Fmcw radar ranging method, system, and medium based on power-weighted centroid

Abstract

A FMCW radar ranging method, system and medium based on power-weighted centroid, which comprises obtaining the beat signal of the FMCW radar and performing Fourier transform thereon to obtain an initial distance spectrum composed of multiple distance units; performing interpolation processing on at least two distance units corresponding to a preset near distance range in the initial distance spectrum to generate a refined distance spectrum with a higher frequency resolution than the initial distance spectrum; performing power-weighted centroid calculation on a continuous frequency spectrum interval corresponding to the preset near distance range based on the refined distance spectrum to obtain a first distance estimation value; performing peak value detection on the initial distance spectrum to obtain a second distance estimation value; performing time domain filtering processing on the first distance estimation values and the second distance estimation values obtained from continuous multiple frames of signals respectively; obtaining a target distance value based on the smoothed first distance estimation value and the smoothed second distance estimation value according to a preset fusion rule, and outputting.

Description

Technical Field

[0001] This invention relates to the field of distance detection technology, and in particular to an FMCW radar ranging method, system and medium based on power-weighted centroid. Background Technology

[0002] Frequency Modulated Continuous Wave (FMCW) radar calculates target range by transmitting a linear frequency modulated signal and analyzing the frequency difference between the echo and the transmitted signal. However, within the controlled frequency band, the radar bandwidth cannot be increased indefinitely, which limits its range resolution.

[0003] For targets at normal distances, peak detection methods based on Fourier transform are effective and can achieve good ranging accuracy. To further improve ranging accuracy, the industry has developed various improvement methods, such as parabolic interpolation of the spectral peaks and their adjacent frequencies to achieve sub-frequency beat frequency estimation; or using Chirp-Z transform to obtain higher resolution spectra within the frequency band of interest.

[0004] However, when the target is at extremely close range (e.g., tens of centimeters), its beat frequency signal is extremely low, corresponding to the lowest few frequency points in the Fourier transform spectrum. In practical applications, the direct coupling between the radar's transmitted and received signals, as well as surrounding static clutter, generates a strong DC component. This component forms a high energy peak near zero frequency, easily overwhelming the echo signal of nearby close-range targets. At this time, it may be impossible to form a recognizable target peak in the spectrum, or the peak may be completely covered by the DC component, making it difficult for traditional peak detection methods to achieve accurate range estimation, forming the so-called "near-field blind zone." If a high-pass filter is used to suppress the DC component, it will destroy the integrity of the low-frequency signal, causing the range estimation to shift to a more distant location. Therefore, in the blind zone at extremely close range, traditional ranging methods that rely on the presence of peaks face challenges.

[0005] In summary, existing technologies lack effective FMCW radar ranging methods in the extremely close-range blind zone. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides an FMCW radar ranging method based on power-weighted centroids, comprising the following steps:

[0007] The beat frequency signal of the FMCW radar is acquired and subjected to Fourier transform to obtain the initial range spectrum composed of multiple range cells; Interpolation processing is performed on at least two distance units corresponding to a preset near-range range in the initial distance spectrum to generate a refined distance spectrum with a higher frequency resolution than the initial distance spectrum; Based on the refined distance spectrum, a power-weighted centroid calculation is performed on the continuous spectrum interval corresponding to the preset near-distance range to obtain a first distance estimate; peak detection is performed on the initial distance spectrum to obtain a second distance estimate; wherein, the power-weighted centroid calculation and peak detection are performed in parallel based on the beat frequency signal of the same frame; The first distance estimate and the second distance estimate obtained from multiple consecutive frames of signal are subjected to time-domain filtering to obtain a smoothed first distance estimate and a smoothed second distance estimate. According to the preset fusion rules, the target distance value is obtained based on the smoothed first distance estimate and the smoothed second distance estimate, and then output.

[0008] Optionally, interpolation processing is performed on at least two range units in the initial range spectrum corresponding to a preset near-range range, including: The amplitudes of the initial distance unit, the first distance unit, and the second distance unit are extracted from the initial distance spectrum, and the spectrum curves with frequency intervals smaller than the corresponding frequency point intervals of the initial distance spectrum are fitted using Newton polynomial interpolation to generate a refined distance spectrum.

[0009] Optionally, based on the refined distance spectrum, a power-weighted centroid calculation is performed on the continuous spectrum intervals corresponding to the preset near-distance range to obtain a first distance estimate, including: Based on the refined distance spectrum, the centroid position is calculated for the continuous spectrum intervals corresponding to the preset near-distance range. The specific formula for calculating the centroid position is as follows: ;in, This represents the i-th distance unit. This represents the initial distance cell, n≥2. Indicates the position of the center of mass. This represents the magnitude corresponding to the i-th distance unit. The preset power exponent; The first distance estimate is calculated by multiplying the centroid position by the distance resolution corresponding to each distance cell.

[0010] Optionally, peak detection is performed on the initial distance spectrum to obtain a second distance estimate, including: The maximum peak value is searched on the initial distance spectrum. The amplitude of the maximum peak value and its adjacent frequency points is subjected to parabolic interpolation to obtain the beat frequency at the sub-frequency level, and then converted into a second distance estimate.

[0011] Optionally, the first distance estimate and the second distance estimate obtained from multiple consecutive frames of signal are subjected to time-domain filtering to obtain a smoothed first distance estimate and a smoothed second distance estimate. Specifically, for each frame of signal, the first distance estimate and the second distance estimate are calculated using a sliding window to obtain the mean or median, respectively, to obtain the smoothed first distance estimate and the smoothed second distance estimate.

[0012] Optionally, according to a preset fusion rule, the target distance value is obtained based on the smoothed first distance estimate and the smoothed second distance estimate, including: If the smoothed first distance estimate is less than or equal to the first threshold, then the smoothed first distance estimate is taken as the target distance value. If the smoothed first distance estimate is greater than the second threshold, then the smoothed second distance estimate is taken as the target distance value. If the smoothed first distance estimate is greater than the first threshold but less than or equal to the second threshold, and the smoothed second distance estimate is greater than the third threshold, then the smoothed second distance estimate will be used as the target distance value. If the smoothed first distance estimate is greater than the first threshold but less than or equal to the second threshold, and the smoothed second distance estimate is less than or equal to the third threshold, then a weighted fusion is performed based on the smoothed first distance estimate and the smoothed second distance estimate to obtain the target distance value. The first threshold, the second threshold, and the third threshold are preset according to the range resolution of the FMCW radar.

[0013] Optionally, the weights of the weighted fusion are dynamically adjusted based on the variance of the smoothed first distance estimate and the smoothed second distance estimate within the time window, with the side with smaller variance corresponding to a higher weight.

[0014] Optionally, multi-target processing steps may also be included: When multiple local maxima are detected in the portion of the refined distance spectrum corresponding to the preset near-distance range, a local spectrum interval is defined with each local maxima as the center. The power-weighted centroid calculation is performed independently within each local interval to obtain the first distance estimate for multiple near targets.

[0015] Corresponding to the aforementioned FMCW radar ranging method based on power-weighted centroids, this invention provides an FMCW radar ranging system based on power-weighted centroids, comprising: The spectrum acquisition module is used to acquire the beat frequency signal of the FMCW radar and perform Fourier transform on it to obtain the initial range spectrum composed of multiple range cells. The spectrum refinement module is used to interpolate at least two distance units corresponding to a preset near-distance range in the initial distance spectrum to generate a refined distance spectrum with a higher frequency resolution than the initial distance spectrum. The parallel estimation module includes a first distance estimation unit and a second distance estimation unit. The first distance estimation unit is used to perform power-weighted centroid calculation on a continuous spectrum interval corresponding to a preset near-distance range based on a refined distance spectrum to obtain a first distance estimate. The second distance estimation unit is used to perform peak detection on the initial distance spectrum to obtain a second distance estimate. The first distance estimation unit and the second distance estimation unit perform their corresponding functions in parallel based on the beat frequency signal of the same frame. The filtering module is used to perform time-domain filtering on the first distance estimate and the second distance estimate obtained from multiple consecutive frames of signal to obtain a smoothed first distance estimate and a smoothed second distance estimate. The output module is used to obtain the target distance value based on the smoothed first distance estimate and the smoothed second distance estimate according to the preset fusion rules, and output it.

[0016] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing an FMCW radar ranging program based on power-weighted centroids, wherein the FMCW radar ranging program based on power-weighted centroids, when executed by a processor, implements the steps of the FMCW radar ranging method based on power-weighted centroids as described above.

[0017] Compared with the prior art, the present invention has the following beneficial effects: (1) By generating a refined range spectrum through interpolation, the frequency resolution of the near range is improved without increasing the radar bandwidth, so that the near target information, which was originally limited by the spectral resolution, can be presented on a more refined spectral scale; by calculating the centroid of the power weighted calculation, the energy distribution of the continuous spectral interval is weighted and summed, and the centroid of the spectral energy is used as the basis for range estimation, avoiding the dependence of the traditional peak detection method on obvious spectral peaks, so that effective range estimates can still be obtained when the echo of the near target is submerged or covered by the DC component and cannot form an identifiable peak; by executing the power weighted centroid calculation and peak detection in parallel, the same frame signal can be used for ranging requirements of both near blind zone and conventional range without the need for frame processing or mode switching; by suppressing the fluctuation of the inter-frame estimate caused by random noise through time domain filtering, the stability of the range estimate is improved; by adaptively selecting or weighting and fusing the two estimates through fusion rules, the high-precision ranging of the near blind zone and the reliable ranging of the conventional range can be taken into account, realizing high-precision range measurement in the full range.

[0018] (2) By using Newton polynomial interpolation to fit the amplitude of a finite number of discrete frequency points with polynomials, a higher resolution refined distance spectrum can be generated with a lower computational load, providing a refined data foundation for subsequent power-weighted centroid calculation.

[0019] (3) By using the power exponent, the frequency points with higher amplitude (more likely to correspond to the real position of the target) are given a greater weight contribution in the centroid calculation, while the influence of noise or interference frequency points with lower amplitude is suppressed, thereby improving the sensitivity of the centroid position to the real target position; by using the linear mapping between the centroid position and the distance resolution, the centroid position in the frequency domain is directly converted into the physical distance unit, and a stable and continuous distance estimation output can still be obtained when the spectral peak is not obvious.

[0020] (4) By refining the peak values ​​of the discrete spectrum through parabolic interpolation, the frequency resolution limitation of the FFT spectrum is broken, which significantly improves the ranging accuracy within the normal distance range. By performing local interpolation only on the maximum peak value and its adjacent frequency points, the computational overhead caused by global high-resolution spectrum calculation is avoided, and the computational complexity is reduced while ensuring ranging accuracy.

[0021] (5) The sliding window temporal smoothing mechanism effectively suppresses the abnormal jumps caused by random noise in single frame estimation, and improves the temporal continuity and stability of distance estimation. Mean filtering is suitable for Gaussian noise scenarios to maximize signal-to-noise ratio gain, and median filtering is suitable for impulse noise scenarios to suppress outlier interference, thus enhancing the system's environmental adaptability.

[0022] (6) The adaptive switching of the range estimation strategy is realized through the threshold judgment logic; the switching boundary is matched with the inherent resolution capability of the radar system by the preset association between the threshold and the range resolution, thus avoiding the jump in the estimated value or the loss of accuracy caused by improper threshold setting.

[0023] (7) Using variance as a quantitative indicator of the reliability of the estimate, the data-driven adaptive allocation of the fusion weight is realized: when a certain estimation method shows better stability (smaller variance) under the current environment or distance conditions, its weight is automatically increased, and vice versa, so that the fusion result always tends to the more reliable estimation source; through variance statistics within the time window, the statistical significance of the reliability assessment is guaranteed, and the reliability drift caused by environmental changes is tracked, realizing the dynamic optimization of the fusion strategy and improving the robustness of ranging in complex scenarios.

[0024] (8) Through the local maximum detection and partitioning processing mechanism, the power-weighted centroid calculation method can be extended from single-target scenarios to multi-target scenarios. By defining independent processing intervals centered on each local maximum, the centroid shift problem caused by the spectral energy aliasing of multiple targets is avoided, so that multiple close-range targets can be simultaneously distinguished and ranged in the same frame signal, thereby improving the multi-target resolution capability of the radar system and the application range of close-range scenarios. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a simplified flowchart of an embodiment of the FMCW radar ranging method based on power-weighted centroids of the present invention; Figure 2 This invention presents the performance of the power-weighted centroid calculation method for FMCW radar ranging based on power-weighted centroids in the range of 20-127 cm. Figure 3 This invention demonstrates the peak detection performance of the FMCW radar ranging method based on power-weighted centroids in the distance range of 20-127 cm. Figure 4 This is a schematic diagram of the smoothed effect of the first distance estimate in an embodiment of the FMCW radar ranging method based on power-weighted centroids of the present invention; Figure 5 This is a schematic diagram of the smoothed effect of the second distance estimate in an embodiment of the FMCW radar ranging method based on power-weighted centroids of the present invention; Figure 6 This is a schematic diagram of the target distance value output by an embodiment of the FMCW radar ranging method based on power-weighted centroid of the present invention; Figure 7 This is a framework diagram of an embodiment of the FMCW radar ranging system based on power-weighted centroids of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] like Figure 1As shown, the present invention provides an FMCW radar ranging method based on power-weighted centroids, which includes the following steps: The beat frequency signal of the FMCW radar is acquired and subjected to Fourier transform to obtain the initial range spectrum composed of multiple range cells; Interpolation is performed on at least two range units in the initial range spectrum that correspond to a preset near range range to generate a refined range spectrum with a higher frequency resolution than the initial range spectrum. Based on the refined distance spectrum, a power-weighted centroid calculation is performed on the continuous spectrum interval corresponding to the preset near distance range to obtain a first distance estimate; peak detection is performed on the initial distance spectrum to obtain a second distance estimate; wherein, the power-weighted centroid calculation and peak detection are performed in parallel based on the beat frequency signal of the same frame; The first and second distance estimates obtained from multiple consecutive frames of signal are subjected to time-domain filtering to obtain smoothed first and second distance estimates. According to the preset fusion rules, the target distance value is obtained based on the smoothed first distance estimate and the smoothed second distance estimate, and then output.

[0028] This invention generates a refined range spectrum through interpolation, improving the frequency resolution of the near-range range without increasing radar bandwidth. This allows near-range target information, previously limited by spectral resolution, to be presented at a more refined spectral scale. By using power-weighted centroid calculation to weightedly sum the energy distribution across continuous spectral intervals, the energy centroid is used as the basis for range estimation. This avoids the dependence of traditional peak detection methods on obvious spectral peaks, thus enabling effective range estimates even when near-range target echoes are submerged or covered by DC components and cannot form identifiable peaks. Parallel execution of power-weighted centroid calculation and peak detection allows the same frame signal to be applied to both near-range blind zones and conventional ranges, eliminating the need for frame splitting or mode switching. Time-domain filtering suppresses inter-frame estimation fluctuations caused by random noise, improving the stability of range estimation. Adaptive selection or weighted fusion of the two estimates using fusion rules balances high-precision ranging in near-range blind zones with reliable ranging in conventional ranges, achieving high-precision range measurement across the entire range.

[0029] refer to Figure 2 and Figure 3 The results show the ranging results of power-weighted centroid calculation and peak detection in the range of 20-127 cm. The vertical axis represents the radar ranging result in centimeters; the horizontal axis represents the frame number / time series of continuous radar sampling; and the discrete points / lines on the curves represent the real-time ranging values ​​of the corresponding algorithms in each sampling frame.

[0030] Figure 2 Power-weighted centroid calculation is used, without relying on peak criterion, from Figure 2 The curve shows that there are no jumps or missed detections, indicating that even when the target is at a very close position of 20cm (traditional blind zone), the power-weighted centroid calculation can stably output the ranging value without numerical discontinuity or no results. This directly reflects the core advantages of the power-weighted centroid calculation, which does not rely on spectral peaks and can eliminate near-range blind zones.

[0031] Figure 3 Peak detection is used to... Figure 3 It can be seen that when the target is at a very close position of 20cm (traditional blind zone), the curve shows obvious jumps, numerical distortion, and even failure to detect. The ranging values ​​of some sampling frames deviate from the true values ​​(such as suddenly dropping to 0 or jumping sharply), which directly reflects the problem that the traditional method is affected by DC leakage and stray noise, and fails to detect weak signals at close range. When the target is close to 127cm (slightly farther), the curve is relatively stable. The overall curve has a higher degree of dispersion, and its noise resistance is weaker than the power-weighted centroid calculation method. The continuity and accuracy of target detection are relatively poor.

[0032] Existing technologies rely on peak detection, requiring information about peak points to calculate accurate distances. When a target is located between the 0th distance cell (initial distance cell) and the 1st distance cell, the peak in the spectrum may appear in the 0th distance cell, leading to a failure to calculate the distance due to peak search failure. Furthermore, reducing the low-frequency energy at the 0th distance cell using methods such as high-pass filtering will cause the distance estimate to shift further away.

[0033] Therefore, in this embodiment, interpolation processing is performed on at least two range units in the initial range spectrum corresponding to a preset near-range range, including: The amplitudes of the initial distance unit, the first distance unit, and the second distance unit are extracted from the initial distance spectrum, and the spectrum curves with frequency intervals smaller than the corresponding frequency point intervals of the initial distance spectrum are fitted using the Newton polynomial interpolation method to generate a refined distance spectrum.

[0034] This invention uses Newton's polynomial interpolation to perform polynomial fitting on the amplitude of a finite number of discrete frequency points, which can generate a higher resolution refined distance spectrum with lower computational cost, providing a refined data foundation for subsequent power-weighted centroid calculation.

[0035] In this embodiment, based on the refined distance spectrum, a power-weighted centroid calculation is performed on continuous spectral intervals corresponding to a preset near-distance range to obtain a first distance estimate, including: Based on the refined distance spectrum, the centroid position is calculated for continuous spectral intervals corresponding to a preset near-distance range. The specific formula for calculating the centroid position is as follows: ;in, This represents the i-th distance unit. This represents the initial distance cell, n≥2. Indicates the position of the center of mass. This represents the magnitude corresponding to the i-th distance unit. The preset power exponent; The first distance estimate is calculated by multiplying the centroid position by the distance resolution corresponding to each distance cell.

[0036] In this embodiment, the power exponent is mainly used to adjust the contribution of different distance unit amplitudes to the final centroid result. The larger the power exponent, the more it is biased towards low-frequency points with high energy. When the power exponent is small, the weight difference of each distance unit in the centroid calculation is relatively gentle, which is beneficial to obtaining smooth distance estimation results when the energy distribution is relatively dispersed or the signal fluctuation is large. When the power exponent is large, high-amplitude distance units occupy a higher weight in the centroid calculation, thereby highlighting the echo component of the main target and suppressing the influence of low-energy noise or sidelobe components on the estimation result.

[0037] Preferably, the preset close range is within one range resolution unit of the FMCW radar (extremely close range blind zone), which is extended to two range resolution units. Based on the actual parameters of a typical 24G FMCW radar, the core physical distance is 0-70 cm, and the extended range is 0-140 cm.

[0038] This invention uses a power-law approach to give higher-amplitude frequencies (more likely corresponding to the target's true location) greater weight in centroid calculation, while suppressing the influence of lower-amplitude noise or interference frequencies, thus improving the sensitivity of the centroid position to the true target position. Through a linear mapping between the centroid position and range resolution, the frequency-domain centroid position is directly converted into physical distance units, achieving stable and continuous range estimation output even when spectral peaks are not prominent. Specifically, the amplitudes of several adjacent frequencies in the refined range spectrum are used as weights, and a weighted average is used to calculate the fine frequency corresponding to the spectral centroid, thereby estimating the target's sub-frequency range value. This soft centroid estimation fully utilizes spectral energy distribution, effectively reducing the ranging blind zone, and does not rely on spectral peaks for judgment. In this embodiment, peak detection is performed on the initial distance spectrum to obtain a second distance estimate, including: The maximum peak value is searched on the initial distance spectrum. The amplitude of the maximum peak value and its adjacent frequency points is parabolically interpolated to obtain the beat frequency at the sub-frequency level, and then converted into the second distance estimate.

[0039] In this embodiment, the sub-frequency beat frequency is substituted into the following formula to convert it into a second distance estimate: ; in, This is the second distance estimate, where c is the speed of light and S is the radar frequency modulation slope (unit: Hz / s). This refers to the beat frequency at the sub-frequency level.

[0040] This invention refines the sub-frequency points of discrete spectrum peaks through parabolic interpolation, breaking through the frequency resolution limitation of FFT spectrum and significantly improving ranging accuracy within the conventional distance range. By performing local interpolation only on the maximum peak and its adjacent frequency points, the computational overhead caused by global high-resolution spectrum calculation is avoided, thus reducing computational complexity while ensuring ranging accuracy.

[0041] In this embodiment, the first and second distance estimates obtained from multiple consecutive frames of signal are subjected to time-domain filtering to obtain smoothed first and second distance estimates. Specifically, for each frame of signal, the mean or median of the first and second distance estimates is calculated using a sliding window to obtain a smoothed first distance estimate (the effect can be referenced). Figure 4 ) and the smoothed second distance estimate (the effect can be referenced) Figure 5 ).

[0042] This invention effectively suppresses abnormal jumps caused by random noise in single-frame estimation through a sliding window temporal smoothing mechanism, thereby improving the temporal continuity and stability of distance estimation. Mean filtering is suitable for Gaussian noise scenarios to maximize signal-to-noise ratio gain, while median filtering is suitable for impulse noise scenarios to suppress outlier interference, thus enhancing the system's environmental adaptability.

[0043] Under normal circumstances, the centroid method can provide continuous and smooth results even for weak signals at very close range, while the conventional peak method may produce jumps or misses when the echo energy at close range is weakened by filtering. Conversely, when the target is slightly farther away or the signal is not distorted, the estimate of the conventional method is closer to the truth. Therefore, by comparing whether the two results fall within their respective confidence intervals, the more reliable one can be selected as the output distance of the frame, or a combination of the two methods can be used according to weights. In this embodiment, according to the preset fusion rule, the target distance value is obtained based on the smoothed first distance estimate and the smoothed second distance estimate, including: If the smoothed first distance estimate is less than or equal to the first threshold (the target is very close and belongs to the traditional blind zone), then the smoothed first distance estimate is used as the target distance value. If the smoothed first distance estimate is greater than the second threshold (the target is outside the near-field blind zone), then the smoothed second distance estimate is used as the target distance value. If the smoothed first distance estimate is greater than the first threshold, but less than or equal to the second threshold (the target is at the boundary between the close range and the conventional detection area), and the smoothed second distance estimate is greater than the third threshold, then the smoothed second distance estimate will be used as the target distance value. If the smoothed first distance estimate is greater than the first threshold but less than or equal to the second threshold, and the smoothed second distance estimate is less than or equal to the third threshold, then a weighted fusion is performed based on the smoothed first distance estimate and the smoothed second distance estimate to obtain the target distance value. The first, second, and third thresholds are preset based on the range resolution of the FMCW radar. Preferably, the first threshold is 0.5, and the second and third thresholds are both 1, with the unit of the thresholds being a range unit.

[0044] Figure 6 For the target distance value output by this invention, please refer to the following: for close range, the smoothed first distance estimate is used as the target distance value, which has no blind spots and no jumps, and can stably detect very close targets; for medium and long range, the smoothed second distance estimate is used as the target distance value, which inherits the advantages of conventional methods for medium and long range ranging, and the smooth transition of weighted fusion in the boundary interval reflects the advantages of fusion decision.

[0045] This invention achieves adaptive switching of range estimation strategies through threshold judgment logic; by presetting the association between the threshold and range resolution, the switching boundary is matched with the inherent resolution capability of the radar system, avoiding estimation value jumps or accuracy loss caused by improper threshold settings.

[0046] In this embodiment, the weights of the weighted fusion are dynamically adjusted based on the variance of the smoothed first distance estimate and the smoothed second distance estimate within the time window, with the side with smaller variance corresponding to a higher weight.

[0047] This invention uses variance as a quantitative indicator of the reliability of the estimate, and realizes data-driven adaptive allocation of fusion weights: when an estimation method shows better stability (smaller variance) under the current environment or distance conditions, its weight is automatically increased, and vice versa, so that the fusion result always tends to the more reliable estimation source; through variance statistics within the time window, it not only ensures the statistical significance of the reliability assessment, but also tracks the reliability drift caused by environmental changes, realizes the dynamic optimization of the fusion strategy, and improves the robustness of ranging in complex scenarios.

[0048] In this embodiment, a multi-target processing step is also included: When multiple local maxima are detected in the part of the distance spectrum corresponding to the preset near range, the local spectrum interval is defined with each local maximum as the center. Power-weighted centroid calculations are performed independently within each local interval to obtain the first distance estimates for multiple near targets.

[0049] This invention extends the power-weighted centroid calculation method from single-target scenarios to multi-target scenarios through local maximum detection and partitioning processing mechanisms. By defining independent processing intervals centered on each local maximum, it avoids the centroid shift problem caused by spectral energy aliasing among multiple targets, enabling multiple near-range targets to be simultaneously resolved and ranged in the same signal frame. This improves the multi-target resolution capability of the radar system and the application range in near-range scenarios. For long-range target detection, traditional FFT peak detection can be used in parallel. This invention is particularly suitable for application scenarios with near-range targets and requiring high-precision measurements.

[0050] like Figure 7 As shown, the present invention also provides an FMCW radar ranging system based on power-weighted centroids, which includes: The spectrum acquisition module 10 is used to acquire the beat frequency signal of the FMCW radar and perform Fourier transform on it to obtain an initial range spectrum composed of multiple range cells. The spectrum refinement module 20 is used to interpolate at least two distance units corresponding to a preset near-distance range in the initial distance spectrum to generate a refined distance spectrum with a higher frequency resolution than the initial distance spectrum. The parallel estimation module 30 includes a first distance estimation unit 31 and a second distance estimation unit 32. The first distance estimation unit 31 is used to perform power-weighted centroid calculation on a continuous spectrum interval corresponding to a preset near-distance range based on a refined distance spectrum to obtain a first distance estimate. The second distance estimation unit 32 is used to perform peak detection on the initial distance spectrum to obtain a second distance estimate. The first distance estimation unit 31 and the second distance estimation unit 32 perform their corresponding functions in parallel based on the beat frequency signal of the same frame. The filtering module 40 is used to perform time-domain filtering on the first distance estimate and the second distance estimate obtained from multiple consecutive frames of signal to obtain a smoothed first distance estimate and a smoothed second distance estimate. The output module 50 is used to obtain the target distance value based on the smoothed first distance estimate and the smoothed second distance estimate according to the preset fusion rules, and then output it.

[0051] This invention also provides a computer-readable storage medium, which may be a computer-readable storage medium included in the memory described in the above embodiments; or it may be a standalone computer-readable storage medium not assembled into a device. The computer-readable storage medium stores at least one instruction, which is loaded and executed by a processor to implement... Figure 1The illustrated FMCW radar ranging method is based on power-weighted centroids. The computer-readable storage medium can be a read-only memory, a hard disk, or an optical disk, etc.

[0052] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For system embodiments and storage medium embodiments, since they are basically similar to method embodiments, the descriptions are relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0053] Furthermore, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0054] The foregoing description illustrates and describes preferred embodiments of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the inventive concept by means of the foregoing teachings or techniques or knowledge in related fields. Any modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. An FMCW radar ranging method based on power-weighted centroids, characterized in that, Includes the following steps: The beat frequency signal of the FMCW radar is acquired and subjected to Fourier transform to obtain the initial range spectrum composed of multiple range cells; Interpolation processing is performed on at least two distance units corresponding to a preset near-range range in the initial distance spectrum to generate a refined distance spectrum with a higher frequency resolution than the initial distance spectrum; Based on the refined distance spectrum, a power-weighted centroid calculation is performed on the continuous spectrum interval corresponding to the preset near-distance range to obtain a first distance estimate; peak detection is performed on the initial distance spectrum to obtain a second distance estimate; wherein, the power-weighted centroid calculation and peak detection are performed in parallel based on the beat frequency signal of the same frame; The first distance estimate and the second distance estimate obtained from multiple consecutive frames of signal are subjected to time-domain filtering to obtain a smoothed first distance estimate and a smoothed second distance estimate. According to the preset fusion rules, the target distance value is obtained based on the smoothed first distance estimate and the smoothed second distance estimate, and then output.

2. The FMCW radar ranging method based on power-weighted centroids according to claim 1, characterized in that, Interpolation processing is performed on at least two range cells in the initial range spectrum corresponding to a preset near-range range, including: The amplitudes of the initial distance unit, the first distance unit, and the second distance unit are extracted from the initial distance spectrum, and the spectrum curves with frequency intervals smaller than the corresponding frequency point intervals of the initial distance spectrum are fitted using Newton polynomial interpolation to generate a refined distance spectrum.

3. The FMCW radar ranging method based on power-weighted centroids according to claim 1, characterized in that, Based on the refined distance spectrum, a power-weighted centroid calculation is performed on the continuous spectrum intervals corresponding to the preset near-distance range to obtain a first distance estimate, including: Based on the refined distance spectrum, the centroid position is calculated for the continuous spectrum intervals corresponding to the preset near-distance range. The specific formula for calculating the centroid position is as follows: ;in, This represents the i-th distance unit. This represents the initial distance cell, n≥2. Indicates the position of the center of mass. This represents the magnitude corresponding to the i-th distance unit. The preset power exponent; The first distance estimate is calculated by multiplying the centroid position by the distance resolution corresponding to each distance cell.

4. The FMCW radar ranging method based on power-weighted centroids according to claim 1, characterized in that, Peak detection is performed on the initial distance spectrum to obtain a second distance estimate, including: The maximum peak value is searched on the initial distance spectrum. The amplitude of the maximum peak value and its adjacent frequency points is subjected to parabolic interpolation to obtain the beat frequency at the sub-frequency level, and then converted into a second distance estimate.

5. The FMCW radar ranging method based on power-weighted centroids according to claim 1, characterized in that, The first distance estimate and the second distance estimate obtained from multiple consecutive frames of signal are subjected to time-domain filtering to obtain a smoothed first distance estimate and a smoothed second distance estimate. Specifically, for the first distance estimate and the second distance estimate obtained from each frame of signal, the mean or median is calculated using a sliding window to obtain a smoothed first distance estimate and a smoothed second distance estimate.

6. The FMCW radar ranging method based on power-weighted centroids according to claim 1, characterized in that, According to a preset fusion rule, the target distance value is obtained based on the smoothed first distance estimate and the smoothed second distance estimate, including: If the smoothed first distance estimate is less than or equal to the first threshold, then the smoothed first distance estimate is taken as the target distance value. If the smoothed first distance estimate is greater than the second threshold, then the smoothed second distance estimate is taken as the target distance value. If the smoothed first distance estimate is greater than the first threshold but less than or equal to the second threshold, and the smoothed second distance estimate is greater than the third threshold, then the smoothed second distance estimate will be used as the target distance value. If the smoothed first distance estimate is greater than the first threshold but less than or equal to the second threshold, and the smoothed second distance estimate is less than or equal to the third threshold, then a weighted fusion is performed based on the smoothed first distance estimate and the smoothed second distance estimate to obtain the target distance value. The first threshold, the second threshold, and the third threshold are preset according to the range resolution of the FMCW radar.

7. The FMCW radar ranging method based on power-weighted centroids according to claim 6, characterized in that, The weights of the weighted fusion are dynamically adjusted based on the variance of the smoothed first distance estimate and the smoothed second distance estimate within the time window, with the side with smaller variance corresponding to a higher weight.

8. The FMCW radar ranging method based on power-weighted centroids according to claim 1, characterized in that, It also includes multi-target processing steps: When multiple local maxima are detected in the portion of the refined distance spectrum corresponding to the preset near-distance range, a local spectrum interval is defined with each local maxima as the center. The power-weighted centroid calculation is performed independently within each local interval to obtain the first distance estimate for multiple near targets.

9. An FMCW radar ranging system based on power-weighted centroids, characterized in that, include: The spectrum acquisition module is used to acquire the beat frequency signal of the FMCW radar and perform Fourier transform on it to obtain the initial range spectrum composed of multiple range cells. The spectrum refinement module is used to interpolate at least two distance units corresponding to a preset near-distance range in the initial distance spectrum to generate a refined distance spectrum with a higher frequency resolution than the initial distance spectrum. The parallel estimation module includes a first distance estimation unit and a second distance estimation unit. The first distance estimation unit is used to perform power-weighted centroid calculation on a continuous spectrum interval corresponding to a preset near-distance range based on a refined distance spectrum to obtain a first distance estimate. The second distance estimation unit is used to perform peak detection on the initial distance spectrum to obtain a second distance estimate. The first distance estimation unit and the second distance estimation unit perform their corresponding functions in parallel based on the beat frequency signal of the same frame. The filtering module is used to perform time-domain filtering on the first distance estimate and the second distance estimate obtained from multiple consecutive frames of signal to obtain a smoothed first distance estimate and a smoothed second distance estimate. The output module is used to obtain the target distance value based on the smoothed first distance estimate and the smoothed second distance estimate according to the preset fusion rules, and output it.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an FMCW radar ranging program based on power-weighted centroids, which, when executed by a processor, implements the steps of the FMCW radar ranging method based on power-weighted centroids as described in any one of claims 1 to 8.

Images