Lidar ocean wind field low signal-to-noise ratio inversion method and system

Abstract

The application belongs to the technical field of wind measurement laser radar, and particularly relates to a laser radar ocean wind field low signal-to-noise ratio inversion method and system. In view of the problem that the coherent Doppler wind measurement laser radar has low signal-to-noise ratio and it is difficult to accurately estimate the radial wind speed under the condition of weak echo in the ocean measurement environment, the method firstly standardizes the received complex signal and converts the Doppler spectrum to extract the initial radial wind speed. Then, a mixed probability density model composed of Gaussian distributed signal components and uniformly distributed noise components is established to realize signal and noise separation. On this basis, by constructing an objective function and performing grid optimization search in a two-dimensional wind vector space, the best wind speed vector estimation value is determined. The application can effectively filter out the influence of the ocean atmospheric background noise on the inversion result, and significantly improve the wind measurement accuracy and effective detection distance of the laser radar in the low signal-to-noise ratio ocean environment.

Description

Technical Field

[0001] This invention belongs to the field of wind-measuring lidar technology, specifically relating to a lidar method and system for low signal-to-noise ratio inversion of marine wind fields. Background Technology

[0002] Coherent Doppler wind lidar uses aerosol particles moving with the wind field in the atmosphere as tracers, and achieves rapid inversion of the atmospheric wind field by detecting the Doppler frequency shift between the emitted laser and the backscattered light signal from the aerosol. Compared to the atmospheric boundary layer over land, the aerosol content over the ocean is relatively sparse, and the aerosol concentration decreases rapidly with increasing detection altitude (above the atmospheric boundary layer). In this low tracer measurement environment, the echo signal received by the lidar is very weak, resulting in an extremely low signal-to-noise ratio, and the inverted wind field results are highly susceptible to interference from background noise.

[0003] Traditional direct sine wave fitting algorithms, when processing such data, suffer from increased erroneous peak probabilities due to noise, leading to significant deviations in estimated wind speeds. Under conditions of weak tracers, accurately extracting target wind speed information in the face of low signal-to-noise ratios and difficulties in radial wind speed estimation is a critical issue that urgently needs to be addressed to improve the detection performance of coherent Doppler lidar in marine or high-altitude environments. Summary of the Invention

[0004] To achieve high-precision wind field inversion in low signal-to-noise ratio marine environments, this invention provides a weak signal wind field inversion algorithm based on optimal estimation. The technical solution is a low signal-to-noise ratio marine wind field inversion method based on optimal estimation using lidar, comprising the following steps:

[0005] A method for inverting low signal-to-noise ratio ocean wind fields using lidar includes the following steps:

[0006] S1. Before wind speed estimation, the raw signal received by the lidar is preprocessed. For each range gate and scanning azimuth angle, complex signal samples are calculated and normalized using the mean noise power to obtain a normalized signal.

[0007] S2. Using the normalized signal, calculate the estimated value of the autocovariance function, and then convert the autocovariance function into a Doppler spectrum;

[0008] S3. Find the index corresponding to the maximum value in the Doppler spectrum, and extract the radial wind speed estimate for each scan azimuth angle based on the peak position;

[0009] S4. Establish a mixed probability density model for radial velocity, and convert the radial wind speed estimate into a superposition of a Gaussian distributed signal component projected around the true wind speed and a noise component that conforms to a uniform distribution, so as to achieve signal and noise separation.

[0010] S5. Using the hybrid probability density model, construct an objective function that includes all scanning azimuth information, perform grid optimization search in the two-dimensional wind vector space, and determine the optimal estimated wind speed vector;

[0011] S6. Calculate and output the horizontal wind speed, wind direction, and wind field profile based on the components of the optimally estimated wind speed vector.

[0012] Preferably, in step S1, the original echo signal is acquired and normalized preprocessed to obtain the lidar signal at each range gate. and scanning azimuth angle The received raw complex signal sample Using average noise power The complex signal is standardized to obtain the normalized signal. :

[0013] ;

[0014] in, To remove the influence of intermediate frequency on the signal, This represents the number of signal samples corresponding to the distance resolution. The sampling interval is... This represents the number of range gates traversed from the initial detection range to the target. This represents the index of the sampling point within the distance gate, with values ​​ranging from 0, 1, 2, ..., M-1. This represents the azimuth sequence index during the conical scanning process of the lidar. It is the angular difference between two consecutive azimuth angles, i.e., angular resolution; This indicates the precise scanning azimuth position of the current signal sample.

[0015] Preferably, the autocovariance function is calculated. The estimated value:

[0016] ;

[0017] Subsequently, the autocovariance function is converted into a Doppler spectrum using discrete Fourier transform. :

[0018] ;

[0019] in, Represents discrete velocity points. For the Kronecker function, This represents the number of staggered sampling points between signal sequences when calculating correlation. This represents the number of laser pulses accumulated. Represents the pulse index within the current accumulated sector, with a value ranging from 0 to... , It serves as a global index for laser pulses, used to precisely locate the currently processed individual pulse signal within the global data matrix.

[0020] Preferably, the radial wind speed is initially estimated in the Doppler spectrum. Find the index corresponding to the maximum value in the middle. For each scanning azimuth, the initial radial wind speed estimate is determined based on the peak position. The calculation formula is:

[0021] ;

[0022] in, This represents the step size for wind speed resolution.

[0023] Preferably, the probability density function of radial wind speed is established. :

[0024] ;

[0025] variance is The probability distribution of noise remains constant throughout the entire measurement range. , Represents the true wind speed vector. For its in the Projection in each azimuth direction This refers to the wind speed measurement range.

[0026] Preferably, assuming that the radial velocity estimates are independent of each other, at low signal-to-noise ratios, the following conditions are met. Under the given conditions, and in conjunction with the above probability density, construct the objective function. The sum of the Gaussian exponent terms for all measured angles:

[0027] ;

[0028] in Let be the projection of the wind vector to be determined onto the corresponding scanning direction. This represents the total number of effective radial velocity estimates that participate in the optimal estimation.

[0029] Preferably, a radial wind speed lookup table is established and traversed for optimization in a two-dimensional wind speed plane. Divide the grid, set the step size, and traverse each candidate wind vector in the grid. Calculate its corresponding Value, find the objective function Reaching the maximum vector This is the optimal wind vector estimate, then the horizontal wind speed... With wind direction :

[0030] ;

[0031] ;

[0032] A lidar system for retrieving ocean wind fields with low signal-to-noise ratio includes a data acquisition unit, a data processing unit, and an output unit.

[0033] Data acquisition unit: uses lidar to receive signals;

[0034] Data processing unit: preprocesses the data, standardizes and performs Doppler spectrum conversion on the received complex signal, extracts the initial radial wind speed; and establishes a mixed probability density model composed of Gaussian distributed signal components and uniformly distributed noise components to achieve signal and noise separation. It constructs a joint likelihood function through the maximum likelihood estimation method and performs grid optimization search in the two-dimensional wind vector space to determine the optimal wind speed vector estimate.

[0035] Output unit: Visualizes the processed data.

[0036] Compared with the prior art, the beneficial effects of this application are as follows:

[0037] 1. This invention overcomes the bottleneck of traditional direct sine wave fitting failing under weak signals by establishing a hybrid probability density model, and can effectively filter out the influence of noise on the inversion results.

[0038] 2. To improve the system's wind measurement accuracy, a grid search is used to maximize the objective function. Even in extreme environments with a signal-to-noise ratio as low as -20 dB, the wind speed error can still be controlled within an acceptable range, significantly improving the system's imaging and wind measurement accuracy.

[0039] 3. By utilizing the joint probability processing method of this spatial azimuth information, the effective detection range of the system in low signal-to-noise ratio environments is improved, enabling complete and high-precision detection of weak echo targets at higher altitudes or greater distances. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below;

[0041] Figure 1 This is a schematic diagram of a low signal-to-noise ratio inversion method for ocean-atmospheric boundary layer wind field based on optimal estimation, as disclosed in an embodiment of the present invention. Detailed Implementation

[0042] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0043] A method for inverting low signal-to-noise ratio ocean wind fields using lidar includes the following steps:

[0044] Step 1: Acquire the raw echo signal and perform normalization preprocessing. Acquire the LiDAR signal at each range gate. and scanning azimuth angle The received raw complex signal sample To eliminate the influence of signal strength, average noise power is used. The complex signal is standardized to obtain the normalized signal. :

[0045] ;

[0046] in, To remove the influence of intermediate frequency on the signal, This represents the number of signal samples corresponding to the distance resolution. The sampling interval is denoted as .

[0047] Step 2, using the normalized signal The estimated value of the autocovariance function is calculated, and then the autocovariance function is converted into a Doppler spectrum through Fourier transform. The resolution of the spectrum depends on the sampling parameters.

[0048] Calculate the autocovariance function The estimated value:

[0049] .

[0050] Subsequently, the autocovariance function is converted into a Doppler spectrum using discrete Fourier transform. :

[0051] ;

[0052] in, Represents discrete velocity points. Let Kronecker function be used.

[0053] Step 3: Preliminary estimate of radial wind speed in the Doppler spectrum. Find the index corresponding to the maximum value in the middle. For each scan azimuth, an initial radial wind speed estimate is determined based on the peak location. The calculation formula is:

[0054] ;

[0055] in, This represents the step size for wind speed resolution.

[0056] Step 4: Under extremely low signal-to-noise ratio conditions, the radial wind speed estimate determined based on the peak position of the Doppler spectrum is prone to distortion. Therefore, a probability density function for the radial wind speed is established. :

[0057] ;

[0058] In the above formula, the probability density function of radial wind speed consists of two parts. The part that follows a Gaussian distribution represents the effective atmospheric signal, which is concentrated near the projection of the true wind speed, and has a variance of... (Under typical system parameters) (Take 2 m / s). The other part conforms to a uniform distribution, representing noise, whose probability distribution remains constant throughout the entire measurement range. In the formula Represents the true wind speed vector. For its in the Projection in each azimuth direction This refers to the wind speed measurement range.

[0059] Step 5, assuming that the radial velocity estimates are independent of each other, under low signal-to-noise ratio conditions, satisfy... Under the given conditions, and in conjunction with the above probability density, construct the objective function. The sum of the Gaussian exponent terms for all measured angles:

[0060] ;

[0061] in Let be the projection of the wind vector to be determined onto the corresponding scanning direction.

[0062] Step 6: Establish a radial wind speed lookup table and iterate through it for optimization in a two-dimensional wind speed plane. Divide the grid into 500×500 pixels with a step size of 0.1 m / s. Iterate through each candidate wind vector in the grid. Calculate its corresponding Value, find the objective function Reaching the maximum vector This is the optimal wind vector estimate. Then the horizontal wind speed... With wind direction :

[0063] ;

[0064] .

[0065] A lidar system for retrieving ocean wind fields with low signal-to-noise ratio includes a data acquisition unit, a data processing unit, and an output unit.

[0066] Data acquisition unit: uses lidar to receive signals;

[0067] Data processing unit: preprocesses the data, standardizes and performs Doppler spectrum conversion on the received complex signal, extracts the initial radial wind speed; and establishes a hybrid probability density model composed of Gaussian distributed signal components and uniformly distributed noise components to achieve signal and noise separation. By constructing a joint objective function and performing grid optimization search in the two-dimensional wind vector space, the optimal wind speed vector estimate is determined.

[0068] Output unit: Visualizes the processed data.

[0069] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for low signal-to-noise ratio inversion of marine wind fields using lidar, characterized in that, Includes the following steps: S1. Before wind speed estimation, the raw signal received by the lidar is preprocessed. For each range gate and scanning azimuth angle, complex signal samples are calculated and normalized using the mean noise power to obtain a normalized signal. S2. Using the normalized signal, calculate the estimated value of the autocovariance function, and then convert the autocovariance function into a Doppler spectrum; S3. Find the index corresponding to the maximum value in the Doppler spectrum. For each scanning azimuth, the radial wind speed estimate is extracted based on the peak position. The calculation formula is: ； in, The step size for wind speed resolution. This represents the number of signal samples corresponding to the distance resolution. S4. Establish a mixed probability density model for radial velocity, and convert the radial wind speed estimate into a superposition of a Gaussian distributed signal component projected around the true wind speed and a noise component that conforms to a uniform distribution, so as to achieve signal and noise separation. Establish the probability density function of radial wind speed. : ； variance is The probability distribution of noise remains constant throughout the entire measurement range. , Represents the true wind speed vector. For its in the Projection in each azimuth direction, This refers to the wind speed measurement range; S5. Using the hybrid probability density model, construct an objective function that includes all scanning azimuth information, perform grid optimization search in the two-dimensional wind vector space, and determine the optimal estimated wind speed vector; S6. Calculate and output the horizontal wind speed, wind direction, and wind field profile based on the components of the optimally estimated wind speed vector.

2. The method for low signal-to-noise ratio inversion of marine wind fields using lidar according to claim 1, characterized in that, Step S1: Acquire the raw echo signal and perform normalization preprocessing to obtain the lidar signal at each range gate. and scanning azimuth angle The received raw complex signal sample Using average noise power The complex signal is standardized to obtain the normalized signal. : ； in, To remove the influence of intermediate frequency on the signal, This represents the number of signal samples corresponding to the distance resolution. The sampling interval is... This represents the number of range gates traversed from the initial detection range to the target. This represents the index of the sampling point within the distance gate, with values ​​ranging from 0, 1, 2, ..., M-1. This represents the azimuth sequence index during the conical scanning process of the lidar. It is the angular difference between two consecutive azimuth angles, i.e., angular resolution; This indicates the precise scanning azimuth position of the current signal sample.

3. The method for low signal-to-noise ratio inversion of marine wind fields using lidar according to claim 2, characterized in that, Calculate the autocovariance function The estimated value: ； Subsequently, the autocovariance function is converted into a Doppler spectrum using discrete Fourier transform. : ； in, Represents discrete velocity points. For the Kronecker function, This represents the number of staggered sampling points between signal sequences when calculating correlation. This represents the number of laser pulses accumulated. Represents the pulse index within the current accumulated sector, with a value ranging from 0 to... , It serves as a global index for laser pulses, used to precisely locate the currently processed individual pulse signal within the global data matrix.

4. The method for low signal-to-noise ratio inversion of marine wind fields using lidar according to claim 1, characterized in that, Assuming that the radial velocity estimates are independent of each other, at a low signal-to-noise ratio, the following conditions are met: Under the given conditions, and in conjunction with the above probability density, construct the objective function. The sum of the Gaussian exponent terms for all measured angles: ； in Let be the projection of the wind vector to be determined onto the corresponding scanning direction. This represents the total number of valid radial velocity estimates that participate in the optimal estimation.

5. The method for low signal-to-noise ratio inversion of marine wind fields using lidar according to claim 4, characterized in that, Establish a radial wind speed lookup table and iterate through it to find the optimal value in a two-dimensional wind speed plane. Divide the grid, set the step size, and traverse each candidate wind vector in the grid. Calculate its corresponding Value, find the objective function Reaching the maximum vector This is the optimal wind vector estimate, then the horizontal wind speed... With wind direction : ； 。 6. A lidar system for retrieving low signal-to-noise ratio ocean wind fields, adapted to the method described in any one of claims 1-5, characterized in that, It includes a data acquisition unit, a data processing unit, and an output unit; Data acquisition unit: uses lidar to receive signals; Data processing unit: preprocesses the data, standardizes and performs Doppler spectrum conversion on the received complex signal, extracts the initial radial wind speed; and establishes a mixed probability density model composed of Gaussian distributed signal components and uniformly distributed noise components to achieve signal and noise separation. It constructs a joint likelihood function through the maximum likelihood estimation method and performs grid optimization search in the two-dimensional wind vector space to determine the optimal wind speed vector estimate. Output unit: Visualizes the processed data.

Images