CN-121995331-A - Altimeter signal detection method and system based on dynamic parameter OS-CFAR

CN121995331ACN 121995331 ACN121995331 ACN 121995331ACN-121995331-A

Abstract

A altimeter signal detection method and system based on dynamic parameter OS-CFAR, the method collects radar signal and carries on the preconditioning at first, obtain the frequency spectrum data, and calculate frequency spectrum entropy and signal-to-noise ratio; then, a two-dimensional feature vector is formed as input of a Gaussian mixture model, an offline training stage and an online execution stage are carried out, and an environment self-adaptive OS-CFAR parameter combination is generated; then calculating a detection threshold factor based on the parameter combination, and identifying a target signal; finally, carrying out multi-dimensional target verification on the target signal to obtain an altimeter signal; according to the invention, by constructing the mixed characteristic of spectrum entropy and signal-to-noise ratio, the defect that the traditional single power characteristic is difficult to distinguish strong targets from distributed interference is overcome, and the environment characteristic sensing and parameter dynamic mapping mechanism is introduced, so that the detection parameters are self-adaptively adjusted along with the environment, and meanwhile, the mode of off-line training and on-line reasoning is adopted, so that the self-adaptive capacity of the environment is ensured, and meanwhile, the calculation efficiency and stability are also considered.

Inventors

MA JIE
FENG XIAOXUAN
HU PANPAN
ZHANG SHIYING
ZHANG QI
QIU SHIJIN

Assignees

武汉理工大学
武汉万集光电技术有限公司

Dates

Publication Date: 20260508
Application Date: 20251217

Claims (10)

1. The altimeter signal detection method based on the dynamic parameter OS-CFAR is characterized by comprising the following steps of: collecting radar signals, preprocessing to obtain spectrum data, and calculating spectrum entropy and signal-to-noise ratio based on the spectrum data; the spectrum entropy and the signal-to-noise ratio are formed into two-dimensional feature vectors to be used as the input of a Gaussian mixture model, and an offline training stage and an online execution stage are carried out to generate an environment self-adaptive OS-CFAR parameter combination; Calculating a detection threshold factor based on the environment self-adaptive OS-CFAR parameter combination, and carrying out point-by-point detection on the spectrum data based on the detection threshold factor to identify a target signal; And after the target signal is subjected to multi-dimensional target verification, outputting a final detection result to obtain the altimeter signal.
2. The dynamic parameter OS-CFAR based altimeter signal detection method of claim 1, wherein: The calculating the spectrum entropy and the signal-to-noise ratio specifically comprises the following steps: the spectral entropy is calculated, and the expression is as follows: ; ; Wherein: For the spectral entropy of the light, Normalized power spectral density; Is the first Power spectral density of the individual frequency points; The sequence number of the frequency point is calculated currently; A traversal index in the summation operation; The total frequency point number of the spectrum data; The signal-to-noise ratio is calculated, which comprises the following steps: Performing self-adaptive sliding window smoothing on the power spectrum based on the spectrum data to obtain a smoothed power spectrum value, wherein the expression is as follows: ; Wherein: Is the first The power spectrum value after the smoothing of each frequency point; The total number of sample points for the sliding window; Counting the number of the windows; a sum sequence number; Is the first Original power spectrum values of the frequency points; the sequence number of the frequency point is currently processed; is the sampling rate; to smooth physical bandwidth as desired; Counting FFT points; Dividing a spectrum section with continuous spectrum width larger than 2 times of window length in the smoothed power spectrum value into a signal band, and dividing the rest spectrum sections into clutter noise bands; and respectively carrying out power integration calculation on the signal band and the clutter noise band, and obtaining the signal-clutter noise ratio through ratio operation.
3. The dynamic parameter OS-CFAR based altimeter signal detection method of claim 1, wherein: The offline training stage and the online execution stage specifically comprise: In the off-line training stage, environment clustering is carried out based on the two-dimensional feature vectors, a representative environment state is determined, mapping is carried out on the representative environment state and the corresponding ordered statistics constant false alarm OS-CFAR parameter combination, an environment-parameter mapping strategy library is constructed, and the environment-parameter mapping strategy library is solidified in a Gaussian mixture model; And in the online execution stage, extracting a characteristic vector of the current environment state, inputting the characteristic vector into the cured Gaussian mixture model, and calling parameters of an environment-parameter mapping strategy library to perform responsibility weighting and coupling constraint verification to generate an environment self-adaptive OS-CFAR parameter combination.
4. The dynamic parameter OS-CFAR based altimeter signal detection method of claim 3, wherein: the environment clustering method specifically comprises the following steps: inputting the two-dimensional feature vector into a Gaussian mixture model, and carrying out environment clustering by adopting an EM algorithm to determine a clustering center point of each representative environment state, wherein the representative environment states comprise pure, medium, complex and severe; Taking the clustering center points corresponding to the representative environment states as objects, dividing the clustering center points of the highest signal-to-noise ratio value and the lowest spectrum entropy value into pure environment states, and dividing the clustering center points of the lowest signal-to-noise ratio value and the highest spectrum entropy value into severe environment states; Between the clustering center points of the pure environment state and the severe environment state, the clustering center point with relatively high signal-to-noise ratio value is divided into medium environment states according to the centering degree, and the clustering center point with relatively low signal-to-noise ratio value is divided into complex environment states.
5. The dynamic parameter OS-CFAR based altimeter signal detection method of claim 3, wherein: the online execution stage specifically comprises the following steps: The characteristic vector of the current environment state is collected and extracted in real time, and is input into a cured Gaussian mixture model to output a responsibility vector Wherein, the method comprises the steps of, For each component The current environmental state of the characterization belongs to the first Probability weights representing environmental states; Invoking an OS-CFAR parameter combination corresponding to a state representing an environment from an environment-parameter mapping policy library And through responsibility weighting and coupling constraint verification, generating an environment self-adaptive OS-CFAR parameter combination, wherein the expression is as follows: ; ; ; Wherein: the number of training units; Is the number of protection units; An index for ordering.
6. The dynamic parameter OS-CFAR based altimeter signal detection method of claim 5, wherein: the coupling constraint specifically comprises: The training unit The value range of (2) is 16 to 64, and the step length is adjusted to be an integer multiple of 8; The protection unit The value range of (1) to (6), the step length is adjusted to be 1, and the sum of the protection window and the training window does not exceed the processing capacity of the sliding window; the ordering index The range of the value of (2) is 0.3 To 0.8 The step length is adjusted to be 1, and when At the time of increasing from 32 to 48, The upper limit of (2) is automatically adjusted from 25 to 38.
7. The dynamic parameter OS-CFAR based altimeter signal detection method of claim 6, wherein: The method for calculating the detection threshold factor based on the environment self-adaptive OS-CFAR parameter combination specifically comprises the following steps: based on preset false alarm probability Training unit And sort index The detection threshold factor is solved by inverse-pushing the accurate model in the logarithmic domain, and the expression is as follows: ; Wherein: is a detection threshold factor.
8. The dynamic parameter OS-CFAR based altimeter signal detection method of claim 7, wherein: the detecting threshold factor based point-by-point detection is carried out on the spectrum data to identify a target signal, and the method specifically comprises the following steps: the current detection unit is taken as the center, and the two sides of the spectrum data are provided with A protection unit arranged outside the protection unit Each side is selected from reference units A plurality of reference units; The power values around the reference units are sorted in ascending order based on the detection threshold factors, and the first power value is selected The minimum value of the power values is used as an estimated value of the current background noise power; based on detection threshold factors Estimation of current background noise power Calculating a detection threshold And power value of each detection unit is calculated With detection threshold Comparison is performed: If it is Judging the signal as a target signal, and recording the frequency and the amplitude of the point; If it is If the noise is determined, the recording is not performed.
9. The dynamic parameter OS-CFAR based altimeter signal detection method of claim 1, wherein: After the target signal is subjected to multi-dimensional target verification, a final detection result is output, and a altimeter signal is obtained, which specifically comprises the following steps: Performing the following target verification on the target signal, and eliminating the target signal which does not meet the verification condition: amplitude verification, namely eliminating target signals with amplitude lower than the maximum amplitude by 10%; frequency range verification, namely eliminating target signals with frequencies not within a frequency threshold; local peak value verification, namely eliminating target signals with insignificant local peak values; the historical continuity verification comprises the steps of eliminating target signals with the deviation of more than 30% from a historical detection result; and taking the target signal meeting the verification condition as a final detection result, outputting corresponding frequency and amplitude information, and converting the frequency difference into a distance according to the frequency difference to obtain the platform ground clearance information.
10. A dynamic parameter OS-CFAR based altimeter signal detection system, characterized in that it is applied to the method of any one of claims 1 to 9, said system comprising: the input vector acquisition module (1) is used for acquiring radar signals, preprocessing the radar signals to obtain spectrum data, and calculating spectrum entropy and signal-to-noise ratio based on the spectrum data; The self-adaptive parameter generation module (2) is used for forming a two-dimensional feature vector by using the spectrum entropy and the signal-to-noise ratio as the input of the Gaussian mixture model, performing an offline training stage and an online execution stage, and generating an environment self-adaptive OS-CFAR parameter combination; The target signal identification module (3) is used for calculating a detection threshold factor based on the environment self-adaptive OS-CFAR parameter combination, and carrying out point-by-point detection on the spectrum data based on the detection threshold factor to identify a target signal; And the altimeter signal acquisition module (4) is used for outputting a final detection result after the target signal is subjected to multi-dimensional target verification to obtain the altimeter signal.

Description

Altimeter signal detection method and system based on dynamic parameter OS-CFAR Technical Field The invention relates to a signal detection means, belongs to the field of target detection algorithms, and particularly relates to a altimeter signal detection method and system based on dynamic parameters OS-CFAR. Background In practical tests, complex and changeable environmental conditions often cause large fluctuation of the dynamic range of signals, saturation distortion is easy to be caused when the signal strength is too high, and effective detection is difficult to be realized when the strength is too low. The traditional OS-CFAR detector adopts a fixed parameter design, and cannot dynamically adjust a detection strategy according to environmental characteristics, so that the adaptability of the traditional OS-CFAR detector in a complex dynamic environment is seriously insufficient. Besides the limitations of the detector, the signal identification link also faces multiple interference challenges, such as false spectrum peaks formed by random noise after FFT spectrum analysis, multipath reflection signals in a low-altitude flight scene, multiple scattering center interference caused by complex terrain and the like, which can seriously influence the accuracy of height measurement and further exacerbate the difficulty of effective signal identification. The method has the advantages that the method is particularly outstanding in defects of short plates of a traditional detection scheme, the fixed threshold is difficult to adapt to clutter characteristic changes of different complex terrains such as cities, mountains and sea surfaces, the environment adaptability is poor, false detection events are easy to generate in a low-altitude multipath propagation environment, false alarm rate is difficult to effectively control, detection sensitivity is always kept constant, self-adaptive switching cannot be conducted between a pure environment and a complex environment, the CFAR technology based on deep learning is introduced and applied, environmental classification is achieved by means of a neural network, and although the method has better detection performance in the complex environment, a significant bottleneck exists in practical engineering application, on the one hand, model training depends on a large-scale labeling data set, calculation complexity is high, real-time processing requirements are difficult to meet, on the other hand, maintenance and updating flow of the model are complex, and moreover, the black box characteristic cannot meet the certainty requirement of a high-reliability system, and engineering landing of the technology is severely restricted. Therefore, a altimeter signal detection means capable of adapting to environmental changes, having low computational complexity and stable detection performance is needed to solve the above-mentioned drawbacks in the prior art. Disclosure of Invention The invention aims to overcome the defects and problems in the prior art and provide a altimeter signal detection method and a altimeter signal detection system based on dynamic parameters OS-CFAR, which can adapt to environmental changes and balance efficiency and stability. In order to achieve the above purpose, the technical solution of the present invention is that a altimeter signal detection method based on dynamic parameters OS-CFAR comprises: collecting radar signals, preprocessing to obtain spectrum data, and calculating spectrum entropy and signal-to-noise ratio based on the spectrum data; the spectrum entropy and the signal-to-noise ratio are formed into two-dimensional feature vectors to be used as the input of a Gaussian mixture model, and an offline training stage and an online execution stage are carried out to generate an environment self-adaptive OS-CFAR parameter combination; Calculating a detection threshold factor based on the environment self-adaptive OS-CFAR parameter combination, and carrying out point-by-point detection on the spectrum data based on the detection threshold factor to identify a target signal; And after the target signal is subjected to multi-dimensional target verification, outputting a final detection result to obtain the altimeter signal. Optionally, the calculating the spectrum entropy and the signal-to-noise ratio specifically includes: the spectral entropy is calculated, and the expression is as follows: ; ; Wherein: For the spectral entropy of the light, Normalized power spectral density; Is the first Power spectral density of the individual frequency points; The sequence number of the frequency point is calculated currently; A traversal index in the summation operation; The total frequency point number of the spectrum data; The signal-to-noise ratio is calculated, which comprises the following steps: Performing self-adaptive sliding window smoothing on the power spectrum based on the spectrum data to obtain a smoothed power spectrum value, wherein the ex