Radar ranging precision improving method based on high-frequency narrow pulse

Abstract

The invention discloses a radar ranging precision improvement method based on high-frequency narrow pulses, and relates to the technical field of radar ranging and ultra-wideband signal processing, and the method comprises the steps: carrying out the gating interception of a radar receiving sequence, and obtaining an intra-gate sequence; performing cepstrum analysis to determine comb tooth intervals and ripple intensity; constructing a reverse comb ripple frequency domain weight sequence based on the ripple intensity, and performing weighted matched filtering on the intra-gate sequence to extract candidate peaks; calculating sidelobe cluster energy corresponding to the candidate peak according to the comb tooth interval; and calculating a group peak suppression score value based on the ripple intensity and the sidelobe cluster energy, and selecting the candidate peak with the maximum score value as a final peak position to calculate the distance. According to the method, side lobe group peak lifting caused by the wet radome can be inhibited, and the distance measurement precision and stability under the complex meteorological condition are improved.

Description

Technical Field

[0001] This invention relates to the field of radar ranging and ultra-wideband signal processing technology, and in particular to a method for improving radar ranging accuracy based on high-frequency narrow pulses. Background Technology

[0002] High-frequency narrow-pulse radar typically estimates target range by transmitting short-duration broadband pulses and performing matched filtering on the echoes. It exhibits good resolution and ranging accuracy in dry environments. However, in practical engineering applications, radar antennas and transceiver components are usually equipped with radomes that serve both protective and shape-integrated functions. Under complex weather conditions such as rain, fog, dew, and snowmelt, the radome surface is prone to forming a wet coating of water film and droplets. This wet coating alters the equivalent electromagnetic boundary conditions of the radome surface, causing quasi-periodic amplitude-frequency response fluctuations in the receiving link within the ultra-wideband, manifested as comb-like ripples. For ranging systems that rely on correlation peak decision, in-band comb-like ripples can alter the sidelobe structure in the matched filter output, specifically manifested as the rise of sidelobe peak clusters and the appearance of clustered peaks. This phenomenon can easily trigger peak decision reversal, causing the ranging algorithm to incorrectly identify sidelobes as main peaks, resulting in clustering of range points and significant degradation in accuracy.

[0003] Existing technologies typically rely solely on fixed transmit pulse templates for matched filtering when addressing such issues, lacking methods to suppress in-band comb ripples caused by wet radomes. Conventional health indicator monitoring often focuses only on macroscopic quantities such as noise floor, transmit power, or gain, making it difficult to perceive the frequency domain comb undulation structure caused by the water film effect. Therefore, when wet radome conditions cause a significant increase in the sidelobe peaks in the relevant domain, existing algorithms cannot adaptively adjust the frequency domain weights or effectively remove sidelobe interference from candidate peaks. This results in the inability to guarantee the ranging stability of the system under adverse weather conditions, making it difficult to meet the application requirements of all-weather high-precision ranging. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies where wet radomes cause comb-like ripples in the receiving band, leading to sidelobe peak rise in the matched filter correlation output, resulting in peak decision reversal and deterioration of ranging accuracy. The invention proposes a radar ranging accuracy improvement method based on high-frequency narrow pulses.

[0005] To address the problems existing in the prior art, the present invention adopts the following technical solution: Methods for improving radar ranging accuracy based on high-frequency narrow pulses include: S1. Collect the radar received sequence, and perform gated interception of the radar received sequence according to the preset ranging range to obtain the sequence within the gate; S2. Perform frequency domain transformation and cepstral analysis on the in-gate sequence, determine the comb spacing by searching for the main peak in the cepstral domain, and calculate the ripple intensity based on the amplitude of the main cepstral peak. S3. Construct a reverse combing frequency domain weight sequence based on ripple intensity, and perform weighted matched filtering on the in-gate sequence using the reverse combing frequency domain weight sequence to obtain the relevant sequence and extract the candidate peak set; S4. Determine the equally spaced distribution of the sidelobe clusters in the relevant domain based on the comb tooth spacing, and calculate the energy of the sidelobe clusters corresponding to each candidate peak in the candidate peak set based on the equally spaced distribution. S5. Calculate the peak suppression score of each candidate peak based on ripple intensity and sidelobe cluster energy, and take the candidate peak with the largest peak suppression score as the final peak position to calculate the target distance value.

[0006] Preferably, gating interception of the radar received sequence includes: Based on the radar system's sampling period, preset ranging range, and light speed value, calculate the ranging gate start index and ranging gate end index respectively. Generate a gated window sequence with the same length as the radar received sequence; The sampling points in the gated window sequence between the starting index and the ending index of the ranging gate are assigned non-zero window coefficient values, and the remaining sampling points in the gated window sequence are assigned zero values. The radar received sequence is multiplied point by point with the gated window sequence to obtain the in-gate sequence.

[0007] Preferably, the sequence within the gate is subjected to frequency domain transformation and cepstral analysis, and the comb spacing is determined by searching for the main peak in the cepstral domain, including: Perform a fast Fourier transform on the sequence within the gate and calculate the logarithmic amplitude spectrum; The cepstral sequence is obtained by performing an inverse Fourier transform on the logarithmic amplitude spectrum; The cepstral search interval is deduced by inversely from the estimated range of comb tooth spacing; Within the cepstral search interval, the point with the largest modulus is taken as the cepstral main peak, and the comb spacing is calculated based on the position index of the cepstral main peak and the sampling period.

[0008] Preferably, calculating the ripple intensity based on the cepstral main peak amplitude includes: Calculate the average modulus of all points within the cepstral search interval, and determine the ripple intensity by the ratio of the modulus of the main cepstral peak to the average modulus.

[0009] Preferably, constructing a reverse comb ripple frequency domain weight sequence based on ripple intensity includes: Calculate the amplitude spectrum of the sequence within the gate, and perform a moving average on the amplitude spectrum to obtain a smoothed baseline spectrum; The ripple ratio sequence is obtained by calculating the ratio of the amplitude spectrum to the smoothed baseline spectrum; The exponential parameter is determined based on the ripple intensity. The ripple ratio sequence is then subjected to a power operation with the exponential parameter as the power and the reciprocal is taken to obtain the reverse comb ripple frequency domain weight sequence.

[0010] Preferably, weighted matched filtering is performed on the in-gate sequence, including: Obtain the spectrum of the transmitted pulse template; The weighted correlation spectrum is obtained by multiplying the spectrum of the anti-comb pattern frequency domain weight sequence, the spectrum of the in-gate sequence, and the conjugate of the transmitted pulse template spectrum. The inverse Fourier transform of the weighted correlation spectrum yields the correlation sequence.

[0011] Preferably, the candidate peak set is extracted, including: Within the range gate index interval, search for local maxima in the relevant sequences; Calculate the median amplitude of the relevant sequence within the range gate index interval, and set an amplitude threshold based on the median amplitude; Local maxima points with amplitudes greater than the amplitude threshold are retained to form a candidate peak set.

[0012] Preferably, calculating the sidelobe cluster energy corresponding to each candidate peak in the candidate peak set includes: The sidelobe distance of the correlation domain is calculated based on the reciprocal of the comb tooth spacing and the sampling period. For each candidate peak position in the candidate peak set, several sampling points are selected on the left and right sides of the candidate peak position with an offset of an integer multiple of the side lobe distance step length. Calculate the sum of squared amplitudes of several sampling points in the relevant sequence, and use it as the energy of the sidelobe cluster corresponding to the candidate peak.

[0013] Preferably, the peak suppression score of each candidate peak is calculated based on the ripple intensity and sidelobe cluster energy, including: Obtain the preset base penalty coefficient and dynamic penalty coefficient; Calculate the product of ripple intensity and dynamic penalty coefficient, and add the product to the base penalty coefficient to obtain the comprehensive penalty weight; The product of the comprehensive penalty weight and the energy of the sidelobe cluster corresponding to the candidate peak is calculated to obtain the normalized denominator; Calculate the squared amplitude of the candidate peak, and obtain the peak suppression score of the candidate peak based on the ratio of the squared amplitude to the normalized denominator.

[0014] Preferably, the candidate peak with the largest peak suppression score is used as the final peak position to calculate the target distance value, including: The candidate peak with the highest peak suppression score is taken as the final peak position; The target distance is determined based on the final peak position, sampling period, and speed of light.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention, through frequency domain transformation and cepstral analysis of the in-gate sequence, can directly quantify the in-band comb-like ripple characteristics caused by wet radomes. It accurately obtains the comb tooth spacing and ripple intensity using the cepstral main peak, and adaptively constructs an anti-comb ripple frequency domain weight sequence based on the ripple intensity. This processing specifically shapes and flattens the spectrum of the received signal, effectively offsetting the amplitude-frequency response fluctuations caused by the water film medium. Thus, it suppresses energy dispersion and related sidelobe structure distortion caused by spectral distortion in the matched filtering stage, reducing the risk of sidelobe peak rise from the source and ensuring signal quality under severe weather conditions.

[0016] 2. This invention utilizes the comb-tooth spacing to derive the equally spaced distribution law of sidelobe clusters in the correlation domain, and establishes a correlation evaluation mechanism between the candidate peak and the energy of the surrounding sidelobe cluster peaks. By calculating the energy of the sidelobe clusters and combining it with the ripple intensity to construct a cluster peak suppression score, differentiated screening of candidate peaks is achieved. This mechanism can automatically enhance the penalty for pseudo-peaks with significant sidelobe characteristics under high ripple intensity conditions, effectively preventing the peak decision from flipping between the main peak and the raised sidelobe, solving the problem of range measurement clustering, and significantly improving the range measurement stability and accuracy of the radar system under wet cover conditions such as rain, fog, and dew. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating a method for improving radar ranging accuracy based on high-frequency narrow pulses, provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] Example: This example provides a method for improving radar ranging accuracy based on high-frequency narrow pulses. See [link to example]. Figure 1 Specifically, including: S1. Collect the radar received sequence, and perform gated interception of the radar received sequence according to the preset ranging range to obtain the sequence within the gate; In an embodiment of the present invention, gating and intercepting the radar received sequence according to a preset ranging range includes: Based on the radar system's sampling period, preset ranging range, and light speed value, calculate the ranging gate start index and ranging gate end index respectively. Collect the radar received sequence and generate a gated window sequence of the same length as the radar received sequence; The sampling points in the gated window sequence between the starting index and the ending index of the ranging gate are assigned non-zero window coefficient values, and the remaining sampling points in the gated window sequence are assigned zero values. Specifically, the sampling period of a radar system refers to the time interval between two adjacent sampling points on the time axis when the receiving link performs discrete sampling of the echo signal. It determines the fineness of the echo waveform's characterization in the time dimension and corresponds to a resolvable range step. The preset ranging range refers to the spatial interval for which range estimation is desired, defined by the minimum and maximum ranging distances. This interval corresponds to the time window allowed for the echo's round-trip propagation, allowing processing resources to be concentrated within the range of interest. The speed of light refers to the propagation speed of electromagnetic waves in a vacuum and serves as a reference constant for converting propagation time and spatial distance. The radar receiving sequence refers to the discrete amplitude sequence arranged in time, obtained by sampling the echo signal after a single transmission at the receiver. It includes components such as target echo and clutter noise. The range gate start index and the range gate... The termination index represents the two sampling position boundaries corresponding to the minimum and maximum ranging distances in the discrete sampling sequence, respectively, and is used to map the continuous range window to the index interval on the discrete sequence. The gated window sequence is a weighted sequence of the same length as the radar receiving sequence, which is used to selectively retain and suppress the receiving sequence in the index dimension. The sampling point between the starting index and the ending index of the ranging gate refers to the discrete sampling position falling within the time window corresponding to the preset ranging range. The non-zero window coefficient value refers to the weight amplitude assigned at the sampling point inside the gate. It can take a constant value or a smooth transition value to control the signal retention ratio inside the gate and reduce the spectral leakage caused by the sudden change at the gate edge. The remaining sampling points are assigned a value of zero to indicate that the echo component outside the gate is completely suppressed, thereby avoiding interference from irrelevant range segments to subsequent processing.

[0020] In detail, the sampling period of the radar system is obtained and recorded as the sampling period value. At the same time, the minimum and maximum ranging distances within the preset ranging range are obtained and recorded as the minimum and maximum ranging distance values. The speed of light is read or pre-stored as a reference constant for the conversion between distance and sampling index. The distance step corresponding to a single sampling point is calculated based on the sampling period value. The distance step is preferably the product of the speed of light and the sampling period value divided by two to reflect the distance conversion relationship of electromagnetic wave round-trip propagation. Then, the minimum ranging distance value is divided by the distance step and rounded down to obtain the starting index of the ranging gate. The maximum ranging distance value is divided by the distance step and rounded up to obtain the ending index of the ranging gate. This ensures that the preset ranging range is completely covered and avoids missed detections caused by threshold truncation. When the minimum ranging distance value is less than zero, it is preferably clamped to zero to satisfy the non-negative constraint of physical distance. When the ending index of the ranging gate is less than the starting index of the ranging gate, it is preferably swapped or the ending index of the ranging gate is set as the starting index of the ranging gate to ensure that the index interval is valid.

[0021] The radar received sequence is acquired and its length is determined as the sequence length value. Based on the sequence length value, a gated window sequence with the same length as the radar received sequence is generated and initialized as an all-zero sequence. In the gated window sequence, sampling points between the starting index and the ending index of the ranging gate are assigned non-zero window coefficient values ​​to form a ranging gate. The window coefficient value is preferably one as a rectangular gate to obtain the maximum signal retention ratio within the gate. It is further preferred to set a transition band on each side of the ranging gate to reduce spectral leakage caused by abrupt changes at the gate edge. The width of the transition band is preferably sixteen to sixty-four sampling points and can be appropriately increased with the increase of the sampling rate. Within the transition band, the window coefficient value is smoothly increased from zero to one and smoothly decreased from one to zero at the end. The smoothing method is preferably cosine smoothing to suppress high-frequency leakage while ensuring energy retention within the gate. The remaining sampling points in the gated window sequence are kept at zero to completely suppress the echo component of the range segment outside the gate. Thus, a gated window sequence with the same length as the radar received sequence and non-zero only within the corresponding index interval of the preset ranging range is obtained for subsequent point-by-point multiplication to generate the in-gate sequence.

[0022] The radar received sequence is multiplied point by point with the gated window sequence to obtain the in-gate sequence; Specifically, the in-gate sequence refers to the discrete echo sequence obtained by multiplying the radar received sequence and the gated window sequence point by point according to the same sampling index. This sequence retains the echo information of the target range segment within the index interval corresponding to the starting index and ending index of the ranging gate, and can be weighted according to the window coefficient value. Outside the index interval, it is assigned zero to suppress the echo and clutter interference of the non-interest range segment. Therefore, the in-gate sequence is equivalent to selectively truncate and shape the energy of the echo in the spatial range window, so that the subsequent frequency domain analysis and matched filtering processing focus on the preset ranging range and reduce the influence of the out-of-gate components on the correlation peak and sidelobe structure. Among them, the sidelobe structure refers to a series of peak-valley combinations with small amplitudes but definite shapes and distribution patterns that still appear in the correlation sequence after matched filtering or correlation processing of radar echoes, in addition to the main peak. It originates from the energy diffusion effect caused by the combined effects of factors such as the limited time and bandwidth of the transmitted pulse waveform, windowing or coding methods, and frequency response fluctuations of the receiving link. The sidelobe structure usually manifests as multiple secondary peaks on both sides of the main peak and the valleys between them, with a certain degree of symmetry or quasi-periodicity. When the sidelobe structure is raised or appears in clusters, it will reduce the amplitude comparison between the main peak and non-main peaks and increase the probability of peak decision misselection, thus adversely affecting the ranging stability and accuracy.

[0023] In detail, the radar received sequence is treated as discrete echo data arranged by sampling index and denoted as the radar received sequence. A gated window sequence is treated as a weighted sequence of the same length as the radar received sequence and denoted as the gated window sequence. First, the length consistency of the two sequences is checked. If the lengths are inconsistent, the shorter one is preferred as the effective length, and the longer one is truncated or padded with zeros at the end to ensure a one-to-one correspondence in point-by-point calculations. Then, using the sampling index as a reference, the process iterates from the zero index to the effective length minus one. At each sampling index, the current sampled value of the radar received sequence and the current window coefficient value of the gated window sequence are read and multiplied. The product is used as the sampled value of the gated sequence at that sampling index, thus obtaining the gated sequence. Where, when sampling... When the sampling index falls outside the starting and ending indices of the ranging gate, the window coefficient of the gated window sequence is zero, making the product zero to achieve complete suppression of external echoes. When the sampling index falls within the ranging gate index interval, the window coefficient of the gated window sequence is non-zero, allowing the internal sequence to retain the internal echo amplitude and achieving weighted retention based on the window coefficient value. Furthermore, when the gated window sequence sets a smooth transition band at the edge of the ranging gate, the amplitude of the internal sequence at the edge will gradually change according to the window coefficient value, thereby reducing spectral leakage caused by hard truncation. The final output internal sequence is a discrete sequence of the same length as the radar received sequence and contains effective echo information only within the index interval corresponding to the preset ranging range, serving as the input for subsequent frequency domain analysis and weighted matched filtering.

[0024] S2. Perform frequency domain transformation and cepstral analysis on the in-gate sequence, determine the comb spacing by searching for the main peak in the cepstral domain, and calculate the ripple intensity based on the amplitude of the main cepstral peak. In an embodiment of the present invention, the comb spacing is determined by searching for the main peak in the cepstral domain, and the ripple intensity is calculated based on the amplitude of the main cepstral peak, including: Perform a fast Fourier transform on the sequence within the gate and calculate the logarithmic amplitude spectrum; The cepstral sequence is obtained by performing an inverse Fourier transform on the logarithmic amplitude spectrum; In detail, the in-gate sequence is obtained and its length is determined as the sequence length value. To ensure frequency domain resolution and computational efficiency, the Fast Fourier Transform (FFT) length is preferably set to an integer power of two, not less than the sequence length value, and denoted as the transform length value. When the in-gate sequence length is less than the transform length value, zeros are padded to the end of the in-gate sequence to the transform length value to avoid additional distortion introduced by time domain truncation. Subsequently, an FFT is performed on the in-gate sequence to obtain the frequency domain complex spectrum of the in-gate sequence, and the amplitude spectrum is obtained by taking the modulus of the frequency domain complex spectrum at each frequency point. To ensure that the amplitude spectrum satisfies the additivity of subsequent cepstral analysis and to transform multiplicative fluctuations into additive fluctuations... The natural logarithm of the amplitude spectrum is taken to obtain the logarithmic amplitude spectrum. To avoid logarithmic divergence caused by zero amplitude, a very small positive number is preferably superimposed on the amplitude spectrum as a numerically stable term. The very small positive number is preferably on the order of one to ten to the power of negative twelve to balance numerical stability and avoid introducing visible bias. After obtaining the logarithmic amplitude spectrum, the inverse fast Fourier transform is performed on the logarithmic amplitude spectrum and the real part is taken to obtain the cepstrum sequence. The sequence index of the cepstrum sequence corresponds to the concentrated representation of the quasi-periodic structure of amplitude fluctuations in the echo band in the cepstrum domain. When there is comb-like ripple in the amplitude frequency response of the in-gate sequence, the cepstrum sequence will show a significant peak.

[0025] The cepstral search interval is deduced by inversely from the estimated range of comb tooth spacing; Within the cepstral search interval, the point with the largest modulus is taken as the cepstral main peak, and the comb tooth interval is calculated based on the position index of the cepstral main peak and the sampling period. It should be noted that the comb spacing refers to the repetition interval on the frequency axis of the quasi-periodic ripples introduced by the wet radome in the in-band amplitude spectrum of the in-gate sequence. It reflects the average frequency interval of adjacent peaks and valleys appearing in pairs when the equivalent amplitude-frequency response of the receiver link presents a comb-like ripple. The smaller the comb spacing, the denser the in-band ripples and the finer the energy distribution of the ripples on each frequency band of the broadband pulse. This makes it easier to form an equally spaced sidelobe cluster structure in the matched filter or correlation output and reduce the amplitude contrast between the main peak and non-main peaks. Therefore, the comb spacing can be used as a key scale to characterize the periodicity of the in-band ripples caused by the wet radome and to derive the equally spaced distribution of the sidelobe clusters in the correlation domain. Among them, wet radome refers to the radome covering the outside of the radar antenna and transceiver components for protection and shape integration. Under the influence of environmental factors such as rain splashing, fog droplet condensation, condensation, or snow melting seepage, its outer or inner surface forms a water-covered state in which a continuous water film and discrete water droplets coexist. This water-covered medium has a higher dielectric constant and loss characteristics than air, which will change the electromagnetic boundary conditions and equivalent thickness distribution of the radome surface. This will cause electromagnetic waves to generate additional reflection, scattering, and absorption when penetrating the radome, and introduce transmission fluctuations that vary with frequency. As a result, the receiving link will show quasi-periodic amplitude-frequency response ripple in the ultra-wideband and further induce the sidelobe structure of the matched filter correlation output to rise, which will have an adverse effect on ranging stability and accuracy.

[0026] In detail, the estimated range of the comb tooth spacing is obtained and represented as the minimum and maximum comb tooth spacing. This estimated range is used to constrain the search for the main peak in the cepstral domain to avoid misjudging low-order envelope or random noise peaks as the main peak of the comb tooth. The minimum and maximum comb tooth spacing are preferably obtained through observation and statistics of the amplitude spectrum of the sequence within the gate under dry and wet conditions. Specifically, the repetition interval between adjacent peaks and valleys can be identified in the amplitude spectrum within the band, and its robust statistical value can be used as the estimated range. Further preferably, a margin of 20% to 50% is added to this range to cover the redistribution of water film on different radome surfaces. The spacing drift; after obtaining the estimated range, the minimum comb spacing and the maximum comb spacing are converted into cepstral search boundaries by using the one-to-one correspondence between the cepstral index and the frequency fluctuation period. Specifically, the reciprocal of the maximum comb spacing is used as the minimum fluctuation frequency and its reciprocal is converted into the lower bound of the cepstral index, and the reciprocal of the minimum comb spacing is used as the maximum fluctuation frequency and its reciprocal is converted into the upper bound of the cepstral index, thus obtaining the start index and end index of the cepstral search interval. The conversion basis of the cepstral index is that the time scale corresponding to the index in the cepstral sequence is equal to the product of the sampling period and the cepstral index.

[0027] Within the cepstral search interval, the modulus of the cepstral sequence is calculated point by point, and the point with the largest modulus is searched. This point is identified as the cepstral main peak, and its position index is recorded as the main peak index value. After determining the cepstral main peak, the period scale corresponding to the cepstral main peak is calculated based on the main peak index value and the sampling period. Specifically, the main peak index value is multiplied by the sampling period to obtain the cepstral main peak period value, and then one is divided by the cepstral main peak period value to obtain the comb tooth interval. Thus, the comb tooth interval that matches the comb-like undulation of the amplitude spectrum of the gate sequence is output for subsequent derivation of the equal interval of the side lobe clusters and construction of the anti-comb weight.

[0028] Calculate the average magnitude of all points within the cepstral search interval, and determine the ripple intensity by the ratio of the magnitude of the main cepstral peak to the average magnitude. It should be noted that ripple intensity is an indicator used to quantify the significance of comb-like fluctuations in the amplitude spectrum of an in-band sequence. It is determined by the ratio of the magnitude of the main peak in the cepstrum to the average level of the background magnitude in the cepstrum search interval. When there are stable quasi-periodic fluctuations in the in-band amplitude-frequency response of an in-band sequence, these fluctuations will form a prominent main peak in the cepstrum, increasing the ripple intensity. The greater the ripple intensity, the more clearly the in-band energy is cut into a comb-like shape with alternating peaks and valleys on the frequency axis, making it easier to disperse the relevant output energy into equally spaced sidelobe peaks and reduce the amplitude contrast between the main peak and the sidelobe. Therefore, ripple intensity can be used to characterize the degree of comb-like ripples caused by wet radomes.

[0029] In detail, after generating the cepstral sequence and determining the cepstral search interval and cepstral peak, the start and end indices of the cepstral search interval are read, and all sampling points of the cepstral sequence within that index interval are traversed accordingly. For each sampling point, its magnitude is calculated and accumulated to obtain the cumulative sum of magnitudes. Simultaneously, the number of sampling points involved in the accumulation is counted to obtain the interval point count. After the traversal, the cumulative sum of magnitudes is divided by the interval point count to obtain the average magnitude of all points within the cepstral search interval. To avoid the average magnitude being too small and causing unstable ratios, a very small positive number is preferably superimposed on the average magnitude as... The numerically stable term and the smallest positive number are preferably on the order of one to ten to the power of negative twelve. Then, the modulus value of the sampling point corresponding to the cepstrum main peak is read as the modulus value of the cepstrum main peak. The modulus value of the cepstrum main peak is divided by the average value of the modulus values ​​to obtain the ratio, and this ratio is defined as the ripple intensity. Thus, the ripple intensity characterizes the prominence of the cepstrum main peak relative to the background level of the search interval. When the ripple intensity is close to one, it indicates that the cepstrum main peak is not significant and the corresponding comb undulations in the band are weak. When the ripple intensity is significantly greater than one, it indicates that the cepstrum main peak is prominent and the corresponding comb undulations in the band are significant and more likely to induce the rise of related sidelobe peaks.

[0030] S3. Construct a reverse combing frequency domain weight sequence based on ripple intensity, and perform weighted matched filtering on the in-gate sequence using the reverse combing frequency domain weight sequence to obtain the relevant sequence and extract the candidate peak set; In an embodiment of the present invention, a reverse combing ripple frequency domain weight sequence is constructed based on the ripple intensity, and the in-gate sequence is weighted matched filtering using the reverse combing ripple frequency domain weight sequence, including: Calculate the amplitude spectrum of the sequence within the gate, and perform a moving average on the amplitude spectrum to obtain a smoothed baseline spectrum; The ripple ratio sequence is obtained by calculating the ratio of the amplitude spectrum to the smoothed baseline spectrum; The exponential parameter is determined based on the ripple intensity. The ripple ratio sequence is then subjected to a power operation with the exponential parameter as the power and the reciprocal is taken to obtain the reverse comb ripple frequency domain weight sequence. It should be noted that the ripple ratio sequence refers to the normalized fluctuation sequence obtained by comparing the amplitude spectrum of the in-gate sequence with the smoothed baseline spectrum at the same frequency point in the frequency domain. It reflects the local enhancement or attenuation of the in-band amplitude-frequency response relative to the slowly varying envelope. It is usually greater than one at the comb peak and less than one at the comb valley, thus explicitly expressing the quasi-periodic comb-shaped ripple caused by wet radome in a directly computable numerical form. The reverse comb-shaped ripple frequency domain weight sequence refers to the frequency point weight sequence obtained by exponentially operating the ripple ratio sequence according to the exponential parameter determined by the ripple intensity and taking the reciprocal. This weight sequence assigns a smaller weight at the comb peak to suppress the energy proportion of the excessively strong frequency band and a larger weight at the comb valley to compensate for the energy proportion of the suppressed frequency band. This flattens the comb-shaped fluctuation in the frequency domain and reduces the risk of sidelobe structure lifting and peak clustering in subsequent weighted matched filtering and correlation processing.

[0031] In detail, the in-gate sequence is obtained and the transform length for frequency domain processing is determined. Preferably, the transform length is set to an integer power of two, not less than the length of the in-gate sequence, to balance frequency domain resolution and computational efficiency. A fast Fourier transform is performed on the in-gate sequence to obtain a complex spectrum in the frequency domain. The amplitude spectrum is obtained by taking the modulus of the complex spectrum at each frequency point. To avoid introducing instability in the amplitude spectrum at extremely small amplitude values, a very small positive number is preferably superimposed on the modulus result, and the very small positive number is preferably on the order of one to ten to the power of -twelfth. Subsequently, the amplitude spectrum is subjected to a moving average to obtain a slow-responding signal. The smoothed baseline spectrum with variable envelope is preferably matched with the comb tooth spacing to achieve ripple-free baseline estimation of comb tooth undulations. Specifically, the comb tooth spacing can be converted into the frequency domain frequency span and twice the comb tooth span plus one is taken as the window length, so that the smoothed baseline spectrum retains the overall attenuation trend in the band without excessively following the comb tooth peaks and valleys. After obtaining the smoothed baseline spectrum, the ratio of the amplitude spectrum to the smoothed baseline spectrum is calculated for each frequency point to obtain the ripple ratio sequence, so that the ripple ratio sequence is greater than one at the comb tooth peaks and less than one at the comb tooth valleys, thus explicitly characterizing the comb-like undulations in the band.

[0032] Further, an exponential parameter is determined based on the ripple intensity to achieve adaptive adjustment of the anti-combing ripple level. The exponential parameter is preferably proportional to the ripple intensity and subject to an upper limit constraint. Specifically, the ripple intensity is multiplied by a preset proportional coefficient to obtain a candidate exponent, wherein the proportional coefficient is preferably between 0.5 and 2 to match the dynamic range of ripple intensity of different systems. The ripple ratio sequence is subjected to a power operation with the exponential parameter as the power at each frequency point, and the reciprocal is taken to obtain the anti-combing ripple frequency domain weight sequence, so that the weight at the comb tooth peak is less than one and the weight at the comb tooth valley is greater than one, thereby smoothing out the frequency domain peak and valley fluctuations and suppressing the sidelobe peak rise caused by the in-band comb ripple in the subsequent weighted matched filtering.

[0033] Obtain the spectrum of the transmitted pulse template; The weighted correlation spectrum is obtained by multiplying the spectrum of the anti-comb pattern frequency domain weight sequence, the spectrum of the in-gate sequence, and the conjugate of the transmitted pulse template spectrum. Perform an inverse Fourier transform on the weighted correlation spectrum to obtain the correlation sequence; Specifically, the spectrum of the transmitted pulse template refers to the complex frequency domain representation obtained by transforming the pre-saved transmitted pulse template sequence in the frequency domain. It characterizes the amplitude and phase distribution of the transmitted pulse at each frequency component and serves as a reference for matched filtering, ensuring that the received echo accumulates energy in the frequency domain in phase with the template to form a sharp main peak. The weighted correlation spectrum refers to the complex spectrum obtained by multiplying the spectrum of the in-gate sequence and the spectrum of the transmitted pulse template conjugate point-by-point in the frequency domain, and further introducing the anti-comb ripple frequency domain weight sequence for point-by-point weighting. It reflects the in-band comb ripple compensation weight. The contribution of each frequency component to the correlation accumulation is then used to smooth out the peak-valley energy imbalance caused by wet radome in the frequency domain and suppress the sidelobe peak rise. The correlation sequence refers to the time-domain discrete sequence obtained by inverse frequency domain transformation of the weighted correlation spectrum. The value of each sampling point represents the similarity or matching degree between the in-gate echo and the transmitted pulse template under the corresponding relative displacement. The main peak corresponds to the most matching position and is used to determine the target distance. The secondary peaks and valleys on both sides of the main peak constitute the sidelobe structure and may exhibit a clustered rise pattern under wet radome conditions, thus requiring further candidate peak screening and peak suppression decision.

[0034] In detail, a transmit pulse template sequence is pre-saved on the digital baseband side of the radar transmit link and used as the transmit pulse template. The transmit pulse template is preferably an equivalent baseband waveform obtained by re-sampling the transmit waveform under dry calibration conditions or by system calibration, to ensure that the template accurately represents the main peak shape under ideal matching conditions and avoids solidifying the comb-like ripples introduced by wet overlay conditions into the template. During one-frame processing, the transform length of the frequency domain processing is determined, ensuring that the anti-comb ripple frequency domain weight sequence, the gate sequence, and the transmit pulse template have the same number of frequency points at that transform length. When the length of the gate sequence or the transmit pulse template is insufficient for the transform length, zeros are preferably padded to the end to the transform length to avoid unwanted aliasing introduced by frequency domain circular convolution. Subsequently, a fast Fourier transform is performed on the transmit pulse template to obtain its spectrum, and the spectrum is then processed point by point. The frequency domain coefficients of the matched filter are constructed using the yoke. Simultaneously, a Fast Fourier Transform (FFT) is performed on the in-gate sequence to obtain its spectrum, which is then aligned with the obtained anti-comb ripple frequency domain weight sequence at frequency points. Then, at each frequency point, the current weight value of the anti-comb ripple frequency domain weight sequence, the current complex value of the in-gate sequence spectrum, and the current complex value of the conjugate of the transmitted pulse template spectrum are multiplied frequency-by-frequency to obtain a weighted correlation spectrum. This allows the frequency points corresponding to the in-band comb peaks and valleys to be adaptively weighted during correlation accumulation, thereby suppressing the amplification effect of the comb ripples caused by the wet radome on the correlation sidelobe structure. Finally, an Inverse Fast Fourier Transform (IFFT) is performed on the weighted correlation spectrum to obtain the correlation sequence. The correlation sequence reflects the similarity distribution between the in-gate echo and the transmitted pulse template in the sampling index dimension, forming the main peak and sidelobe structure, providing direct input for subsequent candidate peak extraction and peak suppression score calculation based on sidelobe cluster energy.

[0035] Within the range gate index interval, search for local maxima in the relevant sequences; Calculate the median amplitude of the relevant sequence within the range gate index interval, and set an amplitude threshold based on the median amplitude; Local maxima points with amplitudes greater than the amplitude threshold are retained to form a candidate peak set; Specifically, the candidate peak set refers to the set of peak positions retained after searching for local maxima of the relevant sequences within the ranging gate index interval and filtering by amplitude threshold. Each candidate peak corresponds to the local maximum point of the matching degree of the relevant sequence at a certain relative displacement and may represent the target main peak or a secondary peak formed by the sidelobe structure and clutter. By retaining only local maxima points with amplitudes significantly higher than the background level within the gate, the candidate peak set narrows the subsequent decision range from all sampling points to a finite number of high-confidence candidate points. This facilitates the calculation of the sidelobe cluster energy of each candidate peak based on the equally spaced distribution of the sidelobe clusters derived from the comb spacing in subsequent steps and the calculation of the peak group suppression score to suppress the interference of the sidelobe peak group rise caused by the wet radome on the main peak decision.

[0036] In detail, the relevant sequences are obtained, and the starting index and ending index of the ranging gate are obtained simultaneously to determine the ranging gate index interval. The relevant sequence samples within the ranging gate index interval are used as candidate search objects, and at least one sampling point is reserved at each end of the interval to avoid the inability to complete the neighborhood comparison at the boundary. Then, the relevant sequences are scanned point by point within the index interval. The amplitude of each current sampling point is calculated and compared with the amplitudes of its left and right adjacent sampling points. When the amplitude of the current sampling point is not less than the amplitude of its left neighbor and not less than the amplitude of its right neighbor, it is determined as a local maximum point and its index and amplitude are recorded, thereby obtaining the initial set of local maximum points.

[0037] After completing the local maximum search, the amplitude of all sampling points within the ranging gate index interval is extracted and the amplitude median is calculated. The amplitude median is used to characterize the background level within the gate and is robust to a small number of strong peaks. An amplitude threshold is set based on the amplitude median. The amplitude threshold is preferably the product of the amplitude median and a preset in-gate significance coefficient. The in-gate significance coefficient is preferably two to six to maintain adjustable detection sensitivity in weak echo and strong clutter environments and suppress the false retention of noise spikes. Finally, threshold filtering is performed on the set of local maximum points, retaining local maximum points with amplitudes greater than the amplitude threshold and sorting their indices in ascending order to form a candidate peak set.

[0038] S4. Determine the equally spaced distribution of the sidelobe clusters in the relevant domain based on the comb tooth spacing, and calculate the energy of the sidelobe clusters corresponding to each candidate peak in the candidate peak set based on the equally spaced distribution. In embodiments of the present invention, the equal-interval distribution of sidelobe clusters in the relevant domain is determined based on the comb tooth spacing, and the energy of the sidelobe clusters corresponding to each candidate peak in the candidate peak set is calculated based on the equal-interval distribution, including: The sidelobe distance of the correlation domain is calculated based on the reciprocal of the comb tooth spacing and the sampling period. For each candidate peak position in the candidate peak set, several sampling points are selected on the left and right sides of the candidate peak position with an offset of an integer multiple of the side lobe distance step length. Calculate the sum of squared amplitudes of several sampling points in the relevant sequence, and use it as the energy of the sidelobe cluster corresponding to the candidate peak; It should be noted that sidelobe cluster energy is an energy-type statistical quantity used to characterize the intensity of equally spaced peak groups formed by sidelobe structures around a candidate peak. It is obtained by selecting several sampling points on the left and right sides with the candidate peak position as the center and the sidelobe dispersion step size of the correlation domain as the interval, and summing the squares of the amplitudes of these sampling points in the correlation sequence. Thus, the overall energy level of the sidelobe cluster in the neighborhood of the candidate peak is expressed as a single value. The larger the sidelobe cluster energy, the stronger the sidelobe cluster of peaks belonging to the same category as the candidate peak and the worse the amplitude contrast between the main peak and the sidelobe. When the wet radome causes comb-like ripples in the band, the sidelobe cluster energy often increases significantly with the rise of the sidelobe cluster. Therefore, it can be used for subsequent peak suppression scoring to penalize candidate peaks that are susceptible to the influence of sidelobe clusters, so as to reduce the degradation of ranging accuracy caused by misselection of sidelobe peaks.

[0039] In detail, the comb tooth spacing and the sampling period of the radar system are obtained. The reciprocal of the comb tooth spacing is used as the repetition rate of the comb-like undulations in the band on the frequency axis. The characteristic time scale corresponding to this repetition rate is used to derive the equally spaced structure of the sidelobe clusters in the correlation domain. Specifically, one is divided by the comb tooth spacing to obtain the comb tooth period value, and this comb tooth period value is divided by the sampling period to obtain the interval counted by the sampling points. The result is then rounded to obtain the sidelobe offset step size of the correlation domain, so that the sidelobe offset step size corresponds to the adjacent interval of the sidelobe peaks in the correlation sequence. Then, each candidate peak position in the candidate peak set is traversed and recorded as the candidate peak index value. The sidelobe cluster order is set to limit the left and right sides. The number of side sampling points and the order of the sidelobe cluster are preferably two to six to cover the main energy distribution of the sidelobe peaks and avoid introducing irrelevant clutter energy by sampling too much. For each candidate peak index value, sampling points are selected on the right and left sides of the candidate peak index value by using an integer multiple of the sidelobe offset as the offset. The right sampling point index is the candidate peak index value plus one offset, plus two offsets, and so on until the offset is multiple of the sidelobe cluster order. The left sampling point index is the candidate peak index value minus one offset, minus two offsets, and so on until the offset is multiple of the sidelobe cluster order. When the sampling point index exceeds the ranging gate index interval, it is preferable to skip the sampling point to keep the energy statistics from only the effective interval within the gate.

[0040] After obtaining several sampling points on the left and right sides, the amplitude corresponding to these sampling points is read from the relevant sequence and the amplitude square is calculated respectively. Then, all amplitude squares are summed to obtain the amplitude sum and the amplitude sum is determined as the sidelobe cluster energy corresponding to the candidate peak. The sidelobe cluster energy reflects the intensity of the sidelobe peaks distributed at equal intervals around the candidate peak. When the wet radome causes the sidelobe peaks to rise, the sidelobe cluster energy will increase significantly and can be used for subsequent peak suppression scoring to penalize and screen candidate peaks.

[0041] S5. Calculate the peak suppression score of each candidate peak based on ripple intensity and sidelobe cluster energy, and take the candidate peak with the largest peak suppression score as the final peak position to calculate the target distance value. In an embodiment of the present invention, the peak suppression score of each candidate peak is calculated based on ripple intensity and sidelobe cluster energy, and the candidate peak with the largest peak suppression score is taken as the final peak position to calculate the target distance value, including: Obtain the preset base penalty coefficient and dynamic penalty coefficient; Calculate the product of ripple intensity and dynamic penalty coefficient, and add the product to the base penalty coefficient to obtain the comprehensive penalty weight; In detail, to enable the peak suppression score to adaptively change with the degree of comb-like ripple caused by the wet radome, a base penalty coefficient and a dynamic penalty coefficient are pre-stored in the radar processor as configurable parameters and read during each frame processing. The base penalty coefficient is used to apply a minimum penalty to the sidelobe cluster energy when the ripple is weak to avoid misselection of sidelobe peaks under low ripple conditions. The dynamic penalty coefficient is used to map the change in ripple intensity to the penalty enhancement amplitude to strengthen the suppression of sidelobe peak lifting when the ripple is significant. The base penalty coefficient is preferably set to 0.1 to 0.2 to balance robustness under low ripple conditions and non-significant weakening of the main peak decision, and the dynamic penalty coefficient is preferably set to zero. Points one through five are used to cover the dynamic range of ripple intensity under different radome structures and environmental wet coverage. Furthermore, offline parameter tuning can be performed by statistically analyzing the misselection rate and outlier clustering on both dry and wet-covered samples to achieve better discrimination. After reading two preset coefficients, the product of the ripple intensity and the dynamic penalty coefficient is calculated to obtain the ripple-driven penalty term. This term is then added to the basic penalty coefficient to obtain the comprehensive penalty weight. This ensures that the comprehensive penalty weight approaches the basic penalty coefficient when the ripple intensity is close to one and increases accordingly as the ripple intensity increases. This allows for the application of an adaptive penalty force to the sidelobe cluster energy in the subsequent scoring denominator construction, matching the degree of wet-covered combing.

[0042] The product of the comprehensive penalty weight and the energy of the sidelobe cluster corresponding to the candidate peak is calculated to obtain the normalized denominator; Calculate the square of the amplitude of the candidate peak, and obtain the peak suppression score of the candidate peak based on the ratio of the square of the amplitude to the normalized denominator. Specifically, the peak suppression score is a decision value used to distinguish between the true main peak and the spurious peaks formed by the uplift of sidelobe peaks within the candidate peak set. It is determined by the ratio of the square of the amplitude of the candidate peak in the relevant sequence to the normalized denominator. The normalized denominator is composed of the comprehensive penalty weight and the energy of the sidelobe clusters corresponding to the candidate peak, and the penalty intensity increases with the increase of ripple intensity. Thus, the score is higher when the amplitude of the candidate peak is large and the energy of the sidelobe clusters distributed at equal intervals around it is low, so as to favor the selection of the main peak. However, when the amplitude of the candidate peak is high but accompanied by a significant increase in the energy of the sidelobe clusters, it is significantly suppressed to suppress the misselection caused by the clustering of sidelobe peaks. Therefore, the peak suppression score can improve the stability of peak decision and improve the ranging accuracy for the sidelobe peak uplift phenomenon induced by the in-band comb ripples introduced by wet radome.

[0043] In detail, the process iterates through each candidate peak position in the candidate peak set and reads the corresponding sidelobe cluster energy. Simultaneously, it reads the amplitude at the candidate peak position from the relevant sequence as the candidate peak amplitude. For each candidate peak, the comprehensive penalty weight is multiplied by the corresponding sidelobe cluster energy to obtain a normalized denominator. This normalized denominator increases with the sidelobe cluster energy and further increases with the comprehensive penalty weight driven by the ripple intensity. This allows for a stronger penalty to be applied to candidate peaks susceptible to sidelobe cluster influence when the wet radome causes the sidelobe cluster peaks to rise. To avoid numerical instability caused by extremely small or zero sidelobe cluster energy, it is preferable to superimpose a very small positive number as a numerical stabilization term on the normalized denominator, preferably on the order of one to ten to the power of -twelfth. Subsequently, the candidate peak amplitude is squared to obtain the square of the candidate peak amplitude, making the score more sensitive to amplitude differences and consistent with the square dimension of the sidelobe cluster energy.

[0044] The ratio of the square of the amplitude of a candidate peak to the normalized denominator is used to obtain the peak suppression score of the candidate peak. The larger the score, the lower the energy of the equally spaced sidelobe clusters, indicating that the candidate peak has a higher matching amplitude and is more likely to correspond to the true main peak. The smaller the score, the stronger the sidelobe cluster peak rise around the candidate peak, indicating that it is more likely to be a sidelobe peak or clutter peak. Based on the peak suppression score, the candidate peak with the largest score can be selected from the candidate peak set as the final peak position for distance estimation.

[0045] The candidate peak with the highest peak suppression score is taken as the final peak position; The target distance value is determined based on the final peak position, sampling period, and light speed value. In detail, after calculating the peak suppression score for each candidate peak in the candidate peak set, a one-to-one correspondence is established between the position index of each candidate peak and its corresponding peak suppression score, and the scores are compared and traversed. The candidate peak with the largest peak suppression score is selected, and its position index is determined as the final peak position. When multiple candidate peaks have the same peak suppression score or the difference is less than the preset tolerance, the candidate peak with the larger amplitude or the candidate peak with the index closer to the center of the ranging gate is preferred to enhance the decision stability. After determining the final peak position, the sampling period and the speed of light of the radar system are obtained. The final peak position is multiplied by the sampling period to obtain the corresponding time scale, and the time scale is converted into spatial distance using the distance conversion relationship of electromagnetic wave round-trip propagation. Specifically, the speed of light is multiplied by the product of the final peak position and the sampling period, and then divided by two to obtain the target distance value. The target distance value is made consistent with the best matching position of the echo in the correlation sequence, so as to output the distance estimation result after peak suppression decision to reduce the adverse effect of sidelobe peak lifting caused by wet radome on ranging accuracy.

[0046] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for improving radar ranging accuracy based on high-frequency narrow pulses, characterized in that, Includes the following steps: S1. Collect the radar received sequence, and perform gated interception of the radar received sequence according to the preset ranging range to obtain the sequence within the gate; S2. Perform frequency domain transformation and cepstral analysis on the in-gate sequence, determine the comb spacing by searching for the main peak in the cepstral domain, and calculate the ripple intensity based on the amplitude of the main cepstral peak. S3. Construct a reverse combing frequency domain weight sequence based on ripple intensity, and perform weighted matched filtering on the in-gate sequence using the reverse combing frequency domain weight sequence to obtain the relevant sequence and extract the candidate peak set; S4. Determine the equally spaced distribution of the sidelobe clusters in the relevant domain based on the comb tooth spacing, and calculate the energy of the sidelobe clusters corresponding to each candidate peak in the candidate peak set based on the equally spaced distribution. S5. Calculate the peak suppression score of each candidate peak based on ripple intensity and sidelobe cluster energy, and take the candidate peak with the largest peak suppression score as the final peak position to calculate the target distance value.

2. The radar ranging accuracy improvement method based on high-frequency narrow pulse according to claim 1, characterized in that, Gated interception of radar received sequences, including: Based on the radar system's sampling period, preset ranging range, and light speed value, calculate the ranging gate start index and ranging gate end index respectively. Generate a gated window sequence with the same length as the radar received sequence; The sampling points in the gated window sequence between the starting index and the ending index of the ranging gate are assigned non-zero window coefficient values, and the remaining sampling points in the gated window sequence are assigned zero values. The radar received sequence is multiplied point by point with the gated window sequence to obtain the in-gate sequence.

3. The radar ranging accuracy improvement method based on high-frequency narrow pulse according to claim 1, characterized in that, Frequency domain transformation and cepstral analysis are performed on the in-gate sequence. The comb spacing is determined by searching for the main peak in the cepstral domain, including: Perform a fast Fourier transform on the sequence within the gate and calculate the logarithmic amplitude spectrum; The cepstral sequence is obtained by performing an inverse Fourier transform on the logarithmic amplitude spectrum; The cepstral search interval is deduced by inversely from the estimated range of comb tooth spacing; Within the cepstral search interval, the point with the largest modulus is taken as the cepstral main peak, and the comb spacing is calculated based on the position index of the cepstral main peak and the sampling period.

4. The radar ranging accuracy improvement method based on high-frequency narrow pulse according to claim 3, characterized in that, The ripple intensity is calculated based on the amplitude of the cepstral main peak, including: Calculate the average modulus of all points within the cepstral search interval, and determine the ripple intensity by the ratio of the modulus of the main cepstral peak to the average modulus.

5. The radar ranging accuracy improvement method based on high-frequency narrow pulse according to claim 1, characterized in that, Constructing a frequency domain weight sequence for the anti-comb ripple based on ripple intensity, including: Calculate the amplitude spectrum of the sequence within the gate, and perform a moving average on the amplitude spectrum to obtain a smoothed baseline spectrum; The ripple ratio sequence is obtained by calculating the ratio of the amplitude spectrum to the smoothed baseline spectrum; The exponential parameter is determined based on the ripple intensity. The ripple ratio sequence is then subjected to a power operation with the exponential parameter as the power and the reciprocal is taken to obtain the reverse comb ripple frequency domain weight sequence.

6. The radar ranging accuracy improvement method based on high-frequency narrow pulse according to claim 1, characterized in that, Weighted matched filtering of the in-gate sequence includes: Obtain the spectrum of the transmitted pulse template; The weighted correlation spectrum is obtained by multiplying the spectrum of the anti-comb pattern frequency domain weight sequence, the spectrum of the in-gate sequence, and the conjugate of the transmitted pulse template spectrum. The inverse Fourier transform of the weighted correlation spectrum yields the correlation sequence.

7. The radar ranging accuracy improvement method based on high-frequency narrow pulse according to claim 2, characterized in that, Extract the candidate peak set, including: Within the range gate index interval, search for local maxima in the relevant sequences; Calculate the median amplitude of the relevant sequence within the range gate index interval, and set an amplitude threshold based on the median amplitude; Local maxima points with amplitudes greater than the amplitude threshold are retained to form a candidate peak set.

8. The radar ranging accuracy improvement method based on high-frequency narrow pulse according to claim 1, characterized in that, Calculate the sidelobe cluster energy corresponding to each candidate peak in the candidate peak set, including: The sidelobe distance of the correlation domain is calculated based on the reciprocal of the comb tooth spacing and the sampling period. For each candidate peak position in the candidate peak set, several sampling points are selected on the left and right sides of the candidate peak position with an offset of an integer multiple of the side lobe distance step length. Calculate the sum of squared amplitudes of several sampling points in the relevant sequence, and use it as the energy of the sidelobe cluster corresponding to the candidate peak.

9. The method for improving radar ranging accuracy based on high-frequency narrow pulses according to claim 1, characterized in that, The peak suppression score for each candidate peak is calculated based on ripple intensity and sidelobe cluster energy, including: Obtain the preset base penalty coefficient and dynamic penalty coefficient; Calculate the product of ripple intensity and dynamic penalty coefficient, and add the product to the base penalty coefficient to obtain the comprehensive penalty weight; The product of the comprehensive penalty weight and the energy of the sidelobe cluster corresponding to the candidate peak is calculated to obtain the normalized denominator; Calculate the squared amplitude of the candidate peak, and obtain the peak suppression score of the candidate peak based on the ratio of the squared amplitude to the normalized denominator.

10. The radar ranging accuracy improvement method based on high-frequency narrow pulse according to claim 1, characterized in that, The candidate peak with the highest peak suppression score is used as the final peak position to calculate the target distance value, including: The candidate peak with the highest peak suppression score is taken as the final peak position; The target distance is determined based on the final peak position, sampling period, and speed of light.

Images