Signal processing method and device of wind measurement laser radar

Abstract

The invention discloses a signal processing method and device for a wind measurement laser radar, and the method comprises the steps: carrying out the windowing interception of an original laser echo signal, and obtaining an intercepted signal; performing dual-tree complex wavelet decomposition transformation on the intercepted signal to obtain each level of approximation wavelet and each level of detail wavelet; performing correlation operation and soft threshold filtering on each level of detail wavelet to obtain a corrected detail wavelet; combining the highest-level approximation wavelet with the corrected detail wavelet to obtain a preliminary noise reduction signal; performing Bayesian maximum posteriori estimation filtering on the preliminary noise reduction signal to obtain a final noise reduction signal; performing dual-tree complex wavelet inverse transformation on the final noise reduction signal to obtain a noise reduction time domain signal; performing fast Fourier transform on the denoised time domain signal to obtain a frequency domain signal; and performing frequency spectrum correction on the frequency domain signal, and then extracting to obtain Doppler frequency. According to the invention, noise filtering of the signals of the wind measurement laser radar is realized, and the Doppler frequency extraction accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of radar signal processing, and in particular to a signal processing method and device for a wind measuring laser radar. Background Art

[0002] The measurement information of atmospheric wind field is widely used in wind shear measurement, atmospheric wind profile measurement, wake wind field measurement and airport wind measurement, etc., which provides a guarantee for the safety of aircraft take-off and landing at airports, and has very important practical significance in the fields of meteorology, aerospace and military. Traditional microwave radar can only observe under special meteorological conditions such as fog, rain and snow, and cannot detect clear sky atmosphere. The laser wavelength of Doppler wind laser radar is in the infrared band, and the aerosol in the atmosphere is used as the detection medium. It uses the Doppler effect of light and calculates the wind speed information by extracting the Doppler frequency in the laser echo. Therefore, it can provide more accurate measurement with higher distance resolution and speed resolution.

[0003] In wind laser radar, the echo signal usually contains a lot of noise, which is similar to Gaussian white noise. Its energy is distributed in the entire frequency band. It is difficult to separate the useful signal and noise, which seriously affects the accurate extraction of Doppler frequency. At present, the Fourier transform of laser echo signal has the problem that the time domain and frequency domain information of the signal cannot be localized at the same time. This method is suitable for processing stationary signals. Although the short-time Fourier transform has the ability of local analysis, due to the limitation of W. Heisenberg uncertainty criterion, the frequency resolution and time resolution of its window function cannot be optimized at the same time. Discrete wavelet transform has the ability of multi-resolution analysis, which can examine the frequency domain characteristics of local time domain processes and the time domain characteristics of local frequency domain processes. However, due to the shortcomings of filter design and the influence of factors such as alternate sampling, discrete wavelet transform still has shortcomings such as frequency aliasing, translation sensitivity, and lack of directional selectivity.

[0004] In summary, the existing wind laser radar signal processing methods have the following technical problems: (1) The general wavelet transform denoising method does not consider the statistical model of the wavelet domain, but only processes the wavelet coefficients of each layer containing random noise obtained by decomposition to achieve the purpose of denoising; (2) The wavelet threshold denoising algorithm analyzes the statistical characteristics of the wavelet coefficients, but ignores the inter-layer correlation of the wavelet coefficients; (3) The existing denoising methods only use traditional wavelets, which have shortcomings such as frequency aliasing and translation sensitivity. Summary of the invention

[0005] The object of the present invention is to provide a signal processing method and device for a wind measuring laser radar, so as to effectively filter out interference noise while better retaining useful characteristic information of the original laser echo.

[0006] To solve the above technical problems, the present invention provides a signal processing method for a wind measurement lidar, including: S1. Windowing and intercepting the original laser echo signal to obtain an intercepted signal; S2. Performing dual-tree complex wavelet decomposition transformation on the intercepted signal to obtain approximation sub-waves and detail sub-waves at each level; S3. Performing correlation operation and soft threshold filtering on the detail sub-waves at each level to obtain corrected detail sub-waves; S4. Combining the highest-level approximation sub-wave and the corrected detail sub-waves to obtain a preliminary noise-reduced signal; S5. Performing Bayesian maximum a posteriori estimation filtering on the preliminary noise-reduced signal to obtain a final noise-reduced signal; S6. Performing inverse dual-tree complex wavelet transformation on the final noise-reduced signal to obtain a noise-reduced time-domain signal; S7. Performing fast Fourier transform on the noise-reduced time-domain signal to obtain a frequency-domain signal; S8. Performing spectrum correction on the frequency-domain signal, and then extracting the Doppler frequency.

[0007] According to the above solution, in step S1, a Hanning window is used for windowing and intercepting.

[0008] According to the above solution, in step S2, the correlation operation on the detail sub-waves at each level is: multiplying the wavelet coefficients at different scales.

[0009] According to the above solution, step S5 specifically includes: S501. Using a robust median estimation method to estimate the noise signal variance estimation value of the preliminary noise-reduced signal; S502. Obtaining the variance estimation value of the preliminary noise-reduced signal according to the preliminary noise-reduced signal at each scale; S503. Obtaining the useful signal estimation value of the preliminary noise-reduced signal according to the noise signal variance estimation value and the variance estimation value of the preliminary noise-reduced signal; S504. Obtaining a filtering threshold according to the noise signal variance estimation value and the useful signal estimation value of the preliminary noise-reduced signal; S505. Filtering the preliminary noise-reduced signal according to the filtering threshold to obtain a final noise-reduced signal.

[0010] According to the above solution, in step S8, a Hanning window and the energy centroid method are used for spectrum correction of the frequency-domain signal.

[0011] The present invention also provides a signal processing device for a wind measurement lidar, including: A windowing and intercepting module, configured to perform windowing and intercepting on the original laser echo signal to obtain an intercepted signal; The dual-tree complex wavelet decomposition and transformation module is used to perform dual-tree complex wavelet decomposition and transformation on the intercepted signal to obtain the approximation sub-waves and detail sub-waves at each level; The first filtering module is used to perform correlation operations and soft-threshold filtering on the detail sub-waves at each level to obtain the corrected detail sub-waves; The preliminary noise reduction signal acquisition module is used to combine the highest-level approximation sub-wave and the corrected detail sub-waves to obtain a preliminary noise reduction signal; The second filtering module is used to perform Bayesian maximum a posteriori estimation filtering on the preliminary noise reduction signal to obtain the final noise reduction signal; The dual-tree complex wavelet inverse transformation module is used to perform dual-tree complex wavelet inverse transformation on the final noise reduction signal to obtain the noise-reduced time-domain signal; The fast Fourier transform module is used to perform fast Fourier transform on the noise-reduced time-domain signal to obtain the frequency-domain signal; The spectrum correction module is used to perform spectrum correction on the frequency-domain signal; The Doppler flat frequency extraction module is used to extract the Doppler frequency from the frequency-domain signal after spectrum correction.

[0012] According to the above scheme, the windowing and intercepting module uses a Hanning window for windowing and intercepting.

[0013] According to the above scheme, the first filtering module performs correlation operations on the detail sub-waves at each level as follows: multiplying the wavelet coefficients at different scales.

[0014] According to the above scheme, the second filtering module specifically performs the following steps: Using the robust median estimation method to estimate the noise signal variance estimate of the preliminary noise reduction signal; According to the preliminary noise reduction signal at each scale, obtain the variance estimate of the preliminary noise reduction signal; According to the noise signal variance estimate and variance estimate of the preliminary noise reduction signal, obtain the useful signal estimate of the preliminary noise reduction signal; According to the noise signal variance estimate and useful signal estimate of the preliminary noise reduction signal, obtain the filtering threshold; Filter the preliminary noise reduction signal according to the filtering threshold to obtain the final noise reduction signal.

[0015] According to the above scheme, the spectrum correction module uses a Hanning window and the energy centroid method to perform spectrum correction on the frequency-domain signal.

[0016] Beneficial effects The present invention performs dual-tree complex wavelet decomposition transformation on the intercepted signal. Compared with traditional wavelets, the method adopted by the present invention has better time-frequency local analysis ability and translation invariance, effectively solving the problems of frequency aliasing and translation sensitivity; by performing correlation operations on the detail wavelets at each level, the inter-layer correlation between wavelet coefficients is fully considered, and denoising is performed using the inter-layer correlation, improving the denoising effect; a statistical model in the wavelet domain is established through Bayesian maximum a posteriori estimation filtering, and the signal is filtered based on the statistical results of the wavelet coefficients at each layer, further improving the noise reduction ability.

[0017] Furthermore, the present invention performs windowed interception through a Hanning window, reducing the leakage of the signal in the frequency domain, improving the overall quality of the intercepted signal, and providing a basis for subsequent signal denoising. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a schematic diagram of the dual-tree complex wavelet decomposition transformation of Embodiment 1 of the present invention; Figure 2 is a flowchart of the signal processing method of the wind measurement lidar in Embodiment 1 of the present invention; Figure 3 is a flowchart of the preliminary denoising of the intercepted signal in Embodiment 1 of the present invention; Figure 4 is a flowchart of the signal processing method of the wind measurement lidar in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present disclosure. Obviously, the described embodiments are some, but not all, of the embodiments of the present disclosure. All other embodiments obtained by those of ordinary skill in the art based on the described embodiments of the present disclosure without creative efforts shall fall within the scope of protection of the present disclosure.

[0020] Embodiment 1: Refer to Figures 1 to 4 , this embodiment discloses a signal processing method for a wind measurement lidar, including: S1. Perform windowed interception on the original laser echo signal to obtain an intercepted signal; S2. Perform dual-tree complex wavelet decomposition transformation on the intercepted signal to obtain approximation wavelets and detail wavelets at each level; S3. Perform correlation operations and soft threshold filtering on the detail wavelets at each level to obtain corrected detail wavelets; S4. Combine the highest-level approximation wavelet and the corrected detail wavelets to obtain a preliminarily denoised signal; S5. Perform Bayesian maximum a posteriori estimation filtering on the preliminary noise-reduced signal to obtain the final noise-reduced signal; S6. Perform inverse dual-tree complex wavelet transform on the final noise-reduced signal to obtain the noise-reduced time-domain signal; S7. Perform fast Fourier transform on the noise-reduced time-domain signal to obtain the frequency-domain signal; S8. Perform spectrum correction on the frequency-domain signal, and then extract the Doppler frequency.

[0021] Further, in step S1, a Hanning window is used for windowing and truncating.

[0022] Further, in step S2, the correlation operation is performed on each level of detail wavelet respectively as: multiplying wavelet coefficients of different scales; In this embodiment, the signal model of the truncated signal in the dual-tree complex wavelet domain is expressed as:

[0023] In the above formula, is the truncated signal containing noise, are the wavelet coefficients of the useful signal in the truncated signal, are the wavelet coefficients of the noise in the truncated signal; Therefore, the process of multiplying wavelet coefficients of different scales is expressed as:

[0024]

[0025] The above process is achieved by multiplying wavelet coefficients of different wavelet scales (i and j); it can be understood that since the two product terms of the first item and are both time-domain consistent signals, the time-domain property of their product will remain unchanged; since the product terms in the remaining three items have inconsistent time-domain properties, the product terms are weakened or even eliminated.

[0026] Further, step S5 specifically includes: S501. Estimate the variance estimate value of the noise signal of the preliminary noise-reduced signal by using the robust median estimation method; Specifically expressed as: ,

[0027] In the above formula, is the variance estimate value of the noise signal of the preliminary noise-reduced signal, represents the wavelet coefficient of the i-th layer, represents the high-frequency wavelet coefficient; S502. Obtain the variance estimation value of the preliminary noise-reduced signal based on the preliminary noise-reduced signals at each scale; Specifically expressed as:

[0028] In the above formula, is the variance estimation value of the preliminary noise-reduced signal (observed sub-band signal), is the number of wavelet decomposition layers; S503. Obtain the estimated value of the useful signal of the preliminary noise-reduced signal based on the variance estimation value of the noise signal and the variance estimation value of the preliminary noise-reduced signal; Since , , satisfy the following relationship:

[0029] Therefore, the estimated value of the useful signal of the preliminary noise-reduced signal is specifically expressed as:

[0030] S504. Obtain the filtering threshold based on the variance estimation value of the noise signal and the estimated value of the useful signal of the preliminary noise-reduced signal; Based on the soft threshold formula:

[0031] Substitute the above parameters, so the filtering threshold is expressed as:

[0032] S505. Filter the preliminary noise-reduced signal according to the filtering threshold to obtain the final noise-reduced signal.

[0033] Furthermore, in step S8, a Hanning window and the energy centroid method are used to perform spectrum correction on the frequency-domain signal; specifically, the Hanning window function is used to truncate the time-domain signal , and its FFT transformation is performed to obtain the discrete amplitude spectrum line; generally, the signal frequency is between two spectrum lines of the window function. By solving the signal frequency and the central coordinate of the main lobe of the window function spectrum, the frequency and amplitude of the center of the main lobe of the window function can be obtained, and the frequency and amplitude in the signal spectrum can be corrected; the normalized frequency correction amount can be expressed as:

[0034] In the above formula, is the frequency correction amount, is the central frequency of the main lobe of the signal intercepted by the Hanning window, is the spectrum line number of the peak value within the main lobe, is the sampling frequency of the signal, is the number of data points.

[0035] In summary, this embodiment discloses a signal processing method for a wind measurement lidar. The method first performs a dual-tree complex wavelet transform on the windowed and intercepted wind measurement lidar signal to obtain the wavelet coefficients of each layer of the original signal. Then, according to the inter-layer correlation of the wavelet coefficients, a correlation operation is performed on the dual-tree complex wavelet decomposition coefficients containing noise, and the signal components in different frequency band ranges are denoised by the method of soft threshold filtering. On the premise that the wavelet transform coefficients of the noise signal and the useful signal respectively follow the Gaussian distribution and the Laplace distribution probability assumptions, the threshold obtained by the Bayesian maximum a posteriori probability estimation algorithm is used to filter and denoise the noisy wavelet coefficients at each decomposition scale, and the wavelet coefficients of the useful signal are estimated. Finally, the dual-tree complex wavelet inverse transform is performed according to the estimated value of the useful signal to reconstruct the real signal, and combined with the Hanning window energy centroid correction method, the purpose of filtering out noise and accurately extracting the Doppler frequency shift of the wind measurement lidar signal is achieved.

[0036] Embodiment 2: The present invention also provides a signal processing device for a wind measurement lidar, including: A windowed interception module, configured to perform windowed interception on the original laser echo signal to obtain an intercepted signal; A dual-tree complex wavelet decomposition transform module, configured to perform a dual-tree complex wavelet decomposition transform on the intercepted signal to obtain approximation sub-waves and detail sub-waves at each level; A first filtering module, configured to perform a correlation operation and soft threshold filtering on the detail sub-waves at each level respectively to obtain corrected detail sub-waves; A preliminary denoised signal acquisition module, configured to combine the highest-level approximation sub-wave and the corrected detail sub-waves to obtain a preliminary denoised signal; A second filtering module, configured to perform Bayesian maximum a posteriori estimation filtering on the preliminary denoised signal to obtain a final denoised signal; A dual-tree complex wavelet inverse transform module, configured to perform a dual-tree complex wavelet inverse transform on the final denoised signal to obtain a denoised time-domain signal; A fast Fourier transform module, configured to perform a fast Fourier transform on the denoised time-domain signal to obtain a frequency-domain signal; A spectrum correction module, configured to perform spectrum correction on the frequency-domain signal; A Doppler average frequency extraction module, configured to extract the Doppler frequency from the frequency-domain signal after spectrum correction.

[0037] Furthermore, the windowed interception module performs windowed interception using a Hanning window.

[0038] Furthermore, the first filtering module performs a correlation operation on the detail sub-waves at each level respectively as follows: multiplying the wavelet coefficients of different scales.

[0039] Further, the second filtering module specifically performs the following steps: Estimate the noise signal variance estimate of the preliminary noise-reduced signal by using the robust median estimation method; Obtain the variance estimate of the preliminary noise-reduced signal according to the preliminary noise-reduced signals at each scale; Obtain the useful signal estimate of the preliminary noise-reduced signal according to the noise signal variance estimate and the variance estimate of the preliminary noise-reduced signal; Obtain the filtering threshold according to the noise signal variance estimate and the useful signal estimate of the preliminary noise-reduced signal; Filter the preliminary noise-reduced signal according to the filtering threshold to obtain the final noise-reduced signal.

[0040] According to the above solution, the spectrum correction module performs spectrum correction on the frequency-domain signal by using the Hanning window and the energy centroid method.

[0041] It should be noted that according to the needs of implementation, each step / component described in the present application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0042] Those skilled in the art can easily understand that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A signal processing method for a wind measurement lidar, characterized in that Including: S1. Window the original laser echo signal and intercept it to obtain the intercepted signal; S2. Perform dual-tree complex wavelet decomposition transform on the intercepted signal to obtain approximation sub-wavelets and detail sub-wavelets at each level; S3. Perform correlation operation and soft threshold filtering on the detail sub-wavelets at each level respectively to obtain the corrected detail sub-wavelets; S4. Combine the highest-level approximation sub-wavelet and the corrected detail sub-wavelets to obtain the preliminary noise-reduced signal; S5. Perform Bayesian maximum a posteriori estimation filtering on the preliminary noise-reduced signal to obtain the final noise-reduced signal; S6. Perform inverse dual-tree complex wavelet transform on the final noise-reduced signal to obtain the noise-reduced time-domain signal; S7. Perform fast Fourier transform on the noise-reduced time-domain signal to obtain the frequency-domain signal; S8. Perform spectrum correction on the frequency-domain signal, and then extract the Doppler frequency.

2. The signal processing method of the wind measurement lidar according to claim 1, wherein In step S1, a Hanning window is used for windowing and interception.

3. The signal processing method of the wind measurement lidar according to claim 1, wherein In step S2, the correlation operation on the detail sub-wavelets at each level is as follows: multiply the wavelet coefficients at different scales.

4. The signal processing method of the wind measurement lidar according to claim 1, characterized in that Step S5 specifically includes: S501. Use the robust median estimation method to estimate the variance estimation value of the noise signal of the preliminary noise-reduced signal; S502. Obtain the variance estimation value of the preliminary noise-reduced signal according to the preliminary noise-reduced signal at each scale; S503. Obtain the estimated value of the useful signal of the preliminary noise-reduced signal according to the variance estimation value of the noise signal and the variance estimation value of the preliminary noise-reduced signal; S504. Obtain the filtering threshold according to the variance estimation value of the noise signal and the estimated value of the useful signal of the preliminary noise-reduced signal; S505. Filter the preliminary noise-reduced signal according to the filtering threshold to obtain the final noise-reduced signal.

5. The signal processing method of the wind measurement lidar according to claim 1, characterized in that, In step S8, a Hanning window and the energy centroid method are used to perform spectrum correction on the frequency-domain signal.

6. A signal processing device for a wind measurement lidar, characterized in that, Including: Windowing and Interception Module, used to window and intercept the original laser echo signal to obtain the intercepted signal; Dual-tree Complex Wavelet Decomposition Transform Module, used to perform dual-tree complex wavelet decomposition transform on the intercepted signal to obtain approximation sub-wavelets and detail sub-wavelets at each level; First Filtering Module, used to perform correlation operation and soft threshold filtering on the detail sub-wavelets at each level respectively to obtain the corrected detail sub-wavelets; Preliminary Noise-reduced Signal Acquisition Module, used to combine the highest-level approximation sub-wavelet and the corrected detail sub-wavelets to obtain the preliminary noise-reduced signal; Second Filtering Module, used to perform Bayesian maximum a posteriori estimation filtering on the preliminary noise-reduced signal to obtain the final noise-reduced signal; Inverse Dual-tree Complex Wavelet Transform Module, used to perform inverse dual-tree complex wavelet transform on the final noise-reduced signal to obtain the noise-reduced time-domain signal; Fast Fourier Transform Module, used to perform fast Fourier transform on the noise-reduced time-domain signal to obtain the frequency-domain signal; Spectrum Correction Module, used to perform spectrum correction on the frequency-domain signal; Doppler Mean Frequency Extraction Module, used to extract the Doppler frequency from the frequency-domain signal after spectrum correction.

7. The signal processing device of the wind measurement lidar according to claim 6, characterized in that, The Windowing and Interception Module uses a Hanning window for windowing and interception.

8. The signal processing device of the wind measurement lidar according to claim 6, wherein, The correlation operation performed by the First Filtering Module on the detail sub-wavelets at each level is as follows: multiply the wavelet coefficients at different scales.

9. The signal processing device of the wind measurement lidar according to claim 6, characterized in that The Second Filtering Module specifically executes the following steps: Use the robust median estimation method to estimate the variance estimation value of the noise signal of the preliminary noise-reduced signal; Obtain the variance estimation value of the preliminary noise-reduced signal according to the preliminary noise-reduced signal at each scale; An estimated value of the useful signal of the preliminary noise-reduced signal is obtained according to the estimated value of the noise signal variance and the estimated value of the variance of the preliminary noise-reduced signal; A filtering threshold is obtained according to the estimated value of the noise signal variance and the estimated value of the useful signal of the preliminary noise-reduced signal; The preliminary noise-reduced signal is filtered according to the filtering threshold to obtain the final noise-reduced signal.

10. The signal processing device of the wind measurement lidar according to claim 6, characterized in that, The spectrum correction module uses a Hanning window and an energy centroid method to perform spectrum correction on the frequency-domain signal.

Images