Radar wave intensity measurement method and system based on multistage filtering amplification

Abstract

The invention provides a radar wave intensity measurement method and system based on multistage filtering amplification, and the method comprises the steps: carrying out the preprocessing of a radar echo signal, and obtaining a preprocessing signal through time domain wavelet packet decomposition and spatial domain wave beam forming collaborative optimization; performing intelligent dynamic gain reconstruction on the preprocessed signal to obtain a gain adjustment signal; performing nonlinear harmonic tracking compensation on the gain adjustment signal to obtain a compensation calibration signal; and performing dynamic threshold intensity extraction on the compensation calibration signal to obtain a final measurement result. According to the method, an environment factor-attenuation compensation mapping relation is established through a support vector regression model, and a dynamic environment compensation coefficient is generated; in combination with wavelet packet multi-scale energy integration and signal-to-noise ratio dynamic weighting, space attenuation is corrected in nonlinear harmonic tracking compensation, and an adaptive threshold and energy estimation are optimized in dynamic threshold intensity extraction; and finally, through Kalman filtering post-processing and spectrum purity monitoring closed-loop verification, the system can perform high-precision measurement in a dynamic environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar wave intensity measurement, in particular to a radar wave intensity measurement method and system based on multi-stage filtering amplification. BACKGROUND

[0002] Traditional radar wave intensity measurement techniques are mostly based on single-domain signal processing framework, typical methods including fixed threshold detection, linear filter gain control and static environment compensation. For example, early techniques often use band-pass filters to suppress out-of-band noise, adjust the receiver gain through manual or simple adaptive algorithms, and implement intensity estimation with an atmospheric attenuation correction model based on empirical formula. Such methods can achieve basic measurement in stable environment scenarios, but due to the lack of spatio-temporal joint optimization capability, they are prone to problems such as limited signal-to-noise ratio improvement and target signal aliasing in complex electromagnetic environments (such as multipath interference and dynamic clutter). In addition, traditional harmonic compensation relies on linear filters, which have limited effect on nonlinear distortion suppression; environmental factor compensation often uses offline calibrated parameters, which are difficult to adapt to real-time weather conditions, resulting in large fluctuations in measurement accuracy. SUMMARY

[0003] The present application aims to at least solve the technical problems in the prior art of large fluctuations in measurement accuracy, and particularly innovatively proposes a radar wave intensity measurement method and system based on multi-stage filtering amplification.

[0004] In order to achieve the above-mentioned purpose of the present application, the present application provides a radar wave intensity measurement method based on multi-stage filtering amplification, the method comprising: S1, performing spatio-temporal joint adaptive filtering preprocessing on the radar echo signal, improving the signal-to-noise ratio and suppressing multipath interference through the cooperative optimization of time-domain wavelet packet decomposition and spatial beam forming, and obtaining the preprocessed radar echo signal; S2, performing intelligent dynamic gain reconstruction on the preprocessed radar echo signal, adjusting the gain in combination with feedforward prediction and feedback control, and obtaining the gain-adjusted radar echo signal; S3, performing nonlinear harmonic tracking compensation on the gain-adjusted radar echo signal, reducing harmonic distortion through high-order Volterra modeling and reverse compensation, and integrating laser ranging calibration to realize dynamic correction of spatial attenuation, and obtaining the compensated and calibrated radar echo signal; S4, performing dynamic threshold intensity extraction on the compensated and calibrated radar echo signal, calculating the radar wave energy estimation value through adaptive threshold setting and multi-scale energy integration, and realizing measurement accuracy optimization by fusing environmental factor compensation, and obtaining the final radar wave intensity measurement result.

[0005] In another aspect, the present application also provides a radar wave intensity measurement system based on multi-stage filter amplification, the system comprising a processor and a memory for storing processor executable instructions; Wherein, the processor is configured to implement the radar wave intensity measurement method based on multi-stage filter amplification when executing the executable instructions.

[0006] The present application has the following advantages: The present application effectively solves the measurement accuracy fluctuation problem caused by real-time meteorological condition changes in the prior art through a multi-stage filter amplification architecture and a dynamic environment compensation mechanism. Specifically, laser ranging calibration and real-time data of a multi-parameter environment sensor (temperature, humidity, air pressure, and rain and fog concentration) are integrated. An environment factor-attenuation compensation mapping relationship is established through a support vector regression model to generate a dynamic environment compensation coefficient. In the nonlinear harmonic tracking compensation in step S3, real-time correction of spatial attenuation is realized through a wavelet packet multi-scale energy integration and a signal-to-noise ratio dynamic weighting technology. In the dynamic threshold intensity extraction in step S4, the adaptive threshold and energy estimation are cooperatively optimized. Finally, through Kalman filter post-processing and spectrum purity monitoring closed-loop verification, the system can still maintain high-precision measurement in dynamic environments such as rain and fog and temperature gradient. Compared with the traditional static compensation method, the precision is greatly improved, the anti-interference ability is greatly improved, and the environmental adaptability and result reliability of radar wave intensity measurement are significantly enhanced.

[0007] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0008] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, including the accompanying drawings, wherein: Figure 1 is a flowchart of a radar wave intensity measurement method based on multi-stage filter amplification. DETAILED DESCRIPTION

[0009] Embodiments of the present application are described in detail below with reference to the attached drawings, wherein the same or similar components are denoted by the same or similar reference numerals throughout. The embodiments described below with reference to the attached drawings are exemplary and are intended only for the purpose of explaining the present application, and should not be construed as limiting the present application.

[0010] Embodiment 1 As shown in Figure 1 A radar wave intensity measurement method based on multi-stage filter amplification, the method comprising: S1, the radar echo signal is preprocessed by spatio-temporal domain joint adaptive filtering, the signal-to-noise ratio is improved and the multipath interference is suppressed through the cooperative optimization of time domain wavelet packet decomposition and space domain beam forming, and the preprocessed radar echo signal is obtained; S2, the radar echo signal after preprocessing is intelligently dynamically gain reconstructed, the gain is adjusted in combination with feedforward prediction and feedback control, and the radar echo signal after gain adjustment is obtained; S3, the radar echo signal after gain adjustment is compensated for nonlinear harmonic tracking, the harmonic distortion is reduced through high-order Volterra modeling and reverse compensation, and spatial attenuation dynamic correction is realized by integrating laser ranging calibration, and the radar echo signal after compensation and calibration is obtained; S4, the radar echo signal after compensation and calibration is dynamically threshold intensity extracted, radar wave energy estimation value is calculated through adaptive threshold setting and multi-scale energy integration, and environmental factor compensation is fused to realize the optimization of measurement accuracy, and the final radar wave intensity measurement result is obtained.

[0011] The principle of the radar wave intensity measurement method based on multi-stage filtering amplification in the embodiment is as follows: first, high-frequency noise and pulse interference are filtered out through time-domain wavelet packet decomposition, and at the same time, the spatial beam forming technology is combined to suppress the multipath effect and sidelobe interference, so as to realize the cooperative improvement of the signal-to-noise ratio in the time and space domains; then, according to the signal instantaneous power spectral density and the leading pulse envelope characteristics, the optimal gain requirement is predicted by using a feedforward neural network, and at the same time, the overload distortion risk is monitored in real time through a feedback loop, so that the gain parameters of the main amplifier and the variable gain amplifier are dynamically reconstructed, and it is ensured that the signal amplitude is in the optimal linear amplification interval; then, for the nonlinear harmonic distortion introduced by the power amplifier, a high-order Volterra kernel function is used for modeling, the harmonic distortion characteristics are extracted through third-order nonlinear system identification, and an initial harmonic compensation model is generated; based on the initial harmonic compensation model, the inverse compensation coefficient is optimized by using the least mean square error criterion, and the Volterra kernel parameters are dynamically adjusted to suppress the harmonic components; at the same time, the spatial distance data are collected in real time, and the distance information is converted into spatial attenuation coefficients through an attenuation factor calculation module; the inverse compensation coefficient and the spatial attenuation coefficient are fused, and a comprehensive compensation parameter matrix is generated by using a weighted fusion algorithm; the signal is compensated and attenuated by using the comprehensive compensation parameter matrix, and the compensation effect is verified by spectral purity monitoring; when the harmonic residual is lower than a preset threshold, the radar echo signal after compensation and calibration is output; finally, based on the radar echo signal after compensation and calibration, an Otsu adaptive threshold algorithm and a noise power estimation joint optimization strategy are used, the signal amplitude histogram distribution is counted through a sliding window, the optimal threshold of the maximum inter-class variance is calculated, and the noise power is corrected to generate a dynamic adaptive threshold; based on the dynamic adaptive threshold, the radar echo signal after compensation and calibration is subjected to multi-scale decomposition by using a wavelet packet, z sub-bands are generated through l layer decomposition by using a Sym8 wavelet basis, the energy of each sub-band is calculated, and the signal total energy estimation value is obtained based on the dynamic weighted integration of the signal-to-noise ratio; then, the temperature, humidity, air pressure and rain and fog concentration data collected in real time by the multi-parameter environment sensor are fused, the environment factor-attenuation compensation mapping relationship is established by using a support vector regression model, and an environment compensation coefficient is generated; based on the environment compensation coefficient, the signal total energy estimation value is subjected to nonlinear mapping conversion and error propagation correction, the error contribution of each link is quantified through Monte Carlo simulation, and a Kalman filter is used for post-processing optimization to generate a preliminary intensity measurement value; finally, the compensation effect is verified by spectral purity index monitoring, and when the target frequency point energy proportion is lower than r%, the parameter recalibration is triggered, otherwise the radar wave intensity measurement result meeting the accuracy requirement is output.

[0012] As an optional embodiment of the present application, optionally, obtaining the preprocessed radar echo signal in step S1 comprises: S101, performing analog-to-digital conversion on the radar echo signal to obtain a digitized radar echo signal; Detailed description is needed in step S101, the analog-to-digital conversion process is realized by high-speed analog-to-digital converter (ADC), specifically, 16-bit resolution, sampling rate of 200MHz ADC chip is selected, model is AD9265, to ensure that the radar echo signal bandwidth range (typical for 0-80MHz) is realized without aliasing sampling. Before sampling, pre-processing is required through the pre-anti-aliasing filter, the filter is an 8-order chebyshev low-pass filter, the cutoff frequency is set to 1.1 times of the signal bandwidth (i.e. 88MHz), the attenuation slope is ≥60dB / octave, to suppress high-frequency noise and out-of-band interference; The sampling clock is generated by the radar system main clock synchronization, the phase jitter is controlled within 0.5ps, to ensure the sampling time accuracy. In the process of digitization, the input signal amplitude needs to be normalized to 70%-90% of the full scale of ADC, which is realized by the pre-adjustment of automatic gain control (AGC) loop, to avoid quantization saturation or signal-to-noise ratio loss; The final output of the digitized radar echo signal is stored in the FIFO buffer in the form of 16-bit parallel data stream.

[0013] S102, based on the digitized radar echo signal, wavelet packet decomposition and soft threshold filtering are used for time domain noise suppression, and time domain denoising radar signal is obtained; Detailed description is needed in step S102, wavelet packet decomposition adopts Daubechies (db) series wavelet basis, specifically, db8 wavelet is selected for 4-layer decomposition, generating 16 sub-bands. In the decomposition process, the signal is decomposed losslessly by using the orthogonal mirror filter bank (QMF), to ensure the energy conservation of each frequency band. The mean value of the absolute value of the wavelet coefficient of each sub-band is calculated as the noise threshold estimation value, and the soft threshold filtering algorithm is used for shrinkage processing of the wavelet coefficient, and the threshold formula is: T equals to sigma times square root of (2lnN), wherein sigma is the noise standard deviation of the sub-band, and N is the coefficient length of the frequency band. The soft threshold processing formula is: w'=sign(w)(|w|-T)+, wherein w is the original wavelet coefficient, w' is the processed coefficient, the coefficient is zero when |w|≤T, and linear shrinkage is performed when |w|>T. The processed coefficient is recovered to the time domain signal by the wavelet packet reconstruction algorithm, the reconstruction process adopts Mallat algorithm for progressive up-sampling and convolution operation, and finally the time domain denoising radar signal is generated. Through multi-scale decomposition and adaptive threshold processing, the step effectively suppresses the impulse noise and Gaussian white noise, while retaining the mutation characteristics of the signal.

[0014] S103, based on the time domain denoising radar signal, phased array beam forming and DOA estimation are used for spatial domain interference suppression, and spatial domain filtered radar signal is obtained; ULA structure, and the inter-element distance is set to half wavelength (λ / 2) to avoid grating lobe effect. The number of elements is selected as 16 channels according to the actual scene requirement. First, the radar signal after time domain noise reduction is shifted to baseband through digital down conversion (DDC), and the sampling rate is reduced to 10MHz to reduce the complexity of subsequent processing. Then, Capon minimum variance distortion response (MVDR) algorithm is used for beam forming. This algorithm constructs a spatial covariance matrix and solves the optimal weight vector to make the beam main lobe point to the target direction while suppressing the sidelobe interference. In the specific implementation, the spatial covariance matrix is estimated by sliding window statistics method, and the window length is set to 512 sampling points to balance the estimation accuracy and computational efficiency. The optimal weight vector is calculated by Lagrange multiplier method to ensure that the target signal gain is maintained while the interference is suppressed. At the same time, multiple signal classification (MUSIC) algorithm is used for direction of arrival (DOA) estimation. The signal subspace and noise subspace are separated by eigenvalue decomposition, and the direction of the interference source is located by spatial spectrum peak search. The DOA estimation result is fed back to the beam forming module to dynamically adjust the null position to enhance the spatial filtering effect. Finally, the signals of each element are weighted and superimposed through the beam synthesis network to generate the spatially filtered radar signal. This signal effectively suppresses multipath interference and sidelobe clutter in the spatial dimension while preserving the phase information of the target echo.

[0015] S104, the spatially filtered radar signal and the radar signal after time domain noise reduction are jointly optimized by using ADMM algorithm to obtain the preprocessed radar echo signal.

[0016] In step S104, it is necessary to explain in detail that the existing Alternating Direction Multiplier Method (ADMM), as a distributed optimization algorithm, decomposes a large-scale global optimization problem into multiple easily tractable local subproblems, and achieves global convergence through alternating iterative solutions. For the joint optimization scenario of the spatially filtered radar signal and the temporally denoised radar signal, an objective function is first constructed. This function consists of two terms: the first is the mean square error between the spatial signal and the target beam pattern, and the second is the fidelity constraint between the temporal signal and the original signal. These two terms are coupled through Lagrange multipliers. In the specific implementation, the optimization variables are decomposed into two sets: a spatial weight vector and temporal filter coefficients. These are updated alternately in each iteration: first, the temporal filter coefficients are fixed, and the optimal solution of the spatial weight vector is solved using the Capon algorithm to minimize the beamforming error; then, the spatial weight vector is fixed again, and the temporal filter coefficients are updated using the least squares method to ensure the signal fidelity constraint. After each iteration, the Lagrange multipliers are adjusted through residual feedback to accelerate the convergence process. To improve computational efficiency, a parallel computing architecture is adopted, allocating the spatial optimization task of the 16-channel array to an FPGA accelerator card, while the temporal filtering is handled by a DSP core. The two exchange data via a high-speed bus. The convergence condition is set to the difference between the objective function values ​​of two adjacent iterations being less than 10^-6 or the number of iterations reaching a preset upper limit (typically 100). The final preprocessed radar echo signal output demonstrates its spatiotemporal joint optimization effect through dual metrics: signal-to-noise ratio improvement (SNRI) and interference rejection ratio (ISR). In typical scenarios, the SNRI improvement exceeds 12dB, and the ISR is better than 35dB, while the algorithm's real-time performance meets the radar system frame period requirement (≤5ms). This step, through spatiotemporal collaborative processing, effectively overcomes the limitations of single-domain filtering.

[0017] As an optional embodiment of the present invention, optionally, obtaining the gain-adjusted radar echo signal in step S2 includes: S201. Establish a feedforward prediction model based on the historical radar echo signal intensity distribution, and use the Kalman filter in the feedforward prediction model to estimate the current signal gain requirement and generate preliminary gain control parameters. S202. Real-time acquisition of the time-spectrum characteristics of the preprocessed radar echo signal, and dynamic correction of the feedforward prediction deviation based on the least mean square algorithm to generate feedback compensation parameters. The expression for the least mean square algorithm is: in, Indicates the first Constantly provide feedback on the incremental compensation parameters. Indicates the step size factor. This represents the feedforward prediction bias signal. represents a time-frequency spectrum feature vector, represents a first feedback compensation parameter cumulative value at the kth moment; In step S202, the construction of the time-frequency spectrum feature vector is realized by short-time Fourier transform (STFT), and the preprocessed radar echo signal is divided into overlapping time window segments (the window length is set to 256 points, and the overlap rate is 75%). The Fourier transform is performed on each segment to obtain the frequency spectrum distribution, and then the key feature parameters are extracted, including the main frequency energy proportion, the spectrum centroid, the spectrum bandwidth, and the spectrum entropy. The feedforward prediction error signal is defined as the difference between the preliminary gain control parameter output by the feedforward prediction model and the actual signal amplitude normalized value, wherein the actual signal amplitude is calculated by the sliding window statistical method (the window length is set to 1024 points) and is normalized to the 0-1 interval. The step factor μ adopts an adaptive adjustment strategy, and the initial value is set to 0.01. When the feedback compensation parameter increment direction is consistent for 5 consecutive iterations, μ increases by 0.1 times the step (the maximum is not more than 0.1); when the direction is reversed, μ decreases by 0.5 times the step (the minimum is not less than 0.001), to balance the convergence speed and stability. The update formula of the feedback compensation parameter cumulative value is: W(k)=W(k-1)+ΔW(k), where W(k) is the cumulative value at the kth moment, and the initial value is set to a zero vector. This step corrects the feedforward prediction error in real time, so that the gain control parameter can dynamically track the signal amplitude change, avoiding the problem of over-gain or under-gain caused by the error of the feedforward model. The final generated feedback compensation parameter is fused with the preliminary gain control parameter according to the weight coefficient (the front weight is set to 0.7, and the feedback weight is set to 0.3), to generate a fused gain control parameter, which is used to drive the cascade gain adjustment of the main amplifier and the variable gain amplifier (VGA). The main amplifier adopts a high linearity power operational amplifier (model THS4509), and its gain is fixed at 20dB; the VGA selects AD8336 chip, which supports a continuous adjustable gain range of-40dB to +40dB, and the two are connected through an impedance matching network to ensure that the total gain range covers-20dB to +60dB, meeting the linear amplification needs of radar echo signals of different intensities. The gain adjustment process is realized by a 16-bit parallel control word generated by a digital signal processor (DSP), and the control word is converted by a DAC and input to the gain control interface of the VGA, and the update period is set to 10μs to match the frame period requirement of the radar system. The radar echo signal amplitude after gain adjustment is verified by an automatic test equipment (ATE), and the linearity error is ≤±0.5dB, and the gain flatness is better than ±0.3dB within the bandwidth of 0-80MHz, ensuring that the signal amplitude is within the best input range (normalized amplitude 0.3-0.7) of the subsequent nonlinear harmonic tracking compensation module.

[0018] S203, fuse the preliminary gain control parameter and the feedback compensation parameter to generate a gain reconstruction coefficient matrix through a fuzzy PID controller; In step S203, the fuzzy PID controller takes the preliminary gain control parameter and the feedback compensation parameter as double input variables, first performs fuzzy processing through a membership function to map the accurate numerical value to a fuzzy set (such as "low", "medium", "high" three linguistic variables). The membership function adopts a combination of triangle and trapezoid, for example, for the preliminary gain control parameter, the domain is set as [0, 1], and three fuzzy subsets are divided: low (0-0.3), medium (0.2-0.7), and high (0.6-1), and the membership degrees of each subset are calculated through linear interpolation; the membership function of the feedback compensation parameter is designed similarly, but the domain is dynamically adjusted according to the actual deviation range (such as [-0.5, 0.5]). Then reasoning is performed based on a preset fuzzy rule base, which contains 27 rules (3x3x3), for example, "if the preliminary gain is high and the feedback compensation is positive, the gain adjustment direction is slightly reduced", and the rule weights are all set to 1. The reasoning result is de-fuzzied through the center of gravity method to generate the initial gain reconstruction coefficient (the range is set as [0.8, 1.2]). To further improve the dynamic response performance, the fuzzy output is combined with the classic PID control: the fuzzy output is used as a real-time correction factor (Kp_fuzzy) of the PID proportional coefficient, and the fixed integral coefficient (Ki=0.05) and the derivative coefficient (Kd=0.01) are used to form an improved PID controller. The finally generated gain reconstruction coefficient matrix is a 3x3 diagonal matrix, the diagonal elements are filled by the fuzzy PID output, and the non-diagonal elements are set to 0, which is realized through the matrix operation unit of the DSP, and the operation period is ≤500ns, which meets the real-time requirement.

[0019] S204, based on the gain reconstruction coefficient matrix, the preprocessed radar echo signal is segmented and linearly amplified to obtain the preliminary gain adjusted radar echo signal, and an anti-saturation truncation strategy is used to avoid signal overload; Need to be described in detail in step S204 is that the piecewise linear amplification process divides the preprocessed radar echo signal into multiple subintervals (for example, divided into low, medium and high three intervals, and the amplitude threshold is dynamically set according to the actual signal distribution) according to the amplitude range, and each subinterval corresponds to a diagonal element in the gain reconstruction coefficient matrix as a local gain value. In specific implementation, first, the interval selection signal is generated by comparing the input signal amplitude belonging to the interval through the comparator array; then the corresponding local gain value is extracted from the gain reconstruction coefficient matrix according to the interval selection signal, and the multiplication operation is performed with the input signal to realize the segmented amplification. In order to avoid the gain mutation at the segmentation point causing signal distortion, a linear transition band (the transition band width is set to 10% of the interval width) is used at the interval boundary, and the gain value in the transition band is calculated by linear interpolation to ensure the continuity of the gain change. The anti-saturation truncation strategy is realized by monitoring the signal amplitude after amplification in real time: when the signal amplitude exceeds 90% of the ADC full scale, the limiting protection is triggered immediately, the exceeding part is truncated and marked as saturation state; At the same time, the saturation signal characteristics (such as saturation duration, amplitude overrun value) are transmitted to the fuzzy PID controller through the feedback channel, and the gain reconstruction coefficient of the subsequent interval is dynamically adjusted (for example, the gain value of the high amplitude interval is reduced), to prevent continuous saturation. In addition, for the problem of spectrum leakage caused by truncation, the existing frequency domain zero padding interpolation algorithm is used in the signal reconstruction stage: after Fourier transform of the truncated signal, zero is filled to twice the original signal length in the high frequency component area, and then inverse Fourier transform is performed to recover the time domain signal, which effectively suppresses the harmonic interference caused by truncation. The final output is the preliminary gain adjusted radar echo signal.

[0020] S205, verify the gain adjustment effect of the preliminary gain adjusted radar echo signal through the sliding window energy monitoring, trigger parameter re-calibration when the signal-to-noise ratio improvement rate in the window is lower than the preset threshold, and obtain the gain adjusted radar echo signal.

[0021] In step S205, it needs to be described in detail that the sliding window energy monitoring adopts a rectangular window structure, and the window length is dynamically set according to the radar pulse repetition period (PRI) (a typical value is set to 5 PRI time lengths), and the full signal section is covered and monitored through sliding steps (the step size is set to 1 PRI). The signal energy in the window is calculated by the square sum method: E(k)=∑[x²(n)] (n from k to k+N-1), where x(n) is the signal amplitude of the n-th sampling point in the window, and N is the window length; the noise energy is obtained by statistical average of adjacent blank windows (target echo period), and the signal-to-noise ratio (SNR) is defined as the ratio of signal energy to noise energy. The signal-to-noise ratio improvement rate ΔSNR(k) is calculated by the difference between the current window SNR(k) and the previous window SNR(k-1): ΔSNR(k)=SNR(k)-SNR(k-1), and its physical meaning is the improvement degree of signal quality after gain adjustment. The preset threshold is set according to the actual application scene (for example, set to 3dB), when the ΔSNR(k) of the continuous 3 sliding windows is lower than the threshold, it is determined that the gain adjustment effect is deteriorated, and the parameter recalibration process is triggered. The parameter recalibration is divided into two steps: first, the time-frequency spectrum characteristics (including the main frequency energy proportion, the spectrum centroid and other key parameters) of the current radar echo signal are reacquired by the ATE device, and the matching degree analysis (using cosine similarity algorithm, and the matching threshold is set to 0.85) is carried out with the typical signal model in the historical database; if the matching is successful, the preset gain control parameter (such as the conservative gain mode in the strong clutter environment) of the corresponding scene is directly called; if the matching fails, the adaptive learning mode is started: the current signal characteristics are input into the feedforward prediction model (the Kalman filter parameters are reset to the initial state), and the gain control parameters are regenerated through 50 times of iteration training, and the new parameters and the corresponding signal characteristics are stored in the historical database to expand the model adaptability. The gain control parameters after recalibration are regenerated as 16-bit parallel control words by DSP, and are input into the gain control interface of VGA after DAC conversion, and the update period is shortened to 5us (originally 10us) to accelerate convergence. The radar echo signal after gain adjustment is finally output.

[0022] As an optional embodiment of the application, optionally, generating the preliminary gain control parameters in step S201 comprises: S2011, collecting historical radar echo signal intensity data in a preset time window, removing abnormal values by sliding average filtering, and generating a standardized intensity sequence; The length of the preset time window is set according to the fluctuation period of historical data of the radar system, and a typical value is 10 pulse repetition periods. The window length of the moving average filter is set to 5 sampling points, and the arithmetic mean value of the data in the window is calculated to replace the center point value, effectively suppressing pulse noise interference. The standardization intensity sequence generation process is divided into two steps: first, calculate the mean and standard deviation of the intensity sequence, and then perform normalization on each sampling point. To avoid the influence of extreme values, the normalized data is limited to the [-3, 3] interval (values outside the range are forced to be truncated to the boundary value), and finally a standardized intensity sequence of length N is generated, which is used as the input observation value of the Kalman filter.

[0023] S2012, based on the standardized intensity sequence, a Gaussian mixture model is used to fit the intensity distribution characteristics, and the mean, variance and skewness parameters are extracted as prediction inputs; In step S2012, the Gaussian mixture model fits complex data distribution through linear combination of multiple Gaussian distributions, and its probability density function is expressed as the weighted sum of multiple Gaussian distributions. In specific implementation, according to the distribution characteristics of the standardized intensity sequence, the number of components of the Gaussian mixture model is set (usually set to 3-5), each component corresponds to a Gaussian distribution, and has its own mean, variance and weight coefficient. The existing expectation maximization (EM) algorithm is used to iteratively optimize the model parameters, so that the likelihood function of the model output and the observation data is maximized. In the parameter extraction stage, the mean and variance parameters of each component are directly read from the optimized Gaussian mixture model, and the third central moment (skewness) of the entire mixture model is calculated as a characteristic parameter describing the asymmetry of the intensity distribution. These parameters together constitute the prediction input vector, which is used in the subsequent feedforward prediction model. To improve the robustness of the model, the extracted parameters are preprocessed: the mean and variance parameters are smoothed by the sliding window statistical method (the window length is set to 10 sampling points), and the skewness parameter is limited by the absolute value (the range is set to [-1, 1]) to avoid abnormal value interference. The finally generated prediction input vector contains three-dimensional feature information, which can fully reflect the intensity distribution law of the historical radar echo signal.

[0024] S2013, the prediction input is introduced into the existing Kalman filter state equation (the specific calculation expression is as follows step S4043), and the gain requirement estimate value is updated in combination with the system noise covariance matrix; It should be noted that in step S2013, the Kalman filter is a recursive optimal estimation method, which can estimate the system state in real time by using the dynamic model of the system and the observation data. In the specific implementation, first, the state equation is established according to the characteristics of the radar system, which describes the change rule of the gain demand estimation value with time, and the system noise covariance matrix reflects the uncertainty of the system state. The predicted input vector generated in step S2012 is introduced into the state equation as one of the input variables of the equation. At the same time, combined with the pre-set system noise covariance matrix, the value of the matrix is adjusted according to the noise characteristics of the actual system to accurately reflect the fluctuation of the system state. Through the recursive calculation process of the Kalman filter, including the prediction step and the update step, the gain demand estimation value is continuously corrected. In the prediction step, the gain demand estimation value at the current time is predicted according to the state equation; in the update step, the predicted value is corrected using the observation data (i.e. the standardized intensity sequence) to obtain a more accurate gain demand estimation value. The finally generated gain demand estimation value can reflect the change of the radar echo signal demand for gain in real time.

[0025] S2014, according to the gain demand estimation value, a preliminary gain control parameter matrix is generated by using the existing linear interpolation algorithm.

[0026] It should be noted that in step S2014, the linear interpolation algorithm is a method of interpolating between known data points to estimate intermediate values. In the specific implementation, first, the interpolation nodes are set according to the range of the gain demand estimation value, for example, the minimum and maximum values of the gain demand estimation value are taken as the starting point and the ending point of the interpolation, and several interpolation nodes are uniformly divided between them. Subsequently, according to these interpolation nodes and the corresponding gain control parameter values (which can be set according to actual demand or experience), a linear interpolation table is constructed. When a new gain demand estimation value is obtained, the interval where it is located is determined by looking up the interpolation table, and the corresponding preliminary gain control parameter value is calculated by using the linear interpolation formula. Finally, the calculated preliminary gain control parameter value is filled into the matrix to generate a preliminary gain control parameter matrix, which can reflect the change of the control parameter under different gain demands.

[0027] As an optional embodiment of the present application, optionally, the radar echo signal compensated and calibrated in step S3 comprises: S301, the gain-adjusted radar echo signal is modeled by a high-order Volterra kernel function, and the harmonic distortion feature is extracted by a third-order nonlinear system identification to generate an initial harmonic compensation model; The step S301 needs to be described in detail is that the existing high-order Volterra kernel function modeling is an effective method for describing the input-output relationship of a nonlinear system, and is particularly suitable for processing radar echo signals with nonlinear characteristics. In specific implementation, first, the radar echo signal after gain adjustment is sampled to obtain discrete input-output data sequences. Then, a third-order Volterra series model is constructed using these data sequences, which comprehensively describes the nonlinear characteristics of the system by considering the first-order, second-order and third-order interactions of the input signal. In the modeling process, the least squares method is used to estimate the coefficients of the Volterra kernel function, so that the sum of the squared errors between the model output and the actual signal is minimized. Through the identification of the third-order nonlinear system, the harmonic distortion characteristics in the radar echo signal can be accurately extracted, including the amplitude, phase and frequency offset of each harmonic. Based on these characteristic parameters, an initial harmonic compensation model is generated, which takes the input signal as the independent variable and outputs the corresponding harmonic compensation amount for subsequent signal correction processing. To improve the model accuracy, the initial model is iteratively optimized: by comparing the residual error between the compensated signal and the ideal signal, the Volterra kernel function coefficients are dynamically adjusted until the residual error meets the preset convergence condition (such as the residual error energy being lower than the noise floor by 3 dB). The final generated initial harmonic compensation model can accurately depict the nonlinear distortion characteristics of the radar echo signal.

[0028] S302, based on the initial harmonic compensation model, the least mean square error criterion is used to optimize the inverse compensation coefficients, and the Volterra kernel parameters are dynamically adjusted to suppress the harmonic components; The step S302 needs to be described in detail is that the least mean square error criterion is a commonly used optimization method, and its goal is to minimize the mean square error between the compensated signal and the ideal signal. In specific implementation, first, the harmonic compensation amount output by the initial harmonic compensation model is applied to the radar echo signal after gain adjustment to obtain a preliminary compensated signal. Then, the mean square error between the preliminary compensated signal and the ideal signal is calculated as the objective function of optimization. In order to dynamically adjust the Volterra kernel parameters to suppress the harmonic components, the gradient descent method is used to iteratively optimize the objective function. In each iteration, the gradient of the objective function with respect to the Volterra kernel parameters is calculated, and the kernel parameters are updated in the opposite direction of the gradient, so that the objective function value gradually decreases. Through multiple iterations, the Volterra kernel parameters are continuously adjusted until the objective function reaches the preset convergence condition (such as the mean square error being lower than a certain threshold). Finally, the optimized Volterra kernel parameters can more accurately describe the nonlinear characteristics of the radar echo signal, thereby effectively suppressing the harmonic components and improving the signal quality.

[0029] S303, real-time acquisition of spatial distance data, conversion of distance information to spatial attenuation coefficients through an attenuation factor calculation module; The expression of the attenuation factor calculation module in step S303 is: ; wherein, represents the spatial attenuation coefficient, represents the real-time distance of laser ranging, represents the transmitting antenna gain, represents the receiving antenna gain, represents the radar wavelength, represents the atmospheric absorption coefficient, represents the radar beam width.

[0030] In step S303, it needs to be explained in detail that the real-time collection of spatial distance data is realized by a laser range finder, which continuously measures the straight-line distance between the target and the radar at a sampling rate of 10 kHz, and the measurement accuracy is better than ±0.1 m. The collected distance data is first subjected to outlier rejection processing, and a median filtering algorithm is used to sort the continuous 5 sampling points, and the middle value is taken as the effective distance value to avoid fly point interference. Then the effective distance value is input into the attenuation factor calculation module, which calculates the spatial attenuation coefficient according to the preset radar system parameters and the real-time distance. In the calculation process, special consideration is given to the atmospheric absorption effect and the inverse square law of distance, wherein the atmospheric absorption term adopts an exponential decay model, and the distance term adopts a logarithmic decay model, and the two are combined through a weighting coefficient (the weight ratio is set to 3:7) to obtain the final spatial attenuation coefficient.

[0031] S304, fuse the reverse compensation coefficient and the spatial attenuation coefficient, and generate a comprehensive compensation parameter matrix using a weighted fusion algorithm; In step S304, it needs to be explained in detail that the core idea of the weighted fusion algorithm is to give different weights to the reverse compensation coefficient and the spatial attenuation coefficient according to their different influence degrees on the radar echo signal, and then generate a comprehensive compensation parameter matrix. In specific implementation, first, the weight distribution of the reverse compensation coefficient and the spatial attenuation coefficient is determined, and this distribution is set according to the emphasis degree of harmonic suppression and spatial attenuation compensation in the actual application scene, for example, in some scenes where the signal purity requirement is extremely high, the weight of the reverse compensation coefficient can be set to 0.7, and the weight of the spatial attenuation coefficient is set to 0.3. Then, the reverse compensation coefficient and the spatial attenuation coefficient are multiplied by their respective weights to obtain the weighted reverse compensation coefficient and the weighted spatial attenuation coefficient. Next, the two weighted coefficients are added to obtain the comprehensive compensation parameter. Finally, according to the different working modes and signal channel numbers of the radar system, the comprehensive compensation parameter is filled into the matrix to generate a comprehensive compensation parameter matrix. This matrix can consider the influence of harmonic distortion and spatial attenuation on the radar echo signal at the same time.

[0032] S305, the signal is compensated and corrected by using the comprehensive compensation parameter matrix, the compensation effect is verified by spectrum purity monitoring, and the radar echo signal after compensation and calibration is output when the harmonic residual is lower than the preset threshold.

[0033] In step S305, the detailed process of using the comprehensive compensation parameter matrix to compensate and correct the radar echo signal after preliminary compensation is as follows: each element in the comprehensive compensation parameter matrix is multiplied by the corresponding sampling point of the signal to realize joint adjustment of the amplitude and phase of the signal. The nonlinear compensation part mainly corrects the harmonic distortion in reverse, and the attenuation correction part dynamically compensates the signal strength according to the spatial distance. The spectrum purity monitoring converts the time domain signal into frequency domain representation through fast Fourier transform, and then calculates the power ratio of each harmonic component to the fundamental component. When the power ratio of all harmonic residuals is lower than the preset threshold (typical value is-60dBc), it is determined that the compensation effect meets the standard and the final calibrated radar echo signal is output; if there is a harmonic residual exceeding the standard, the iteration optimization process is started, the comprehensive compensation parameter matrix is regenerated by adjusting the weight distribution of the reverse compensation coefficient and the spatial attenuation coefficient, until the spectrum purity requirement is met. This process realizes real-time processing through a hardware acceleration module (FPGA), and the single compensation period is controlled within 20μs, ensuring the measurement accuracy of the radar system in high-speed motion scenes.

[0034] As an optional embodiment of the present application, optionally, obtaining the final radar wave intensity measurement result in step S4 includes: S401, based on the compensated and calibrated radar echo signal, an Otsu adaptive threshold algorithm and a noise power estimation joint optimization strategy are adopted, the signal amplitude histogram distribution is counted through a sliding window, the optimal threshold value of maximizing the inter-class variance is calculated, and the dynamic adaptive threshold value is generated by combining the noise power correction; The expression of the Otsu adaptive threshold algorithm is: ; ; wherein, optimal threshold value, candidate threshold variable, low value region probability, high value region probability, low value region mean, high value region mean, dynamic adaptive threshold value, empirical correction coefficient, noise power estimation, minimum threshold constraint; In step S401, it is necessary to explain in detail that the Otsu adaptive thresholding algorithm, as a classic image segmentation method, is based on the core idea of ​​finding the optimal segmentation threshold by maximizing the inter-class variance. In radar signal processing, this can be transformed into finding the optimal separation threshold between signal and noise. Specifically, the compensated and calibrated radar echo signal is first normalized to map the signal amplitude to the [0,1] interval. Then, a sliding window (with a window length of 512 sampling points) is used to statistically analyze the signal amplitude histogram, resulting in 256 discretized amplitude distribution intervals. All possible candidate thresholds (0~255) are traversed, and the inter-class variance after each threshold divides the signal into low-value and high-value regions is calculated. The candidate threshold that maximizes the inter-class variance is selected as the initial optimal threshold. Considering the dynamic changes in noise power in radar signals, a noise power estimation module is introduced. This module estimates the current noise power level by calculating the statistical characteristics of the low-amplitude region (top 10% quantile) in the signal amplitude histogram. The final dynamic adaptive threshold value is adjusted from the initial optimal threshold using an empirical correction coefficient (typically 0.8), and its rationality is ensured by combining a minimum threshold constraint T_min (set to be 3 times the noise power). This dynamic threshold value can adaptively track the boundary changes between signal and noise, effectively suppressing noise interference while preserving signal characteristics.

[0035] S402. Based on the dynamic adaptive threshold, the radar echo signal after compensation and calibration is decomposed into wavelet packets at multiple scales. The z sub-bands are generated by the Sym8 wavelet base l-level decomposition. The energy of each sub-band is calculated and the total signal energy is estimated based on the dynamic weighted integral of the signal-to-noise ratio. The expression for calculating the energy of each sub-band is: , ; in, Indicates sub-band Energy estimates, Indicates sub-band The length of the coefficient, Indicates sub-band wavelet packet coefficients, This represents the estimated total signal energy. Indicates the total number of sub-bands. Subband Dynamic weights; It needs to be explained in detail in step S402 that the purpose of applying wavelet packet multi-scale decomposition is to decompose the signal in different frequency ranges in multiple levels, so as to more accurately extract the signal characteristics. In detail, first, the radar echo signal after compensation and calibration is preprocessed based on a dynamic adaptive threshold to remove possible outliers. Subsequently, the signal is decomposed using a Sym8 wavelet basis to generate z sub-bands (for example, when l=3, z=8). Each sub-band contains signal components in a specific frequency range, and the energy estimate of the sub-band is obtained by calculating the square sum of the wavelet packet coefficients of each sub-band. In order to more accurately estimate the total energy of the signal, a dynamic weighted integration mechanism of signal-to-noise ratio is introduced. This mechanism dynamically allocates weights according to the signal-to-noise ratio (SNR) of each sub-band. The sub-band with high signal-to-noise ratio is given a higher weight, and the sub-band with low signal-to-noise ratio is given a lower weight. In specific implementation, first, the signal-to-noise ratio of each sub-band is calculated, then the signal-to-noise ratio is converted to weight value through normalization processing. Finally, the energy estimates of each sub-band are multiplied by their corresponding weights, and the sum is taken to obtain the total energy estimate of the signal. This process can effectively suppress noise interference and improve the accuracy of signal energy estimation.

[0036] S403, based on the total energy estimate of the signal, fusing the temperature, humidity, air pressure, rain and fog concentration data collected by the multi-parameter environment sensor in real time, establishing the mapping relationship between environmental factors and attenuation compensation through the support vector regression model, and generating the environmental compensation coefficient; In step S403, it needs to be explained in detail that the multi-parameter environment sensor can collect data such as temperature, humidity, air pressure and rain and fog concentration in the environment in real time and accurately. These data are crucial for accurately measuring the radar wave intensity, because environmental factors will have a significant impact on the propagation of radar waves, and thus affect the accuracy of the measurement results. In this method, the support vector regression model is used to establish the mapping relationship between environmental factors and attenuation compensation. Specifically, the temperature, humidity, air pressure, rain and fog concentration and other environmental parameters are taken as input variables, and the corresponding attenuation compensation value is taken as output variable, and the support vector regression model is trained. During the training process, the model continuously adjusts its parameters to minimize the error between the predicted value and the actual value, so as to learn the complex relationship between environmental factors and attenuation compensation. After training, the real-time collected environmental parameters are input into the trained model, and the model can output the corresponding environmental compensation coefficient. The environmental compensation coefficient can accurately reflect the influence degree of the current environmental factors on the propagation of radar waves.

[0037] S404, based on the environmental compensation coefficient, performing nonlinear mapping conversion and error propagation correction on the total energy estimate of the signal, quantifying the error contribution of each link through Monte Carlo simulation, and performing post-processing optimization using Kalman filter to generate preliminary intensity measurement value; S405, based on the preliminary intensity measurement value, verify the compensation effect through the spectral purity index monitoring, trigger parameter recalibration when the target frequency point energy ratio is lower than r%, re-execute steps S401-S404, and finally output the radar wave intensity measurement result.

[0038] In step S405, it needs to be explained in detail that the spectral purity index monitoring is used to analyze the energy ratio of the target frequency point in the overall spectrum. Specifically, first, the time-domain signal corresponding to the preliminary intensity measurement value is converted into frequency-domain representation by using fast Fourier transform, so as to obtain the spectral distribution of the signal. Then, the target frequency point is located in the spectrum, and the energy value at the frequency point is calculated. At the same time, the total energy value in the entire spectrum range is calculated. By calculating the ratio of the target frequency point energy value to the total energy value, the target frequency point energy ratio is obtained. Set a threshold r (95)%, when the target frequency point energy ratio is lower than the threshold, it indicates that the signal may be seriously interfered or attenuated, resulting in that the target signal characteristics are not obvious. At this time, the parameter recalibration mechanism is triggered, and steps S401-S404 are re-executed. In this process, the system will re-perform dynamic adaptive threshold calculation, signal total energy estimation, environment compensation coefficient generation, nonlinear mapping conversion and error propagation correction and other operations, so as to ensure that each parameter and compensation measure can accurately reflect the current signal and environment state. After recalibration, the spectral purity index monitoring is performed again until the target frequency point energy ratio meets the requirements, and finally the accurate and reliable radar wave intensity measurement result is output.

[0039] As an optional embodiment of the present application, optionally, in step S404, based on the environment compensation coefficient, the signal total energy estimation value is subjected to nonlinear mapping conversion and error propagation correction, the error contribution of each link is quantified through Monte Carlo simulation, and Kalman filter is used for post-processing optimization to generate the preliminary intensity measurement value, including: S4041, based on the environment compensation coefficient and the signal total energy estimation value, the logarithmic scale conversion is performed through the nonlinear mapping function to generate the linear intensity value; The expression of the nonlinear mapping function is: ; wherein, the linear intensity value is represented by L, the signal total energy estimation value is represented by E, the environment compensation coefficient is represented by C, the reference energy threshold is represented by E0; It needs to be explained in detail in step S4041 that the design of the nonlinear mapping function aims to combine the signal total energy estimate value with the environmental compensation coefficient, and generate a linear intensity value that is more consistent with the actual physical meaning through a specific mathematical transformation. Specifically, the function adopts a logarithmic scale conversion method, normalizes the signal total energy estimate value based on a reference energy threshold, and further modifies it in combination with the environmental compensation coefficient. This conversion method can effectively handle the dynamic changes of signal energy under different environmental conditions, making the generated linear intensity value more stable and comparable. In practical applications, the reference energy threshold can be set according to the performance indicators and measurement requirements of the specific radar system to ensure the accuracy and reliability of the conversion results. Through the processing of the nonlinear mapping function, the system can more accurately reflect the actual intensity of the radar wave under different environments.

[0040] S4042, based on the linear intensity value, the statistical distribution of threshold setting error, energy integration error and environmental compensation error is quantified through Monte Carlo simulation, an error propagation model is established, the error contribution of each link is quantified and the total error confidence interval is determined; The expression of the error propagation model is: ; wherein, σtot represents the total error standard deviation, σth represents the dynamic adaptive threshold value, I represents the final intensity measurement value, σtherr represents the standard deviation of the threshold setting error, E represents the signal total energy estimate value, σint represents the standard deviation of the energy integration error, K represents the environmental compensation coefficient, σK represents the standard deviation of the environmental compensation error.

[0041] In step S4042, it is necessary to explain in detail that Monte Carlo simulation simulates the random error distribution in the actual measurement process through a large number of repeated experiments, thereby accurately assessing the impact of errors in each stage on the final measurement result. Specifically, firstly, error statistical models are established for key stages such as threshold setting error, energy integration error, and environmental compensation error. Among them, threshold setting error mainly considers the random fluctuations in the calculation process of dynamic adaptive threshold value, energy integration error focuses on analyzing the discreteness of energy estimates of each sub-frequency band after wavelet packet decomposition, and environmental compensation error focuses on the deviation between the predicted value and the actual value of the support vector regression model. Subsequently, a large number of sample data are randomly generated within the set error range, and the measurement process under different error combinations is simulated through simulation experiments. The calculation results of the linear intensity value are recorded for each experiment, and the distribution characteristics of all experimental values ​​are statistically analyzed. Based on the statistical results, an error propagation model is established, which can clearly show how the errors in each stage gradually accumulate and are propagated to the final measurement value through operations such as nonlinear mapping and energy integration. By analyzing the model output, the contribution of each error component to the total error can be quantified. For example, threshold setting error may account for 25% of the total error, energy integration error for 40%, and environmental compensation error for 35%. Simultaneously, the confidence interval for the total error is determined based on statistical distribution characteristics; for instance, the total error range at a 95% confidence level is ±0.5 dB. This quantitative analysis process effectively identifies weak points in the system, providing data support for optimizing measurement methods and improving measurement accuracy.

[0042] S4043. Based on the total error confidence interval, a Kalman filter is used to smooth and denoise the linear intensity value, and dynamic optimization is achieved through the state equation and the observation equation. The expression for the state equation is: ;in, Represents the state vector. Represents the state transition matrix. Represents the process noise vector; The expression for the observation equation is: ;in, Observation vector, Represents the observation matrix. Represents the measurement noise vector; The Kalman filter is a method for optimal estimation of system states through joint optimization of state equations and observation equations. Specifically, first construct the state equation, which describes the change of system state over time. The state vector usually contains the linear intensity value and its first or higher order derivative to reflect the dynamic characteristics of the signal intensity. The state transition matrix is set according to the system dynamics, for example, in radar wave intensity measurement, the state transition matrix can be assumed to be the identity matrix, indicating that the signal intensity remains stable in a short time. The process noise vector is used to describe the random disturbance in the system state change process, and its covariance matrix can be adjusted according to the actual measurement environment. At the same time, the observation equation is constructed, which establishes the mapping relationship between the state vector and the actual observation value. The observation value vector is usually the directly measured linear intensity value, and the observation matrix is set according to the observation model, for example, in the simple case it can be set to the identity matrix. The measurement noise vector is used to describe the random error in the observation process, and its covariance matrix can be determined by experimental calibration or empirical value. In the filtering process, the Kalman filter realizes dynamic optimization through two steps of prediction and update. The prediction step is based on the state equation and the optimal estimation value at the last time, to calculate the priori estimation value and covariance matrix at the current time. The update step combines the observation equation and the observation value at the current time to correct the priori estimation value, to obtain the posteriori estimation value and covariance matrix. Through the continuous iteration of these two steps, the Kalman filter can gradually reduce the state estimation error and improve the measurement accuracy. In practical applications, the parameter setting of the Kalman filter has an important influence on the filtering effect. For example, the setting of the process noise covariance matrix and the measurement noise covariance matrix needs to consider the system dynamic characteristics and the measurement environment noise level. If the process noise is set too large, the filter may rely too much on the observation value, introducing too much noise; if the measurement noise is set too large, the filter may respond slowly to the state change. Therefore, it is usually necessary to determine the optimal parameter value through experimental calibration or adaptive adjustment method. In addition, the Kalman filter can also handle nonlinear systems through extended forms, such as Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF), to adapt to more complex measurement scenarios. Through the smoothing denoising processing of the Kalman filter, the system can significantly reduce the random error in the linear intensity value and improve the stability and reliability of the measurement results.

[0043] S4044, based on the smoothed and denoised linear intensity value, verify the error correction effect through the spectral purity index monitoring, and trigger the Kalman filter parameter re-calibration when the total error exceeds the preset threshold, and re-execute the steps S4041-S4043 to generate the preliminary intensity measurement value meeting the accuracy requirement.

[0044] In step S4044, it needs to be explained in detail that the spectral purity index monitoring plays a key quality control role in this link. Specifically, the time domain signal corresponding to the linear intensity value after smoothing and denoising is first converted into a frequency domain signal by fast Fourier transform, and then the complete frequency spectrum distribution graph is obtained. In the frequency spectrum distribution, the target frequency point is accurately located, and the energy value of the frequency point and the total energy value of the entire frequency spectrum are calculated respectively. By calculating the ratio of the energy value of the target frequency point to the total energy value, the energy proportion of the target frequency point is obtained, which is used as an important indicator to measure the error correction effect. A reasonable total error preset threshold is set, which needs to consider factors such as the performance indicators of the radar system, the measurement accuracy requirements and the actual application scenarios. When the calculated total error exceeds this preset threshold, it indicates that the current error correction effect does not meet the expected standard, and some errors in the process may not have been effectively controlled or corrected. At this time, the system automatically triggers the Kalman filter parameter recalibration mechanism. Steps S4041-S4043 are re-executed, and in the re-execution process, the system will re-determine the parameters in the non-linear mapping function according to the new state and environmental information, re-construct the error propagation model and quantify the error contribution of each link, and optimize and adjust the parameters in the state equation and observation equation of the Kalman filter. Through this series of re-operations, it ensures that each parameter and correction measure can more accurately adapt to the current signal characteristics and environmental conditions, so as to generate a preliminary intensity measurement value that meets the accuracy requirements.

[0045] Embodiment 2 A radar wave intensity measurement system based on multi-stage filter amplification, comprising a processor and a memory for storing processor executable instructions; wherein the processor is configured to implement the radar wave intensity measurement method based on multi-stage filter amplification when executing the executable instructions.

[0046] It should be noted that the computer device includes a processor, a memory, and the computer device can further include one or more of a multimedia component, an input / output (I / O) interface, and a communication component.

[0047] The processor is configured to control the overall operation of the computer device to complete all or part of the steps of the radar wave intensity measurement method based on multi-stage filter amplification.

[0048] The memory is used to store various types of data to support the operation of the computer device, which can include, for example, instructions for any application or method operating on the computer device, and application-related data; the memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0049] The multimedia component can include a screen, which can be a touch screen, for example, and an audio component for outputting and / or inputting audio signals; for example, the audio component can include a microphone for receiving external audio signals, and the received audio signals can be further stored in the memory or transmitted through the communication component; the audio component also includes at least one speaker for outputting audio signals.

[0050] The I / O interface provides an interface between the processor and other interface modules, which can be a keyboard, a mouse, a button, etc.; these buttons can be virtual buttons or physical buttons.

[0051] The communication component is used for wired or wireless communication between the computer device and other devices; wireless communication, such as Wi-Fi, Bluetooth, near field communication (NFC), 2G, 3G, 4G or 5G, or a combination of one or more of them, so the corresponding communication component can include a Wi-Fi module, a Bluetooth module, an NFC module, and a mobile communication module.

[0052] As a preferred scheme of the embodiment, the computer device can be implemented by one or more Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), controller, microcontroller, microprocessor or other electronic elements for executing the radar wave intensity measurement method based on multi-stage filtering amplification.

[0053] Although the embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made thereto without departing from the principles and spirit of the present application, the scope of which is defined by the claims and their equivalents.

Claims

1. A radar wave intensity measurement method based on multi-stage filtering and amplification, characterized in that, The method includes: S1. Perform joint adaptive filtering preprocessing of the radar echo signal in the temporal and spatial domains. Improve the signal-to-noise ratio and suppress multipath interference through the synergistic optimization of temporal wavelet packet decomposition and spatial beamforming to obtain the preprocessed radar echo signal. S2. Perform intelligent dynamic gain reconstruction on the preprocessed radar echo signal, and adjust the gain by combining feedforward prediction and feedback control to obtain the radar echo signal with adjusted gain. S3. Perform nonlinear harmonic tracking compensation on the radar echo signal after gain adjustment. Reduce harmonic distortion by modeling with a high-order Volterra model and reverse compensation, and integrate laser ranging calibration to achieve dynamic correction of spatial attenuation, and obtain the compensated and calibrated radar echo signal. S4. Dynamic threshold intensity extraction is performed on the compensated and calibrated radar echo signal. The radar wave energy estimate is calculated by adaptive threshold setting and multi-scale energy integration. Environmental factor compensation is then integrated to optimize the measurement accuracy and obtain the final radar wave intensity measurement result.

2. The radar wave intensity measurement method based on multi-stage filtering and amplification as described in claim 1, characterized in that, Obtaining the preprocessed radar echo signal in step S1 includes: S101. Perform analog-to-digital conversion on the radar echo signal to obtain a digitized radar echo signal; S102. Based on the digitized radar echo signal, wavelet packet decomposition and soft threshold filtering are used to suppress time-domain noise and obtain the radar signal after time-domain noise reduction. S103. Based on the radar signal after time-domain noise reduction, spatial interference suppression is performed using phased array beamforming and DOA estimation to obtain the radar signal after spatial filtering. S104. The ADMM algorithm is used to jointly optimize the radar signal after spatial filtering and the radar signal after temporal denoising to obtain the preprocessed radar echo signal.

3. The radar wave intensity measurement method based on multi-stage filtering and amplification as described in claim 1, characterized in that, The gain-adjusted radar echo signal obtained in step S2 includes: S201. Establish a feedforward prediction model based on the historical radar echo signal intensity distribution, and use the Kalman filter in the feedforward prediction model to estimate the current signal gain requirement and generate preliminary gain control parameters. S202. Real-time acquisition of the time-spectrum characteristics of the preprocessed radar echo signal, and dynamic correction of the feedforward prediction deviation based on the least mean square algorithm to generate feedback compensation parameters. S203. Integrate the preliminary gain control parameters and feedback compensation parameters, and generate a gain reconstruction coefficient matrix using a fuzzy PID controller; S204. Based on the gain reconstruction coefficient matrix, the preprocessed radar echo signal is segmented and linearly amplified to obtain the radar echo signal after preliminary gain adjustment, and an anti-saturation truncation strategy is adopted to avoid signal overload. S205. Verify the gain adjustment effect of the radar echo signal after the initial gain adjustment by using sliding window energy monitoring. When the signal-to-noise ratio improvement rate within the window is lower than the preset threshold, trigger parameter recalibration to obtain the radar echo signal after gain adjustment.

4. The radar wave intensity measurement method based on multi-stage filtering and amplification as described in claim 2, characterized in that, The generation of preliminary gain control parameters in step S201 includes: S2011. Collect historical radar echo signal strength data within a preset time window, remove outliers by moving average filtering, and generate a standardized intensity sequence. S2012. Based on the standardized intensity sequence, fit the intensity distribution characteristics using a Gaussian mixture model, and extract the mean, variance, and skewness parameters as prediction inputs. S2013. The predicted input is imported into the state equation of the Kalman filter, and the gain requirement estimate is updated in combination with the system noise covariance matrix. S2014. Based on the estimated gain requirement, a preliminary gain control parameter matrix is ​​generated using a linear interpolation algorithm.

5. The radar wave intensity measurement method based on multi-stage filtering and amplification as described in claim 1, characterized in that, The compensated and calibrated radar echo signal obtained in step S3 includes: S301. The radar echo signal after gain adjustment is modeled using a high-order Volterra kernel function. Harmonic distortion features are extracted by third-order nonlinear system identification to generate an initial harmonic compensation model. S302. Based on the initial harmonic compensation model, the reverse compensation coefficient is optimized using the minimum mean square error criterion, and the Volterra kernel parameters are dynamically adjusted to suppress harmonic components. S303: Real-time acquisition of spatial distance data, and conversion of distance information into spatial attenuation coefficient through the attenuation factor calculation module; S304. The reverse compensation coefficient and the spatial attenuation coefficient are fused together, and a comprehensive compensation parameter matrix is ​​generated using a weighted fusion algorithm. S305. The signal is nonlinearly compensated and attenuated using the comprehensive compensation parameter matrix. The compensation effect is verified by monitoring the spectral purity. When the harmonic residue is lower than the preset threshold, the compensated and calibrated radar echo signal is output.

6. The radar wave intensity measurement method based on multi-stage filtering and amplification as described in claim 5, characterized in that, The expression for the attenuation factor calculation module in step S303 is as follows: in, Indicates the spatial attenuation coefficient. This indicates the real-time distance measured by laser ranging. Indicates the transmit antenna gain. Indicates the receiving antenna gain. Indicates the radar wavelength. Indicates the atmospheric absorption coefficient. Indicates the radar beamwidth.

7. The radar wave intensity measurement method based on multi-stage filtering and amplification as described in claim 1, characterized in that, The final radar wave intensity measurement results obtained in step S4 include: S401. Based on the compensated and calibrated radar echo signal, the Otsu adaptive threshold algorithm and noise power estimation joint optimization strategy are adopted. By statistically analyzing the signal amplitude histogram distribution through a sliding window, the optimal threshold that maximizes the inter-class variance is calculated and combined with noise power correction to generate a dynamic adaptive threshold value. S402. Based on the dynamic adaptive threshold value, the compensated and calibrated radar echo signal is decomposed into wavelet packets at multiple scales. The z sub-bands are generated by the Sym8 wavelet base l-level decomposition. The energy of each sub-band is calculated and the total signal energy is estimated based on the dynamic weighted integral of the signal-to-noise ratio. S403. Based on the estimated total energy of the signal, the temperature, humidity, air pressure, and rain / fog concentration data collected in real time by multi-parameter environmental sensors are fused together, and an environmental factor-attenuation compensation mapping relationship is established through a support vector regression model to generate an environmental compensation coefficient. S404. Based on the environmental compensation coefficient, the total energy estimate of the signal is subjected to nonlinear mapping transformation and error propagation correction. The error contribution of each link is quantified by Monte Carlo simulation, and Kalman filter is used for post-processing optimization to generate preliminary intensity measurement values. S405. Based on the preliminary intensity measurement value, the compensation effect is verified by monitoring the spectral purity index. When the energy percentage of the target frequency point is lower than r%, parameter recalibration is triggered, and steps S401-S404 are executed again. Finally, the radar wave intensity measurement result is output.

8. The radar wave intensity measurement method based on multi-stage filtering and amplification as described in claim 7, characterized in that, The process of generating preliminary strength measurements in step S404 includes: S4041. Based on the environmental compensation coefficient and the estimated total signal energy, a linear intensity value is generated by performing a logarithmic scaling transformation through a nonlinear mapping function. S4042. Based on the linear intensity value, the statistical distribution of threshold setting error, energy integration error, and environmental compensation error is quantified through Monte Carlo simulation. An error propagation model is established to quantify the error contribution of each link and determine the total error confidence interval. S4043. Based on the total error confidence interval, a Kalman filter is used to smooth and denoise the linear intensity value, and dynamic optimization is achieved through the state equation and the observation equation. S4044. Based on the linear intensity value after smoothing and denoising, the error correction effect is verified by monitoring the spectral purity index. When the total error exceeds the preset threshold, the Kalman filter parameters are recalibrated, and steps S4041-S4043 are re-executed to generate preliminary intensity measurement values ​​that meet the accuracy requirements.

9. The radar wave intensity measurement method based on multi-stage filtering and amplification as described in claim 8, characterized in that, The expression for the error propagation model in step S4042 is: in, Indicates the standard deviation of the total error. Indicates a dynamic adaptive threshold value. This represents the final strength measurement value. This represents the standard deviation of the threshold setting error. This represents the estimated total signal energy. The standard deviation of the energy integration error is represented by... Indicates the environmental compensation coefficient. This represents the standard deviation of the environmental compensation error.

10. A radar wave intensity measurement system based on multi-stage filtering and amplification, characterized in that, The system includes a processor and a memory for storing processor-executable instructions; The processor is configured to implement the radar wave intensity measurement method based on multi-stage filtering and amplification as described in any one of claims 1 to 9 when executing the executable instructions.

Images