar based on synchronous extraction transformation according to claim 1, characterized in that, In step S41, the Mahalanobis distance criterion introduced for overlapping clusters is: Among them, X i = [t i , f i T is the time-frequency point coordinate, μ j , Σ j are respectively the mean and covariance matrix of the j-th cluster;​ In step S42, the Doppler frequency shift model is: λ = c / f0 where c = 3×10 8 m / s, f0 is the starting frequency, and Δf d (t) is the target instantaneous frequency change rate, and v(t) is the target motion speed.

5. A method for dynamic separation and anti-interference of multiple targets of a frequency-modulated continuous-wave radar based on synchronous extraction transformation according to claim 1, characterized in that, The specific content of step S5 is: S51. Based on the preset interference feature library and the real-time time-frequency energy distribution, construct a dynamic time-frequency mask filter to filter out noise; S52. For the energy distribution characteristics of the time-frequency matrix, use statistical-driven dynamic threshold adjustment to perform detection threshold calculation and energy aggregation optimization.

6. A method for dynamic separation and anti-interference of multiple targets of a frequency-modulated continuous-wave radar based on synchronous extraction transformation according to claim 5, characterized in that, In step S51, the dynamic time-frequency mask filter is specifically: Among them, R k is an interference area, which is recognized through multi-level matching, including fixed clutter suppression and multipath interference detection; In step S52, the calculation of the detection threshold is: T(t) = μ(t) + k·σ(t) where μ(t) is the mean of the time-frequency energy within the sliding window, i.e., the window length, σ(t) is the standard deviation, and k is the frequency modulation slope, which is dynamically adjusted according to the signal-to-noise ratio; The energy aggregation is optimized for weak target signals across multiple time-frequency units, and the energy integration is achieved through morphological dilation operations: Among them, se is the structural element, se = 3×3, E enhanced (t, f) is the time-frequency energy after processing, and E(t, f) is the original time-frequency energy.

7. A multi-target dynamic separation and anti-interference system for frequency-modulated continuous-wave radar based on synchronous extraction transformation, characterized in that, A method for dynamic separation and anti-interference of multiple targets in a frequency-modulated continuous-wave radar based on synchronous extraction transformation according to any one of claims 1-6, comprising: a signal transmission module, a signal reception and preprocessing module, a synchronous extraction transformation time-frequency analysis module, a dynamic target separation and decoupling module, and an anti-interference processing module; The signal transmission module is used to generate a linearly frequency-modulated continuous wave and transmit a high-frequency signal through a radio frequency front end; The signal reception and preprocessing module is used to receive the target reflection signal, mix it with the transmitted signal through a mixer, extract the intermediate-frequency beat signal and perform denoising processing; The synchronous extraction transformation time-frequency analysis module is used to perform time-frequency decomposition on the signal by using synchronous extraction transformation to generate a high-resolution instantaneous frequency-time distribution matrix; The dynamic target separation and decoupling module is used to separate overlapping targets based on the time-frequency matrix by using a density clustering algorithm, combine a dynamic decoupling compensation algorithm to eliminate the range-velocity coupling error, and output the time-frequency trajectories of independent targets; The anti-interference processing module is used to construct a dynamic time-frequency domain mask filter to filter out noise, optimize the detection sensitivity through an adaptive threshold detection mechanism and perform false alarm control.

Images