Machine learning-based average wave period estimation method of coherent X wave band

Abstract

The invention relates to a coherent X-waveband microwave radar average wave period estimation method based on machine learning. The method comprises the following steps: acquiring a single echo Doppler spectrum and a time Doppler spectrum of a sea surface; acquiring a left boundary and a right boundary from the echo Doppler spectrogram, and extracting Doppler frequency shift of an echo Doppler spectrum; obtaining a space-time sequence of the radial speed, and extracting characteristic parameters of the wave characteristics; calculating the minimum peak distance of the wave features according to the average wave period data, and inputting the feature parameters and the minimum peak distance into a random forest and linear regression integrated model for training; and then inputting the characteristic parameters into the trained integrated model for prediction, and identifying a wave crest and a wave trough according to a prediction result so as to estimate an average wave period. According to the method, the direct relation between the radar echo speed space-time sequence and the wave period parameters is established, the problems of large calculation amount, long time consumption and low precision of a traditional method are solved, the method is not affected by historical data quality caused by extreme weather conditions, and the accuracy of average wave period estimation is improved.

Description

Technical Field

[0001] The present invention relates to the field of radar technology, and in particular to a method for estimating an average wave period of a coherent X-band based on machine learning. Background Art

[0002] In ocean monitoring, wave parameters are of great significance for marine engineering, hazard prediction and oceanographic research. As an effective ocean detection tool, coherent microwave radar extracts Doppler velocity information from echoes and uses the relationship between wave orbit velocity and wave height to obtain wave parameters such as wave spectrum and average wave period. However, due to the influence of breaking waves, nonlinear characteristics such as group lines and high-order harmonic energy in the wave number frequency spectrum measured by radar reduce the accuracy of wave parameter inversion, especially the estimation of average wave period.

[0003] The current method of estimating the average wave period using coherent microwave radar ignores the uncertainty of nonlinear wave energy and energy distribution of different frequency components in radar echoes, resulting in large errors in the measurement results; predicting ocean wave parameters through machine learning requires a large amount of computing resources and input data, and the efficiency of long-term wave parameter prediction is low; the data-driven model that uses historical wave data for short-term and long-term wave parameter forecasting has large errors in predicting nonlinear and non-stationary wave parameters due to data variability and the limitations of linear and stationary assumptions. Summary of the invention

[0004] The technical problem of the present invention is to provide a method for estimating the average wave period of the coherent X-band based on machine learning, estimate the average wave period from the perspective of velocity time-space series and machine learning, improve the precision and accuracy of coherent microwave radar in predicting the average wave period, and provide a more accurate method for the inversion of wave parameters.

[0005] The purpose of the present invention is to solve the above problems. The technical solution of the present invention is a method for estimating the average wave period of the coherent X-band based on machine learning, comprising the following steps: S1: Use coherent X-band radar to obtain the Doppler spectrum and time Doppler spectrum of a single echo on the sea surface; S2: Obtain the left boundary and the right boundary from the echo Doppler spectrum, and extract the Doppler frequency shift of the echo Doppler spectrum by moment estimation method; S3: The time-space sequence of radial velocity is obtained through Doppler frequency shift, and divided into a training data set and a test data set, and then the characteristic parameters of the wave characteristics are extracted to obtain the training data set D1 and the test data set D2; S4: Calculate the minimum peak distance of wave characteristics based on the average wave period data, and then input the data set D1 and the minimum peak distance into the random forest and linear regression integrated model for training; S5: Input the dataset D2 into the trained integrated model to predict the minimum peak distance of the wave characteristics, identify the positions of the crests and troughs based on the prediction results, and then estimate the average wave period based on the relationship between wavelength and period.

[0006] Preferably, step S2 includes calculating the center frequency through the echo Doppler spectrum, and the calculation formula is: ; In the formula, represents the Doppler frequency, represents the Doppler spectrum.

[0007] Furthermore, characteristic parameters of the wave characteristics are extracted in S3, specifically, the mean, standard deviation, maximum value and minimum value of the wave characteristics are extracted.

[0008] Preferably, the calculation formula of the space-time series of radial velocity is: ; In the formula, represents the electromagnetic wavelength, represents the radial distance, represents the sampling time, f d represents the average Doppler shift.

[0009] Furthermore, step S4 also includes: performing correlation analysis on the fluctuation features in the data set D1, screening feature subsets that are significantly correlated with the target variable, and then eliminating low-correlation feature values ​​through a set correlation threshold; Preferably, the random forest and linear regression integrated model, the random forest sub-model integrates multiple independent regression trees and bagging strategies to enhance the tolerance to noise and outliers and the prediction accuracy and stability; the linear regression sub-model introduces a regularization method, including Lasso or Ridge regression method, and the calculation formula of the random forest and linear regression integrated model is: ; In the formula, represents the nonlinear prediction of random forest, represents the number of decision trees, represents the input feature value, represents independent and identically distributed random variables, represents the decision tree output, represents the linear prediction of linear regression, is the regression coefficient, which indicates the influence of each feature on the target variable. represents the error term, and Represent the weights of random forest regression and linear regression respectively.

[0010] Furthermore, the average wave period is estimated based on the relationship between wavelength and period. Specifically, the average wave period is calculated based on linear wave theory. The calculation formula is: ; In the formula, represents the wavelength, Represents the cycle, Represents the phase velocity of the wave.

[0011] Compared with the prior art, the beneficial effects of the present invention include: 1) The average wave period estimation method of the coherent X-band based on machine learning proposed in the present invention extracts characteristic parameters of wave characteristics by analyzing the velocity space-time series, and inputs them into the constructed random forest and linear regression integrated model to predict the minimum peak distance of wave characteristics. A direct relationship between the radar echo velocity space-time series and the wave period parameters is established, which solves the problems of large calculation amount, long time consumption and low precision in traditional methods, and is not affected by the quality of historical data caused by extreme weather conditions, thereby improving the accuracy of the average wave period estimation.

[0012] 2) The random forest and linear regression integrated model proposed in this invention can effectively identify the locations of wave peaks and troughs, and use the relationship between wavelength and period to accurately evaluate the average wave period, thereby improving the accuracy of average wave period estimation of coherent microwave radar in ocean remote sensing and improving the effectiveness of remote sensing data.

[0013] 3) The present invention retains most of the echo information in the process of effectively removing interference, balances noise suppression and information retention, ensures the efficiency and accuracy of wave parameter estimation, and provides more valuable data support for marine engineering and monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0015] Figure 1 Schematic diagram of the mean wave period estimation method according to an embodiment of the present invention.

[0016] Figure 2 This is a radar irradiated sea surface scattering geometry diagram according to an embodiment of the present invention.

[0017] Figure 3 It is a single echo Doppler spectrum obtained by detecting the echo signal of the sea surface using the coherent microwave radar in an embodiment of the present invention.

[0018] Figure 4 The time Doppler spectrum is obtained when the coherent microwave radar in an embodiment of the present invention detects the echo signal of the sea surface.

[0019] Figure 5It is a space-time series of displacement of the simulated sea surface according to an embodiment of the present invention.

[0020] Figure 6 It is the space-time series of the velocity of the simulated sea surface according to the embodiment of the present invention.

[0021] Figure 7 This is a comparison chart of the results of evaluating the average sea surface wave period by the method of an embodiment of the present invention and the Berbi and simulation methods. DETAILED DESCRIPTION

[0022] The sea detection data used in the present invention comes from the data set "Radar Sea Detection Data 2022 Issue 1" published by the Journal of Radar.

[0023] like Figure 1 As shown, the average wave period estimation method of the coherent X-band based on machine learning includes the following steps: S1: Use coherent multi-band radar to obtain the Doppler spectrum and time Doppler spectrum of a single echo on the sea surface; like Figure 2 As shown in Figure 1, the range resolution of the radar is set to 2.5m and the Doppler spectrum sampling rate is 0.0005s. Water particles on the sea surface propagate in the form of continuous waves, and the orbital motion causes the frequency shift of the Doppler spectrum.

[0024] The calculation formula of radar track velocity is: ; In the formula, represents radial distance, t represents time, u( ) represents the horizontal component of orbital velocity, and w( ) represents the vertical component of orbital velocity.

[0025] ; ; In the formula, represents the incident angle of the radar illuminating the sea surface, U 0 represents the initial orbital velocity, k represents the wave number, and w represents the wave approaching or moving away from the radar.

[0026] Radial velocity of radar echo and center frequency of Doppler spectrum The expression of the relationship between is: ; In the formula, represents the radial velocity of the radar echo, The center frequency of the Doppler spectrum.

[0027] The single echo Doppler spectrum and time Doppler spectrum obtained by coherent microwave radar detecting the echo signal of the sea surface are as follows: Figure 3 and Figure 4 shown.

[0028] S2: Obtain the left boundary and the right boundary from the echo Doppler spectrum, and extract the Doppler frequency shift of the echo Doppler spectrum by moment estimation method; In step S2, the center frequency is calculated by the echo Doppler spectrum, and the calculation formula is: ; In the formula, represents the Doppler frequency, represents the Doppler spectrum, Indicates the left border, Indicates the right border.

[0029] S3: The time-space sequence of radial velocity is obtained through Doppler frequency shift, and divided into a training data set and a test data set, and then the characteristic parameters of the wave characteristics are extracted to obtain the training data set D1 and the test data set D2; The displacement sequence of the sea surface reflects the wave characteristics of the sea surface. The space-time sequence of the simulated sea surface radial velocity is as follows: Figure 3 As shown; The velocity time series of the simulated sea surface is as follows Figure 6 As shown, the characteristic parameters of the wave characteristics are extracted, specifically the mean, standard deviation, maximum value and minimum value of the wave characteristics.

[0030] The mean wave period data are obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF).

[0031] The calculation formula of the space-time series of radial velocity is: ; In the formula, represents the electromagnetic wavelength, represents the radial distance, represents the sampling time, f d represents the average Doppler shift.

[0032] S4: Calculate the minimum peak distance of wave characteristics based on the average wave period data, and then input the data set D1 and the minimum peak distance into the random forest and linear regression integrated model for training; Step S4 also includes: performing correlation analysis on the fluctuation features in the data set D1, screening feature subsets that are significantly correlated with the target variable, and then eliminating low-correlation feature values ​​through a set correlation threshold; The random forest and linear regression integrated model, the random forest sub-model integrates multiple independent regression trees and bagging strategies to enhance the tolerance to noise and outliers and the prediction accuracy and stability; the linear regression sub-model introduces regularization methods, including Lasso or Ridge regression methods. The calculation formula of the random forest and linear regression integrated model is: ; In the formula, represents the nonlinear prediction of random forest, represents the number of decision trees, represents the input feature value, represents independent and identically distributed random variables, represents the decision tree output, represents the linear prediction of linear regression, is the regression coefficient, which indicates the influence of each feature on the target variable. represents the error term, and Represent the weights of random forest regression and linear regression respectively.

[0033] S5: Input the dataset D2 into the trained integrated model to predict the minimum peak distance of the wave characteristics, identify the positions of the crests and troughs based on the prediction results, and then estimate the average wave period based on the relationship between wavelength and period.

[0034] The average wave period is estimated based on the relationship between wavelength and period. Specifically, the average wave period is calculated based on linear wave theory. The calculation formula is: ; In the formula, represents the wavelength, Represents the cycle, Represents the phase velocity of the wave.

[0035] like Figure 7 As shown, in order to verify the effectiveness of this method for estimating the average wave period, the evaluation results of this method are compared with the Berbi method. The comparison results are shown in Table 1; Table 1

[0036] The correlation coefficient CC, absolute value error AMD and root mean square error RMSD of the present invention are 0.9642, 0.0935 and 0.1098; the correlation coefficient CC, absolute value error AMD and root mean square error RMSD of the Berbi method are 0.9115, 0.1341 and 0.1619 respectively. The present invention is superior to the Berbi method in terms of correlation coefficient CC, absolute value error AMD and root mean square error RMSD.

[0037] The experimental results show that the present invention can not only accurately capture the nonlinear characteristics of waves, but also maintain a high prediction accuracy in complex wave phenomena, proving that the improved method is effective. By combining the advantages of multiple algorithms, the integrated model significantly improves the estimation accuracy of wave period and provides reliable technical support for practical applications.

[0038] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limitations of the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A method for estimating the average wave period of the coherent X-band based on machine learning, characterized in that: The following steps are involved: S1: Use coherent X-band radar to obtain the Doppler spectrum and time Doppler spectrum of a single echo on the sea surface; S2: Obtain the left boundary and the right boundary from the echo Doppler spectrum, and extract the Doppler frequency shift of the echo Doppler spectrum by moment estimation method; S3: The time-space sequence of radial velocity is obtained through Doppler frequency shift, and divided into a training data set and a test data set, and then the characteristic parameters of the wave characteristics are extracted to obtain the training data set D1 and the test data set D2; S4: Calculate the minimum peak distance of wave characteristics based on the average wave period data, and then input the data set D1 and the minimum peak distance into the random forest and linear regression integrated model for training; S5: Input the dataset D2 into the trained integrated model to predict the minimum peak distance of the wave characteristics, identify the positions of the crests and troughs based on the prediction results, and then estimate the average wave period based on the relationship between wavelength and period.

2. The method for estimating the average wave period of the coherent X-band based on machine learning according to claim 1, characterized in that: In the step S2, the center frequency is calculated by the echo Doppler spectrum, and the calculation formula is: ； In the formula, represents the Doppler frequency, represents the Doppler spectrum, Indicates the left border, Indicates the right border.

3. The method for estimating the average wave period of the coherent X-band based on machine learning according to claim 1, characterized in that: In step S3, the characteristic parameters of the extracted wave characteristics are specifically the mean, standard deviation, maximum value and minimum value of the extracted wave characteristics.

4. The method for estimating the average wave period of the coherent X-band based on machine learning according to claim 3, characterized in that: The calculation formula of the space-time series of radial velocity is: ； In the formula, represents the electromagnetic wavelength, represents the radial distance, represents the sampling time, f d represents the average Doppler shift.

5. The method for estimating the average wave period of the coherent X-band based on machine learning according to claim 1, characterized in that: Step S4 also includes: performing correlation analysis on the fluctuation features in the data set D1, screening feature subsets that are significantly correlated with the target variable, and then eliminating low-correlation feature values ​​through a set correlation threshold.

6. The method for estimating the average wave period of the coherent X-band based on machine learning according to claim 5, characterized in that: The random forest and linear regression integrated model, the random forest sub-model integrates multiple independent regression trees and bagging strategies to enhance the tolerance to noise and outliers and the prediction accuracy and stability; The linear regression sub-model introduces a regularization method, and the calculation formula of the random forest and linear regression integrated model is: ； In the formula, represents the nonlinear prediction of random forest, represents the number of decision trees, represents the input feature value, represents independent and identically distributed random variables, represents the decision tree output, represents the linear prediction of linear regression, is the regression coefficient, which indicates the influence of each feature on the target variable. represents the error term, and Represent the weights of random forest regression and linear regression respectively.

7. The method for estimating the average wave period of the coherent X-band based on machine learning according to claim 1, characterized in that: The average wave period is estimated based on the relationship between wavelength and period, specifically, the average wave period is calculated based on linear wave theory, and the calculation formula is: ； In the formula, represents the wavelength, Represents the cycle, Represents the phase velocity of the wave.

Images