A tunnel scene vehicle-mounted millimeter wave radar signal reconstruction method

Abstract

The application discloses a kind of tunnel scene under vehicle-mounted millimeter wave radar signal reconstruction method, it is related to signal processing and radar technical field, first, introduce the multi-modal measurement matrix Φ of different radar scene characteristics, to enhance the expression ability of sparse projection;Second, based on the sparse level estimation method of reconstruction error dynamic adjustment, improve the adaptability of algorithm to signal actual sparsity;Again cross stability analysis in the process of multiple iterations is combined with support set;While introducing the regularized residual feedback mechanism, strengthen the identification and compensation of weak component in each iteration;Finally, through the final reconstruction tuning strategy integration, further improve reconstruction precision and robustness;Experimental results show that the method of the application has better detection rate, lower error and stronger generalization ability under different sparsity levels, signal-to-noise ratio and sampling rate, especially suitable for resource-constrained environment under vehicle-mounted radar system signal processing.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of signal processing and radar technology, in particular to a vehicle-mounted millimeter wave radar signal reconstruction method in a tunnel scene. BACKGROUND

[0002] With the rapid development of automatic driving and intelligent transportation, vehicle-mounted millimeter wave radar gradually becomes an important sensor for vehicle environment perception due to its high resolution, strong anti-interference and all-weather working characteristics. In the tunnel scene, due to the closed space and smooth wall surface, radar signals are prone to strong multipath effects. Multipath interference not only causes false targets and energy dispersion, but also leads to a decrease in signal sparsity, thereby affecting the reliability of target detection and tracking.

[0003] In terms of signal acquisition, traditional millimeter wave radar signal acquisition is usually based on the Nyquist sampling theorem, which requires sampling at a rate much higher than the signal bandwidth, resulting in high sampling rate, large data volume, long processing delay and high hardware cost. Especially in the resource-constrained embedded vehicle scene, due to the limitation of computing resources and energy consumption, these problems are more prominent, which seriously restricts the deployment and application of high-resolution radar systems.

[0004] In recent years, CS (Compressed Sensing, compressed sensing) technology provides a new solution for radar signal acquisition and reconstruction. Based on the sparsity of signals in a certain domain, the method obtains observation values with a sampling number much lower than the Nyquist rate by designing a suitable measurement matrix Φ, and restores the original signal using a reconstruction algorithm.

[0005] In terms of reconstruction algorithm, the mainstream methods include greedy algorithms (such as orthogonal matching pursuit OMP, compressed sampling matching pursuit CoSaMP), convex optimization methods (such as basis pursuit BP, L1 minimization) and threshold iteration algorithms (such as iterative hard thresholding IHT, alternating direction method of multipliers ADMM, etc.). Among them, CoSaMP has high practical value in engineering implementation due to its faster reconstruction speed and moderate accuracy. However, these algorithms generally rely on a pre-set sparse level parameter, and lack the ability to adaptively adjust the sparsity of actual signals, affecting their adaptability and robustness in complex dynamic scenarios.

[0006] In the vehicle-mounted millimeter wave radar scene, some research has applied compressed sensing to target detection, range-Doppler spectrum estimation and moving target recognition tasks. For example, some scholars have introduced compressed sensing for sparse echo reconstruction in FMCW radar systems, significantly reducing the number of sampling points while preserving key target information; some research has combined multi-channel sparse reconstruction methods to jointly process target echoes, thereby improving multi-target resolution.

[0007] Although the current achievements have made positive progress, there are still the following problems: most algorithms fix the sparse level setting and cannot adapt to the changing signal sparse characteristics, resulting in a decline in recovery performance in strong multipath or resource-limited scenarios; during the iterative reconstruction process, the support set often fluctuates, there is a lack of judgment and feedback mechanism based on the stability measurement of the support set, which affects the overall reconstruction effect; there is a lack of feedback optimization mechanism integrating multi-dimensional criteria, making it difficult to perform dynamic correction based on residual distribution; existing researches mainly focus on static radar scenarios, and there is a lack of customized optimization design for vehicle-mounted dynamic multi-target environment. SUMMARY

[0008] To solve the above technical problems, the present application provides a vehicle-mounted millimeter wave radar signal reconstruction method in a tunnel scene, comprising the following steps:

[0009] S1, radar echo signal preprocessing and multi-dimensional feature extraction: after collecting the original echo signal from the vehicle-mounted millimeter wave radar, perform signal preprocessing operation;

[0010] S2, measurement matrix construction and selection: based on the number of multipath interference, compression ratio, hardware resource occupancy rate and target radial velocity, select different measurement matrix construction strategies;

[0011] S3, compressed sensing observation: using the measurement matrix constructed in step S2 to subsample the preprocessed radar echo signal;

[0012] S4, initial sparse level estimation and pruning processing: estimating the initial sparse level of the signal according to the projection energy proportion, and screening the initial support set to eliminate low-confidence candidates; optimizing and updating the sparse level according to the correlation peak value ratio, projection energy concentration degree and energy distribution entropy;

[0013] S5, support set extraction and stability judgment: based on the estimated sparse level and compressed observation, using an iterative algorithm to extract the support set of the signal, and judging its stability, if it meets the requirements, step S7 is executed, otherwise step S6 is executed;

[0014] S6, re-estimation and update of unstable support set: according to the variation amplitude of the support set, the estimated value is adaptively relaxed or tightened, the measurement weight or residual error is adjusted and fed back, and the support set extraction process is re-entered, i.e. returning to step S4;

[0015] S7, regularized residual feedback update: after the stability of the support set, a regularized residual feedback mechanism is introduced to update the signal residual error;

[0016] S8, error judgment and loop feedback: judging whether the current reconstructed signal meets the error threshold, if it meets the requirements, step S9 is executed, otherwise it goes to step S4;

[0017] S9. Final Reconstruction and Tuning Output and Reconstruction Signal Output: Under the premise of meeting the error requirements, the final reconstruction and tuning steps are executed to generate the final estimated signal.

[0018] The technical solution further defined in this invention is:

[0019] Furthermore, in step S1, the target radial velocity is estimated during preprocessing using the radar range-Doppler spectrum, assuming the original radar echo signal is... N represents the length of the sparse signal to be recovered. The following is represented as:

[0020]

[0021] Where s represents the sparse coefficient vector; The number of non-zero components in vector s is represented by K, and the sparsity level of the signal is represented by K. The sparse basis Ψ is used to transform the original signal x into an orthogonal or incomplete basis of the sparse representation s.

[0022] As described above, in the method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario, the measurement matrix construction strategy in step S2 is as follows:

[0023] When the number of multipath interferences is ≤4 and the hardware resource utilization rate is ≤60%, the Gaussian-Fourier mixture mode is selected, and the Gaussian random matrix is... With partial Fourier matrices Perform dimensional stitching:

[0024]

[0025] Where M1 represents the number of compressed observations taken in Gaussian measurement mode, M2 represents the number of compressed observations taken in Fourier measurement mode, and M1+M2=M; N represents the length of the original signal, that is, the length of the sparse signal to be recovered.

[0026] When the number of multipath interferences is ≥5 and the hardware resource utilization rate is ≤50%, multimodal weighted fusion is adopted, and the weight α is dynamically adjusted to improve robustness.

[0027]

[0028] The weighting coefficients α1, α2, and α3 are dynamically adjusted based on the signal-to-noise ratio, sampling rate, and ambient noise level, where α1 + α2 + α3 = 1, and α i ≥0.2, α1, α2, and α3 are dynamically updated based on the real-time acquired signal-to-noise ratio and compression ratio, with the update period consistent with the radar signal frame period:

[0029] α1=max(0.2,0.5×(SNR / SNR_max))

[0030] a2 = max(0.2, 0.3 x (1 - CR))

[0031] a3 = 1 - a1 - a2

[0032] wherein a1 represents a correlation threshold adjustment factor, used to control how many strong correlation elements are retained in support set extraction, SNR represents real-time signal-to-noise ratio, SNR_max represents maximum tolerable signal-to-noise ratio of the radar system, when SNR≥10dB, a1≥0.25; a2 represents a stability judgment adjustment factor, used to judge the consistency degree of the support set in multiple iterations, CR represents compression ratio, when CR≤0.5, a2≥0.15; a3 represents a sparsity level estimation update adjustment factor, used to control whether the new sparsity level, i.e. K value estimation, is adjusted up or down, and a3≥0.2;

[0033] When the hardware resource occupancy rate > 70%, the Gaussian matrix Φ is dynamically selected according to the compression ratio G or the Hattley matrix Φ H :

[0034]

[0035] wherein M represents the number of rows of the measurement matrix Φ, and γ represents a threshold parameter for controlling the selection of the measurement matrix.

[0036] The tunnel scene vehicle-mounted millimeter wave radar signal reconstruction method as described above, in step S3, the compressed sensing measurement is performed, and a compressed observation model is constructed as follows:

[0037] y = Φx + n = ΦΨs + n

[0038] wherein represents a compressed observation vector, represents an equivalent sensing matrix, N represents the length of the original signal, n represents a Gaussian white noise, and the length is M, which has the same dimension as the compressed observation vector y; n ~ N0(0, σ 2 I), N0(0, σ 2 I) represents a multi-dimensional Gaussian distribution, wherein the mean vector is 0, and the covariance matrix is σ 2 I, σ represents a standard deviation calculated according to the set signal-to-noise ratio, and I represents a unit matrix with a size of M x M.

[0039] The tunnel scene vehicle-mounted millimeter wave radar signal reconstruction method as described above, in step S4, the adaptive hierarchical coupling multi-index sparsity level judgment mechanism is used for sparsity level estimation and dynamic updating process based on the characteristics of the multipath environment, which specifically includes the following steps:

[0040] S4.1, residual initialization and correlation calculation: before the 0th iteration, there is no any estimation of x, so So the initial residual is set as r (0) =y; the residual is denoted as ;

[0041] Calculate the correlation vector of the current residual and the columns of the measurement matrix: u = Φ T r (0) ; where, denotes the reconstruction importance of each position, Φ T denotes the transpose of the measurement matrix Φ, used for inverse projection to the signal dimension, obtaining the importance measure of each signal component to the current residual;

[0042] S4.2, projection energy calculation and sorting: set is the i-th column of the measurement matrix Φ, then ; calculate the observation residual r (t) and the correlation vector u = Φ T r (t) of the sensing matrix Φ;

[0043] Calculate the projection energy e i of each atom, the square of the projection value at the i-th position is , and arrange it in descending order:

[0044] e (1) ≥e (2) ≥...≥e (N)

[0045] S4.3, adaptive sparse level estimation based on energy coverage: calculate the cumulative energy proportion of the top k largest energies :

[0046]

[0047] According to the target characteristics in the tunnel multipath environment, adaptively select the energy coverage threshold ρ, find the smallest k, so that the top k largest energies have covered ρ proportion of the total energy, as shown in the following formula:

[0048] K init =min{k:E k ≥ρ},ρ∈[0.6,0.9]

[0049] Where, ρ represents the energy coverage threshold, when the number of multipaths n mp ≤2, the radar signal shows single target echo characteristics, and the proportion of non-zero components ≤10%, ρ takes 0.6-0.7; when 3≤n mp ≤4, the radar signal shows multi-target echo characteristics, and the proportion of non-zero components is 10%-30%, ρ takes 0.7-0.8; when n mpWhen the radar signal is ≥5, it shows weak target characteristics, and when the non-zero component ratio is >30%, ρ is taken as 0.8-0.9;

[0050] According to the amplitude sorting pruning, the indexes corresponding to the first maximal amplitude are taken as the initial support set .

[0051] S4.4, calculate the correlation peak ratio:

[0052]

[0053] Where, u i represents the i-th element in the correlation vector, that is, the correlation value of the residual vector and the i-th column of the measurement matrix; u j represents the j-th element in the correlation vector, which is used to calculate the average level of the correlation amplitude of the full signal; N represents the length of the original signal, that is, the total number of elements of the correlation vector u; CPR represents the correlation peak ratio, which is used to measure the sparsity, the more prominent the peak, the sparser the signal;

[0054] S4.5, calculate the energy distribution entropy: normalize the current residual correlation vector u (t) to a probability distribution:

[0055]

[0056] In order to introduce the energy concentration degree judgment, the normalized entropy of the support set amplitude is defined:

[0057]

[0058] Where, ϵ represents a small amount to avoid log0; H (t) is larger, the residual distribution is more uniform, which means that the signal dispersion has not been captured, and the sparsity level is increased; on the contrary, H (t) is smaller, which reduces the sparsity level;

[0059] S4.6, coupling candidate spectrum, calculating ΔK: let the current estimated sparsity level be K (t) , and the step size is ΔK (t) , then:

[0060]

[0061] Where, CPR(t) represents the correlation peak ratio of the t-th iteration, P target represents the expected peak ratio; v (t) represents the projection energy concentration degree of the candidate set, which is defined as:

[0062]

[0063] Where, ci (t) represents the projection energy of the i-th column atom pair residual:

[0064]

[0065] where the energy mean is:

[0066]

[0067] where T t represents the candidate atom set selected in the t-th iteration, τ v represents the stability threshold, H (t) represents the energy entropy of the support set, log2k (t) represents the sparsity level scale factor; β1, β2, and β3 represent the peak ratio adjustment weight, residual projection energy distribution weight, and information entropy normalization adjustment weight, respectively;

[0068] Update the sparsity level:

[0069]

[0070] where sign(ΔK (t) ) represents the sign function, which determines the direction of sparsity level adjustment: +1 indicates increasing the sparsity level, -1 indicates decreasing the sparsity level, and 0 indicates no adjustment; ΔK max represents the maximum allowed value of sparsity level adjustment in each iteration, which is set to 10%-20% of the current sparsity level;

[0071] The updated sparsity level K is taken as the input for the next round, and the residual, peak ratio, and information entropy are repeatedly calculated and the sparsity level is updated, forming an adaptive optimization closed loop to achieve dynamic adjustment of the signal sparsity level.

[0072] The tunnel scene vehicle-mounted millimeter wave radar signal reconstruction method as described above, in step S5, in the support set determination and update process, the concentration index of the candidate set projection energy distribution is introduced, the support set stability is weighted and determined, and the closed loop update is driven, which specifically includes:

[0073] Set the sliding window size to 3, and denote the support set in the window as S t-2 , S t-1 , and S t , t represents the current iteration number, and the base threshold is denoted as J thresh =0.85; calculate the Jaccard similarity:

[0074]

[0075] where S i represents the support set obtained in the i-th iteration, and S jSj represents the support set obtained in the jth iteration;

[0076] The sliding window stability index is set as the average Jaccard similarity of all adjacent support sets in the window:

[0077]

[0078] The support set energy entropy weight is: The residual error descent rate is: v(t) represents the degree of the candidate spectrum set;

[0079] If Stab(t) is greater than or equal to 0.85, it is determined that the support set is stable, and step S7 is performed to enter the final reconstruction stage; if Stab(t) is less than 0.85, it is determined that the support set is unstable, and step S6 is performed to enter the candidate expansion step.

[0080] The method for reconstructing a vehicle-mounted millimeter wave radar signal in a tunnel scene as described above, in the candidate expansion step in step S6, the candidate support set T is constructed based on the residual error projection score , wherein current k represents the size of the current estimated support set; the score mechanism is introduced, and in the tth iteration, the score of each column is calculated according to the residual error, which is used to evaluate the contribution of each basis vector to the residual error; the projection intensity and the matrix column norm are combined to select the candidate index with the highest score from the non-current support set:

[0081]

[0082] , wherein C i represents the correlation projection coefficient; r (t-1) represents the residual energy of the last iteration; a j,i represents the element of the jth row and the ith column of the measurement matrix Φ; ε represents a small positive number to prevent the denominator from being 0;

[0083] The elements of the current support set are excluded, and the score is calculated for all unselected atoms, and then the top L atoms with the highest scores are selected from the candidate set T, wherein L is a function of the current sparse estimate K, so:

[0084]

[0085] , wherein min(3K (t) ,·) represents the first layer limit, represents the second layer correction term, and the support set history feedback is introduced; the set union operation is performed on the candidate set T and the current support set S to form the expanded support set and orthogonal-triangular decomposition of the extended support set, i.e., decomposing a matrix into the product of an orthogonal matrix and an upper triangular matrix, as the input of the subsequent least square estimation.

[0086] The method further comprises the following steps: in step S6, setting a coefficient according to the median and the mean absolute deviation to construct a screening threshold, and eliminating abnormal values; the screening threshold is , retaining the support set corresponding to the coefficients exceeding the threshold, and updating the support set , and reconstructing the signal as , wherein θ represents the signal amplitude estimation obtained after fitting the observation y with the support set column matrix, MAD represents the mean absolute deviation, and σ represents the standard deviation calculated according to the set signal-to-noise ratio.

[0087] The method further comprises the following steps: in step S7, performing regularized feedback updating on the residual error, and the feedback weight λ (t) is dynamically adjusted according to the iteration number in an exponential decay law:

[0088] The basic residual error is calculated , and a regularized feedback term is embedded in the residual error updating from the sixth iteration to enhance the robustness , wherein λ (t) represents an iteration-related exponential decay regularization coefficient, and is defined as:

[0089]

[0090] , wherein λ0=0.05×σ represents an initial coefficient, and τ=5 represents a decay factor.

[0091] The residual error is updated as

[0092]

[0093] Otherwise:

[0094]

[0095] It is determined whether the residual error norm satisfies the following formula:

[0096]

[0097] or the residual error is not reduced to the following formula:

[0098]

[0099] , wherein represents a residual error reduction ratio threshold, and the iteration is stopped in advance when the residual error reduction ratio is less than .

[0100] The vehicle-mounted millimeter wave radar signal reconstruction method in a tunnel scene as described above, in step S9, record the support set S of the last three iterations (t) 、S (t-1) And S (t-2) , by calculating the average Jaccard similarity, determine whether the support set reaches a stable state, Jaccard is obtained by calculating the ratio of the intersection size and the union size of two sets:

[0101]

[0102] Wherein, T represents the total number of iterations, t represents the current iteration index, S (t) The support set of the tth iteration is represented by S

[0103] When J avg <0.85, it is determined that the support set has not reached a stable state, and the final support set optimization is performed, the final support set optimization is performed, the size of the support set is dynamically adjusted based on the signal amplitude, and the least square is solved again to obtain the final reconstruction result:

[0104]

[0105]

[0106] Wherein, A S The column submatrix corresponding to the support set S in the measurement matrix is represented by A The transpose of the column submatrix A S , y represents the observation vector, The optimal solution on the support set is represented by x

[0107] The least square method is used on the support set S to solve the coefficient vector x , and then it is embedded into the sparse vector x * with a full length of N, the coefficients outside the support set are set to zero, and the final complete sparse coefficient vector x * is obtained, and the final reconstruction signal x is recovered according to the final reconstruction signal:

[0108] .

[0109] The beneficial effects of the present application are:

[0110] (1) In the application, by combining the Gaussian random matrix with the Fourier matrix, and combining the hardware resource occupation rate, the number of multipath in the tunnel environment and the compression ratio and other environmental constraints, an adaptive multi-modal measurement matrix selection mechanism is proposed;When the multipath interference is less and the hardware load is moderate, a hybrid matrix of Gaussian and Fourier is used to balance randomness and sparsity, to ensure accurate recovery of the signal;When the multipath is complex or the noise is serious, a weighted fusion strategy is introduced, and the weights of each mode are dynamically adjusted according to the real-time signal-to-noise ratio and compression ratio, to improve the robustness of the system in a strong interference environment;When the hardware resources are tight, automatically switch to a lightweight matrix to reduce the computational overhead and ensure the real-time performance of the vehicle-mounted system;The mechanism can maintain good reconstruction accuracy and computational efficiency in different complexity tunnel scenarios, overcoming the poor adaptability of traditional algorithms under fixed matrix conditions, and greatly enhancing the practicality of the method in resource-constrained environments;

[0111] (2) In the application, a hierarchical coupling judgment mechanism based on correlation peak value ratio, energy concentration degree and energy distribution entropy is proposed, which can adaptively estimate and dynamically adjust the sparsity in each iteration, effectively avoiding reconstruction distortion caused by excessive or insufficient sparsity;At the same time, the application introduces a support set stability analysis method in the iteration process, which considers Jaccard similarity, residual drop rate and energy concentration degree to determine the convergence of the support set;When the support set is detected to be unstable, the system will trigger the candidate expansion and robust threshold screening mechanism to automatically remove false atoms caused by tunnel reflection, thereby ensuring the reliability of the support set;This design significantly reduces the false alarm rate, improves the precision, recall rate and F1 score, and realizes robust adaptation to complex multipath environments;

[0112] (3) In the application, in the iteration update stage, the application introduces a residual-gradient hybrid regularization feedback mechanism, which uses dynamic weights with exponential decay to enhance the identification and compensation of weak target components, and still maintains high detection sensitivity under tunnel multipath interference;In the final stage, the application further introduces a weighted least squares optimization based on amplitude weight, which performs amplitude weighting correction on the coefficients on the support set, significantly improving the accuracy of amplitude estimation and the overall fidelity of the signal;Through comparison experiments with OMP and CoSaMP, the application can achieve lower mean square error (MSE) and higher reconstruction signal-to-noise ratio (RSNR) under the same conditions, while maintaining a faster convergence speed;This design not only improves the recovery ability of the algorithm for weak targets, but also ensures the advantages of the final reconstructed signal in accuracy and stability, and has significant engineering application value. BRIEF DESCRIPTION OF DRAWINGS

[0113] Figure 1 is a schematic diagram of the overall process of the application;

[0114] Figure 2A schematic diagram of the measurement matrix construction strategy in the embodiment of the present application is shown in the figure;

[0115] Figure 3 A flowchart of the sparsity estimation in the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0116] The method provided by the embodiment of the present application is a signal reconstruction method for a vehicle-mounted millimeter wave radar in a tunnel scene, which comprises the following steps as shown in the figure: Figure 1

[0117] S1, radar echo signal preprocessing and multi-dimensional feature extraction: after collecting the original echo signal from the vehicle-mounted millimeter wave radar, the signal preprocessing operation is performed.

[0118] The target radial velocity is estimated by the radar range-Doppler spectrum during the preprocessing, and the original radar echo signal is denoted as x = (x0, x1, x2, …, xN-1)T, where N represents the length of the sparse signal to be recovered. Under the condition that the length of the original signal is N, the signal is denoted as:

[0119]

[0120] where s represents the sparse coefficient vector; K represents the sparsity level of the signal; and the sparse basis Ψ is used to transform the original signal x into the orthogonal or under-complete basis of the sparse representation s.

[0121] S2, measurement matrix construction and selection: based on the number of multipath interference, the compression ratio, the hardware resource occupancy rate and the target radial velocity, different measurement matrix construction strategies are selected.

[0122] As shown in the figure, when the number of multipath interference is small (≤4) and the hardware resource occupancy rate is ≤60%, the number of interference targets is not large at this time, and the randomness (Gaussian) + frequency domain sparsity (Fourier) can ensure the signal recovery accuracy, and the calculation cost is low, so the mixed Gaussian-Fourier mode is selected, and the Gaussian random matrix Figure 2 is dimensionally spliced with the partial Fourier matrix

[0123]

[0124] where M1 represents the number of compressed observations taken in the Gaussian measurement mode, which is much smaller than N, M2 represents the number of compressed observations taken in the Fourier measurement mode, which is much smaller than N, and M1+M2=M; and N represents the length of the original signal, i.e. the length of the sparse signal to be recovered.

[0125] ​​​​​When the number of multipath interference is large (≥5) and the hardware resource occupancy rate is ≤50%, the number of interference sources is too large, the robustness of single matrix combination is insufficient, and the matrix diversity needs to be increased by weight fusion to suppress multipath noise; multi-modal weighted fusion is adopted to dynamically adjust the weight α to improve the robustness:

[0126]

[0127] Among them, the weight coefficients α1, α2, α3 are dynamically adjusted according to the signal-to-noise ratio, sampling rate and environmental noise level, α1+α2+α3=1, and α i ≥0.2, α1, α2 and α3 are dynamically updated according to the real-time collected signal-to-noise ratio and compression ratio, and the update period is consistent with the radar signal frame period:

[0128] α1=max(0.2,0.5×(SNR / SNR_max))

[0129] α2=max(0.2,0.3×(1-CR))

[0130] α3=1-α1-α2

[0131] Among them, α1 represents the correlation threshold adjustment factor, which is used to control how many strong correlation elements are retained in the support set extraction, SNR represents the real-time signal-to-noise ratio, SNR_max represents the maximum tolerable signal-to-noise ratio of the radar system, when SNR≥10dB, α1≥0.25; α2 represents the stability judgment adjustment factor, which is used to judge the consistency of the support set in multiple iterations, CR represents the compression ratio, when CR≤0.5, α2≥0.15; α3 represents the sparse level estimation update adjustment factor, which is used to control whether the new sparse level K value estimation is adjusted up or down, and α3≥0.2.

[0132] When the hardware resources are tight (occupancy rate>70%), the real-time performance and calculation load are guaranteed, and the Gaussian matrix Φ G or the Hartley matrix Φ H is dynamically selected according to the compression ratio:

[0133]

[0134] Among them, M represents the number of rows of the measurement matrix Φ, and γ represents the threshold parameter for controlling the selection of the measurement matrix, which can be set according to the specific application scenario.

[0135] S3, compressed sensing observation: the measurement matrix constructed in step S2 is used to subsample the preprocessed radar echo signal.

[0136] Based on the multi-modal measurement matrix Φ matched with the tunnel multipath environment constructed in step S2, the preprocessed radar echo signal x is linearly projected to construct a compressed observation model as follows:

[0137] y=Φx+n=ΦΨs+n

[0138] where, denotes the compressed observation vector, denotes the equivalent sensing matrix, N denotes the length of the original signal, n denotes the Gaussian white noise, length M, same dimension as the compressed observation vector y; n~N0(0,σ 2 I), N0(0,σ 2 I) denotes a multi-dimensional Gaussian distribution, where the mean vector is 0, Cov(covariance) is σ 2 I, σ denotes the standard deviation calculated according to the set signal-to-noise ratio, I denotes the unit matrix, size MxM.

[0139] S4, initial sparse level estimation and pruning processing: estimate the initial sparse level of the signal according to the projection energy proportion, and perform screening of the initial support set, and eliminate low-confidence candidates; according to the correlation peak value ratio, projection energy concentration degree and energy distribution entropy, the sparse level is optimized and updated.

[0140] Based on the characteristics of the multipath environment, an adaptive hierarchical coupling multi-index sparse level judgment mechanism is adopted for sparse level estimation and dynamic updating process, as shown in the following figure: Figure 3 , which includes the following steps:

[0141] S4.1, residual initialization and correlation calculation: before the 0th iteration, there is no any estimation of x, so set , so the initial value of the residual is set as r (0) =y; the residual after that is ;

[0142] Calculate the correlation vector of the current residual and the column of the measurement matrix: u=Φ T r (0) ; wherein, denotes the reconstruction importance of each position, Φ T denotes the transpose of the measurement matrix Φ, used for inverse projection to the signal dimension, to obtain the importance measure of each signal component to the current residual.

[0143] S4.2, projection energy calculation and sorting: set is the i-th column of the measurement matrix Φ, then ; calculate the correlation vector u=Φ (t) r T of the observation residual r (t) ;

[0144] Calculate the projection energy e i of each atom (column of the measurement matrix), and the square of the projection value at the i-th position is , in descending order:

[0145] e (1) ≥e (2) ≥...≥e (N) .

[0146] S4.3, Adaptive sparse level estimation based on energy coverage ratio: Calculate the cumulative energy proportion of the top k largest energies :

[0147]

[0148] According to the target characteristics in the tunnel multipath environment, the energy coverage threshold p is adaptively selected, and the smallest k is found, so that the top k largest energies have covered p proportion of the total energy, as shown in the following formula:

[0149] K init =min{k:E k ≥ρ},ρ∈[0.6,0.9]

[0150] Where p represents the energy coverage threshold, when the environment is simple (the number of multipaths n mp ≤2), the radar signal shows single target echo characteristics, and the proportion of non-zero components is ≤10%, p takes 0.6-0.7; when the environment is medium (the number of multipaths 3≤n mp ≤4), the radar signal shows multi-target echo characteristics, and the proportion of non-zero components is 10%-30%, p takes 0.7-0.8; when the environment is complex (the number of multipaths n mp ≥5), the radar signal shows weak target characteristics, and the proportion of non-zero components is >30%, p takes 0.8-0.9;

[0151] According to the amplitude sorting pruning, the top largest amplitude corresponding index is taken as the initial support set .

[0152] S4.4, Calculate the correlation peak ratio, which is used to measure the sparsity of the signal:

[0153]

[0154] Where u i represents the i-th element in the correlation vector, i.e. the correlation value of the residual vector and the i-th column of the measurement matrix; u j represents the j-th element in the correlation vector, which is used to calculate the average level of the correlation amplitude of the whole signal; N represents the length of the original signal, i.e. the total number of elements of the correlation vector u; CPR represents the correlation peak ratio, which is used to measure the sparsity, the more prominent the peak, the sparser the signal.

[0155] S4.5, Calculate the energy distribution entropy: Calculate the current residual correlation vector u (t) Normalized to a probability distribution:

[0156]

[0157] To introduce the energy concentration criterion, define the normalized entropy of the support set amplitude:

[0158]

[0159] Where, ϵ represents a small amount to avoid log0; H (t) The larger, the more uniform the residual distribution, indicating that the signal dispersion has not been captured, and the sparsity level should be increased; otherwise, H (t) The smaller, the more concentrated the residual in a small number of elements, and the sparsity level may be too large, and the sparsity level needs to be reduced.

[0160] S4.6, Calculate ΔK (coupling candidate spectrum): Let the current estimated sparsity level be K (t) , the step size is ΔK (t) , then:

[0161]

[0162] Where, CPR(t) represents the correlation peak ratio of the tth iteration, reflecting the contrast of the strongest atom and the average level; P target represents the expected peak ratio, which can be obtained by statistical peak ratio distribution of real radar signals in tunnel environment, and the signal is more dispersed under tunnel multipath, P target ∈[3.5,4.5], more than 4.

[0163] v (t) represents the projection energy concentration of the candidate set, defined as:

[0164]

[0165] Where, c i (t) represents the projection energy of the i-th column atom on the residual:

[0166]

[0167] Where the energy mean is:

[0168]

[0169] Where, T t represents the candidate atom set selected in the tth iteration; τ v represents the stability threshold, which is taken as τ v ∈[2,2.5] to balance the interference of strong targets and multipath false peaks; H(t) represents the energy entropy of the support set, measuring the uncertainty of the energy distribution; log2k (t) represents the sparsity level scale factor, used for normalization to prevent the entropy term from being too large.

[0170] β1, β2 and β3 represent weight coefficients for controlling the contribution of the two indicators; peak ratio adjustment weight: β1 ∈ [0.5-0.6], to ensure that the peak ratio dominates the sparsity level adjustment and is the core driver; residual projection energy distribution weight: β2 ∈ [0.3, 0.8], to avoid excessive increase in sparsity level due to multipath virtual atoms, as an auxiliary adjustment; information entropy normalization adjustment weight: β3 ∈ [0.2, 0.3], to suppress excessive expansion of sparsity level, as a safety valve.

[0171] Update the sparsity level:

[0172]

[0173] where sign(ΔK (t) ) represents the sign function, used to determine the direction of sparsity level adjustment: +1 means increasing the sparsity level, -1 means decreasing the sparsity level, and 0 means no adjustment; ΔK max represents the maximum allowed value of sparsity level adjustment per iteration, used to avoid reconstruction divergence caused by excessive jumps, set to 10%-20% of the current sparsity level.

[0174] Take the updated sparsity level K as the input for the next round, repeat the calculation of residual, peak ratio and information entropy and update the sparsity level, form an adaptive optimization closed loop, and realize the dynamic adjustment of signal sparsity level.

[0175] S5, support set extraction and stability judgment: based on the estimated sparsity level and compressed observation, an iterative algorithm is used to extract the support set of the signal, and its stability is judged, if it meets the requirements, step S7 is executed, otherwise step S6 is executed.

[0176] In the support set determination and update process, the concentration index of the candidate set projection energy distribution is introduced, the support set stability is weighted and judged, and the closed loop update is driven, which specifically includes:

[0177] Set the sliding window size to 3, and denote the support set in the window as S t-2 , S t-1 and S t , t represents the current iteration number, and the basic threshold is denoted as J thresh =0.85; calculate the Jaccard similarity:

[0178]

[0179] where S iS (i) represents the support set obtained in the i-th iteration j S (j) represents the support set obtained in the j-th iteration.

[0180] The sliding window stability index is set as the average Jaccard similarity of all adjacent support sets in the window:

[0181]

[0182] The support set energy entropy weight is: The residual error reduction rate is: v(t) represents the degree of concentration in the candidate spectrum; if Stab(t)≥0.85, it is determined that the support set is stable, step S7 is executed, and the final reconstruction phase is entered; if Stab(t)<0.85, it is determined that the support set is unstable, step S6 is executed, and the candidate expansion step is entered.

[0183] S6, re-estimation and update of unstable support set: the estimated value is adaptively relaxed or tightened according to the support set variation amplitude, the measurement weight or the residual error is feedback adjusted, and the support set extraction process is re-entered, i.e. returning to step S4.

[0184] In the candidate expansion phase, the candidate support set is constructed based on the residual error projection score , wherein current k represents the size of the current estimated support set; in order to overcome the interference of multipath false targets in the tunnel environment on the construction of the candidate set, a scoring mechanism is introduced, and in the t-th iteration, the score of each column is calculated according to the residual error, which is used to evaluate the contribution of each basis vector to the residual error; in combination with the projection intensity and the matrix column norm, the candidate index with the highest score that is not in the current support set is selected:

[0185]

[0186] , wherein C i represents the correlation projection coefficient; r (t-1) represents the residual energy of the last iteration; a j,i represents the element of the j-th row and i-th column of the measurement matrix Φ; ε represents a small positive number to prevent the denominator from being 0.

[0187] Excluding the elements of the current support set, the score is calculated for all unselected atoms, and then the top L atoms with the highest scores are selected to form the candidate set T, wherein L is a function of the current sparse estimate K, so:

[0188]

[0189] , wherein min(3K (t) ,·) represents the first layer limit, represents the second layer correction term, and a support set history feedback is introduced; the candidate set T is merged with the current support set S to form an extended support set , and the extended support set is subjected to a QR decomposition, i.e., a matrix is decomposed into a product of an orthogonal matrix and an upper triangular matrix, as an input of subsequent least square estimation.

[0190] To suppress false atomic interference caused by multipath effect in a tunnel environment, a robust threshold screening of median and mean absolute deviation (MAD) setting is introduced on the extended support set to eliminate abnormal values; the screening threshold is , the support set corresponding to the coefficients exceeding the threshold is retained, the support set is updated , and the reconstructed signal is , θ represents a signal amplitude estimation obtained after the support set column matrix is fitted to the observation y, MAD represents mean absolute deviation, and σ represents a standard deviation calculated according to a set signal-to-noise ratio.

[0191] S7, regularized residual feedback update: after the support set is stabilized, a regularized residual feedback mechanism is introduced to update the signal residual; the residual is subjected to regularized feedback update, and the feedback weight λ (t) is dynamically adjusted according to the iteration number in an exponential decay law.

[0192] The basic residual is calculated , and a regularized feedback term is embedded in the residual update from the 6th iteration to enhance robustness , where λ (t) represents an iteration-related exponential decay regularization coefficient, and is defined as:

[0193]

[0194] where λ0=0.05×σ represents an initial coefficient, and τ=5 represents a decay factor.

[0195] The residual is updated as:

[0196]

[0197] Otherwise:

[0198]

[0199] It is judged whether the residual norm satisfies the following formula:

[0200]

[0201] or the residual is not reduced as follows:

[0202]

[0203] wherein, represents the threshold of the residual reduction ratio, when the residual reduction ratio is less than the iteration is stopped in advance.

[0204] S8, error judgment and loop feedback: judge whether the current reconstructed signal meets the error threshold, if yes, execute step S9, otherwise go to step S4.

[0205] S9, final reconstruction optimization output and reconstructed signal output: execute the final reconstruction optimization step to generate the final estimated signal under the premise of meeting the error requirement.

[0206] Record the support set S (t) , S (t-1) and S (t-2) , determine whether the support set reaches a stable state by calculating the average Jaccard similarity, and the Jaccard is obtained by calculating the ratio of the intersection size to the union size of two sets:

[0207]

[0208] where T represents the total number of iterations, t represents the current iteration index (from 1 to T), S (t) represents the support set of the tthiteration.

[0209] When J avg <0.85, it is determined that the support set has not reached a stable state, and the final support set optimization is executed, the size of the support set is dynamically adjusted based on the signal amplitude, and the final reconstruction result is obtained by re-solving the least square:

[0210]

[0211]

[0212] where A S represents the column submatrix corresponding to the support set S in the measurement matrix, represents the transpose of the column submatrix A S , y represents the observation vector, represents the optimal solution on the support set (in the sense of least square).

[0213] Solve the coefficient vector on the support set S by using the least square method , and then embed it into the sparse vector x * with the full length of N, the coefficients outside the support set are set to zero, and the final complete sparse coefficient vector x * is obtained, and the final reconstructed signal is recovered accordingly:

[0214] .

[0215] The method of the embodiment first introduces a multi-modal measurement matrix Φ suitable for different radar scene characteristics to enhance the sparse projection expression capability; secondly, based on the sparse level estimation method of the reconstruction error dynamic adjustment, the adaptability of the algorithm to the actual sparsity of the signal is improved; combined with the cross stability analysis of the support set in the multi-iteration process; at the same time, the regularization residual feedback mechanism is introduced, and the identification and compensation of weak components are strengthened in each iteration; finally, through the final reconstruction optimization strategy integration, the reconstruction accuracy and robustness are further improved; the experimental results show that the method has better detection rate, lower error and stronger generalization ability under different sparse levels, signal-to-noise ratios and sampling rates, and is especially suitable for signal processing of vehicle-mounted radar systems in resource-constrained environments.

[0216] In addition to the above embodiments, the present application can also have other implementation manners. Any technical solution formed by equivalent replacement or equivalent transformation falls within the protection scope required by the present application.

Claims

1. A method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario, characterized in that: Includes the following steps: S1. Radar echo signal preprocessing and multi-dimensional feature extraction: After acquiring the raw echo signal from the vehicle-mounted millimeter-wave radar, preprocessing operations are performed on the signal. S2. Measurement Matrix Construction and Selection: Based on the number of multipath interferences, compression ratio, hardware resource utilization, and target radial velocity, different measurement matrix construction strategies are selected; the specific measurement matrix construction strategies are as follows: When the number of multipath interferences is ≤4 and the hardware resource utilization rate is ≤60%, the Gaussian-Fourier mixture mode is selected, and the Gaussian random matrix is... With partial Fourier matrices Perform dimensional stitching: Where M1 represents the number of compressed observations taken in Gaussian measurement mode, M2 represents the number of compressed observations taken in Fourier measurement mode, and M1 + M2 = M; N represents the length of the original signal, i.e., the length of the sparse signal to be recovered. Represents the set of real numbers; When the number of multipath interferences is ≥5 and the hardware resource utilization rate is ≤50%, multimodal weighted fusion is adopted, and the weight α is dynamically adjusted to improve robustness. The weighting coefficients α1, α2, and α3 are dynamically adjusted based on the signal-to-noise ratio, sampling rate, and ambient noise level, where α1 + α2 + α3 = 1, and α i ≥0.2, α1, α2, and α3 are dynamically updated based on the real-time acquired signal-to-noise ratio and compression ratio, with the update period consistent with the radar signal frame period: α1=max(0.2,0.5×(SNR / SNR_max)) α2=max(0.2,0.3×(1-CR)) α3=1-α1-α2 Wherein, α1 represents the correlation threshold adjustment factor, used to control how many strongly correlated elements are retained in the support set extraction; SNR represents the real-time signal-to-noise ratio; SNR_max represents the maximum acceptable signal-to-noise ratio of the radar system; when SNR≥10dB, α1≥0.25; α2 represents the stability judgment adjustment factor, used to judge the consistency of the support set in multiple iterations; CR represents the compression ratio; when CR≤0.5, α2≥0.15; α3 represents the sparsity level estimation update adjustment factor, used to control whether the new sparsity level, i.e., the K-value estimate, is adjusted up or down, and α3≥0.2; When hardware resource utilization exceeds 70%, the Gaussian matrix Φ is dynamically selected based on the compression ratio. G Or Hartley matrix Φ H : Where M represents the number of rows in the measurement matrix Φ, and γ represents the threshold parameter that controls the selection of the measurement matrix; S3. Compressed Sensing Observation: The preprocessed radar echo signal is subsampled using the measurement matrix constructed in step S2; S4. Initial Sparsity Level Estimation and Pruning: The initial sparsity level of the signal is estimated based on the projection energy ratio, and the initial support set is screened to remove candidates with low confidence; the sparsity level is optimized and updated based on the correlation peak ratio, projection energy concentration, and energy distribution entropy. S5. Support set extraction and stability judgment: Based on the estimated sparsity level and compressed observations, the support set of the signal is extracted using an iterative algorithm, and its stability is judged. If the requirements are met, step S7 is executed; otherwise, step S6 is executed. S6. Re-estimation and update of unstable support set: The estimated value is adaptively widened or tightened according to the change of support set, the measurement weight or residual is adjusted, and the support set extraction process is re-entered, that is, the process returns to step S4. S7. Regularized Residual Feedback Update: After the support set stabilizes, a regularized residual feedback mechanism is introduced to update the signal residuals; the residuals are updated using regularized feedback, with feedback weight λ. (t) Dynamically adjusts according to an exponential decay law with the number of iterations: Calculate the basic residual Starting from the 6th iteration, a regularization feedback term is embedded in the residual update to enhance robustness. Where y represents the compressed observation vector, r (t) Let A represent the residual value after the t-th iteration, and let A represent the complete compressed observation matrix. This represents the submatrix corresponding to the current support set; This indicates that the support set s is at the t-th iteration. (t) The corresponding reconstructed signal component; s (t) Let A represent the support set after the t-th iteration; A represents the complete compressed observation matrix. λ represents the reconstructed signal at the (t-1)th iteration; (t) The exponential decay regularity coefficient representing the iteration correlation is defined as: Where λ0=0.05×σ represents the initial coefficient, and τ=5 represents the attenuation factor; Update residuals: otherwise: Determine if the residual norm satisfies the following formula: Stop iteration if the residual does not decrease as shown in the following formula: in, This represents the initial residual at the t-th iteration. This indicates the threshold for the percentage decrease in residuals; when the percentage decrease in residuals is less than... Stop the iteration early; S8. Error Judgment and Loop Feedback: Determine whether the current reconstructed signal meets the error threshold. If it does, proceed to step S9; otherwise, go to step S4. S9. Final Reconstruction and Tuning Output and Reconstruction Signal Output: Under the premise of meeting the error requirements, the final reconstruction and tuning steps are executed to generate the final estimated signal.

2. The method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario according to claim 1, characterized in that: In step S1, the target radial velocity is estimated using the radar range-Doppler spectrum during preprocessing. Let the original radar echo signal be... N represents the length of the sparse signal to be recovered. Representing the set of real numbers, in The following is represented as: Where s represents the sparse coefficient vector; The number of non-zero components in vector s is represented by K, and the sparsity level of the signal is represented by K. The sparse basis Ψ is used to transform the original signal x into an orthogonal or incomplete basis of the sparse representation s.

3. The method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario according to claim 1, characterized in that: In step S3, compressed sensing measurement is performed, and a compressed observation model is constructed as follows: y=Φx+n=ΦΨs+n Where Φ represents the selected measurement matrix, x represents the echo signal, s represents the sparse coefficient vector, and Ψ represents the sparse basis; This represents a compressed observation vector. Represents the equivalent sensing matrix. Let N represent the set of real numbers, and let n represent the length of the original signal; n represents Gaussian white noise of length M, with the same dimension as the compressed observation vector y; n ~ N0(0,σ 2 I), N0(0,σ 2 I) represents a multidimensional Gaussian distribution, where the mean vector is 0 and the covariance matrix is ​​σ. 2 I and σ represent the standard deviation calculated based on the set signal-to-noise ratio, and I represents the identity matrix with a size of M×M.

4. The method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario according to claim 1, characterized in that: In step S4, an adaptive hierarchical coupled multi-index sparsity level determination mechanism is used to perform sparsity level estimation and dynamic update based on the characteristics of the multipath environment. It includes the following steps: S4.1 Residual Initialization and Correlation Calculation: Before the 0th iteration, there is no estimate of x, therefore let... Therefore, the initial residual value is set to r. (0) =y; the residuals thereafter are expressed as Where y represents the compressed observation vector, This represents the echo signal at time t=0, i.e., the initial, unestimated signal; Calculate the correlation vector between the current residual and the columns of the measurement matrix: u = Φ T r (0) ;in, This indicates the importance of the reconstruction at each location. Let N represent the set of real numbers, and Φ represent the signal length. T This represents the transpose of the measurement matrix Φ, used for inverse projection onto the signal dimension to obtain a measure of the importance of each signal component to the current residual; S4.2 Projected Energy Calculation and Ranking: Let... If the i-th column is the measurement matrix Φ, then we have u i express The inner product of y and y, i.e., the correlation between the two; calculate the observed residual r. (t) The correlation vector u=Φ with the sensing matrix Φ T r (t) ; Calculate the projected energy e of each atom i The square of the projection value at the i-th position is Sort them in descending order: and (1) ≥e (2) ≥∙∙∙≥e (N) Among them, e (N) This represents the projected energy of the atom at position N; S4.3 Adaptive Sparsity Level Estimation Based on Energy Coverage: Calculate the cumulative energy ratio of the top k largest energies. : Among them, e (i) This represents the projected energy of the atom at position i. Based on the target characteristics in a multipath tunnel environment, an adaptive energy coverage threshold ρ is selected, and the minimum value k is found such that the first k maximum energies cover the proportion of the total energy ρ, as shown in the following formula: K init =min{k:E k ≥ρ},ρ∈[0.6,0.9] Among them, K init The initial value of the sparsity level K is represented by ρ, which represents the energy coverage threshold when the number of multipaths n mp When ρ ≤ 2, the radar signal exhibits single-target echo characteristics, and when the proportion of non-zero components is ≤ 10%, ρ is taken as 0.6-0.7; when 3 ≤ n mp When n ≤ 4, the radar signal exhibits multi-target echo characteristics, and when the proportion of non-zero components is 10%-30%, ρ is taken as 0.7-0.8; when n mp When the radar signal has a value of ≥5, it exhibits characteristics of weak targets, and when the proportion of non-zero components is >30%, ρ is taken as 0.8-0.

9. Prune according to amplitude, taking the top [value]. The index corresponding to the maximum amplitude is used as the initial support set Ω = argTopK i∈{1,...,N} (|u i |,K i ), K i This represents the value of K after the i-th iteration. An estimate representing the sparsity level; S4.4 Calculate the correlation peak ratio: Among them, u i This represents the i-th element in the correlation vector, i.e., the correlation value between the residual vector and the i-th column of the measurement matrix; u j The j-th element in the correlation vector is used to calculate the average level of the correlation amplitude of the entire signal; N represents the length of the original signal, i.e., the total number of elements in the correlation vector u; CPR represents the peak correlation ratio, which measures the sparsity. The more prominent the peak, the sparser the signal. S4.5 Calculate the energy distribution entropy: This involves converting the current residual correlation vector u... (t) Normalized to a probability distribution: Wherein, the numerator represents the current residual correlation vector u. (t) The absolute value of the i-th element reflects the magnitude of the i-th component in the residual correlation; the denominator represents the residual correlation vector u. (t) The sum of the absolute values ​​of all elements serves as a normalization function. To introduce energy concentration discrimination, the normalized entropy of the support set amplitude is defined as follows: Where ϵ represents avoiding small amounts of log0; H (t) The larger the value of H, the more uniform the residual distribution, indicating that the uncaptured signal is dispersed, increasing the sparsity level; conversely, the smaller the value of H, the more uniform the residual distribution. (t) The smaller the value, the lower the sparsity level; S4.6 Couple candidate spectra and calculate ΔK: Let the current estimated sparsity level be K. (t) The step size is ΔK (t) ,but: Where CPR(t) represents the peak correlation ratio in the t-th iteration, P target Indicates the expected peak ratio; v (t) The projected energy concentration of the candidate set is defined as: Among them, c i (t) represents the projected energy of the residual for the i-th column of atoms: The average energy is: Among them, T t τ represents the set of candidate atoms selected in the t-th iteration. v H represents the stability threshold. (t) The energy entropy of the support set, log2k (t) β1 represents the sparsity level scaling factor; β2 and β3 represent the peak ratio adjustment weight, residual projection energy distribution weight, and information entropy normalization adjustment weight, respectively. Update the sparsity level. The value of the sparsity level after the (t+1)th iteration is: Where, sign(ΔK) (t) ) represents the sign function, used to determine the direction of sparsity level adjustment: +1 indicates increasing the sparsity level, -1 indicates decreasing the sparsity level, and 0 indicates no adjustment; ΔK max This represents the maximum allowed value for adjusting the sparsity level in each iteration, set to 10%-20% of the current sparsity level. The updated sparsity level K is used as the input for the next round. The residual, peak ratio, and information entropy are repeatedly calculated and the sparsity level is updated to form an adaptive optimization closed loop, thereby realizing the dynamic adjustment of the signal sparsity level.

5. The method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario according to claim 1, characterized in that: In step S5, during the support set determination and update process, a concentration index of the projected energy distribution of the candidate set is introduced to perform a weighted determination of the support set stability and drive the closed-loop update. Specifically, this includes: Let the sliding window size be 3, and let the support set within the window be S. t-2 S t-1 and S t t represents the current iteration number, S t-2 S t-1 and S t Let J represent the support sets obtained after the (t-2), (t-1), and (t)th iterations, respectively; the basic threshold is denoted as J. thresh =0.85; Calculate Jaccard similarity: Among them, S i Let S represent the support set obtained in the i-th iteration. j This represents the support set obtained in the j-th iteration; The stability metric for the sliding window is set as the mean Jaccard similarity of all adjacent support sets within the window: The supporting set energy entropy weights are: k (t) H represents the sparsity level after the t-th iteration. (t) The energy entropy of the support set after the t-th iteration is: The residual descent rate is: r (t-1) Let r represent the residual value after the (t-1)th iteration. (t) Let represent the residual value after the t-th iteration; v(t) represents the concentration of the candidate spectrum; ω(t-1) and γ(t-1) represent the support set energy entropy weight and the residual decrease rate at the (t-1)-th iteration, respectively; If Stab(t) ≥ 0.85, the support set is determined to be stable, and step S7 is executed to enter the final reconstruction stage; if Stab(t) < 0.85, the support set is determined to be unstable, and step S6 is executed to enter the candidate expansion step.

6. The method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario according to claim 1, characterized in that: In step S6, during the candidate expansion stage, a candidate support set is constructed based on the residual projection score. ,in, The current value represents the estimate of the sparsity level. k This indicates the size of the currently estimated support set; a scores mechanism is introduced, where in the t-th iteration, a score is calculated for each column based on the residuals to evaluate the contribution of each basis vector to the residuals; combining the projection strength and matrix column norm, the candidate index with the highest score outside the current support set is selected: Where M represents the number of rows in the measurement matrix Φ, c i Represents the correlation projection coefficient; r (t-1) a represents the residual energy from the previous iteration; j,i This represents the element in the j-th row and i-th column of the measurement matrix Φ; ε represents a small positive number to prevent the denominator from being 0. Excluding elements from the current support set, calculate the score for all unselected atoms, and then select the atoms with the highest scores (L) to form a candidate set T, where L is a function of the current sparse estimate K. Therefore: Among them, K (t) S represents the sparsity level after the t-th iteration. (t-1) Let represent the support set obtained after the (t-1)th iteration, min(3K) (t) ,·) indicates the first level of restriction. This represents the second-level correction term, and incorporates historical feedback from the support set; the candidate set T is then merged with the current support set S to form an extended support set. Furthermore, an orthogonal-upper triangular decomposition is performed on the extended support set, that is, a matrix is ​​decomposed into the product of an orthogonal matrix and an upper triangular matrix, which serves as the input for subsequent least squares estimation.

7. The method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario according to claim 6, characterized in that: In step S6, a screening threshold is constructed by setting coefficients based on the median and mean absolute deviation to eliminate outliers; the screening threshold is... Retain the support set corresponding to coefficients exceeding the threshold, and update the support set. The reconstructed signal is θ represents the signal amplitude estimate obtained by fitting the observation y with the support set column matrix, MAD represents the mean absolute deviation, σ represents the standard deviation calculated based on the set signal-to-noise ratio, and i represents the index used to identify the i-th element. This represents the candidate support set, which contains a set of indices that may belong to the valid signal support set; Indicates the reconstructed signal The i-th component; τ (t) This represents the filtering threshold at the t-th iteration.

8. The method for reconstructing vehicle-mounted millimeter-wave radar signals in a tunnel scenario according to claim 1, characterized in that: In step S9, the support set S of the three most recent iterations is recorded. (t) S (t-1) and S (t-2) By calculating the average Jaccard similarity, it can be determined whether the support sets have reached a stable state. The Jaccard similarity is obtained by calculating the ratio of the intersection size to the union size of the two sets. Where T represents the total number of iterations, t represents the current iteration index, and S (t) Let S represent the support set for the t-th iteration. (t-1) Let S represent the support set for the (t-1)th iteration. (t-2) Let represent the support set for the (t-2)th iteration; When J avg When the value is less than 0.85, it is determined that the support set has not yet reached a stable state. Final support set optimization is then performed, dynamically adjusting the support set size based on the signal amplitude, and resolving the least squares problem to obtain the final reconstruction result. in, Indicates the final reconstructed signal, A S Let u represent the column submatrix corresponding to the support set S in the measurement matrix, s represent the sparse coefficient vector, and Ψ represent the sparse basis. Represents the column submatrix A S The transpose of , where y represents the observation vector. Denotes the optimal solution on the support set; Solving for coefficients on the support set S using the least squares method Then embed it into a sparse vector x of length N. * In the middle, the coefficients of the support set outside positions are set to zero, resulting in the final complete sparse coefficient vector x. * Based on this, the final reconstructed signal can be recovered. : 。

Images