A method for inversion of tropospheric atmospheric temperature and humidity profiles based on multi-objective genetic algorithm

Abstract

The present invention discloses a tropospheric atmospheric temperature and humidity profile inversion method based on a multi-objective genetic algorithm, and specifically relates to the technical field of computer systems based on a multi-objective genetic algorithm. The method specifically comprises the following steps: S1, first constructing an objective function, setting reasonable constraints such as temperature and humidity lapse rates, so that the errors between the calculated values ​​and the measured values ​​of the simulated brightness temperatures of the water vapor channel and the oxygen channel are minimized; in this embodiment, it is specifically explained that the objective function uses temperature and humidity data variables collected in the troposphere to represent the target form of the temperature and humidity profile; the temperature and humidity lapse rates are such that in the troposphere, the temperature and humidity decrease with increasing altitude; compared with the traditional tropospheric atmospheric temperature and humidity profile inversion method, the present invention adopts a multi-objective genetic NSGA2 algorithm, establishes an inversion model of the tropospheric atmospheric temperature and humidity profile through the NSGA2 algorithm, and improves the inversion accuracy of the tropospheric atmospheric temperature and humidity profile.

Description

Technical Field

[0001] The present invention relates to the technical field of computer systems based on multi-objective genetic algorithms, and more specifically, to a method for inverting tropospheric atmospheric temperature and humidity profiles based on multi-objective genetic algorithms. Background Art

[0002] The atmospheric temperature and humidity profile is obtained by using microwave radiometers and microwave remote sensing technology to receive atmospheric microwave radiation signals at different altitudes in the troposphere. The atmospheric microwave signals are then processed to invert the temperature and humidity curves corresponding to different altitudes in the troposphere.

[0003] The existing inversion algorithm for the atmospheric temperature and humidity profile of the troposphere is completely dependent on historical data. The present invention discloses a new inversion algorithm and process. Based on the statistical analysis of historical sounding data, the empirical range of atmospheric temperature and humidity parameters in each layer is obtained. An atmospheric temperature and humidity profile is randomly constructed from the experience library as the initial value. The simulated brightness temperature is calculated based on the microwave atmospheric radiation transmission model MonoRTM, and the degree of closeness to the measured brightness temperature is compared. The initial value is updated, and the calculation is continuously iterated to screen out feasible solutions. By setting constraints such as the vertical lapse rate of atmospheric temperature and humidity, the Pareto front is constrained from the feasible solution, and then the Pearson coefficient weighted average method is used to obtain a global satisfactory solution from the Pareto front.

[0004] The existing inversion methods for tropospheric atmospheric temperature and humidity profiles have problems such as low inversion accuracy, high reliance on historical data, and inability to be used in areas with a lack of historical meteorological data. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a method for inverting tropospheric atmospheric temperature and humidity profiles based on a multi-objective genetic algorithm, which solves the problems raised in the above-mentioned background technology by adopting a multi-objective genetic algorithm.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for inverting tropospheric atmospheric temperature and humidity profiles based on a multi-objective genetic algorithm, specifically comprising the following steps:

[0007] S1. First, construct the objective function and set the constraints such as temperature and humidity lapse rate so that the error between the calculated and measured values ​​of the simulated brightness temperature of the water vapor channel and the oxygen channel is minimized.

[0008] In a preferred embodiment, the objective function is to use the temperature and humidity data variables collected in the troposphere to represent the target form of the temperature and humidity profile; the temperature and humidity lapse rate is that in the troposphere, the air temperature and humidity decrease with increasing altitude, because in the tropospheric atmosphere, the ground long-wave radiation, the higher the altitude from the ground, the lower the temperature and water vapor humidity; the brightness temperature is a description of the radiation characteristics of the actual object itself, specifically, at the same wavelength, when the spectral radiation intensity of the actual object is consistent with that of the black body, the temperature of the black body is called the brightness temperature of the actual object at this wavelength, wherein the black body is an idealized object that can completely absorb all external electromagnetic radiation, and the constraint condition is to limit the variables that determine the construction of the objective function, and the variable condition that limits the objective function is the constraint condition.

[0009] S2. Conduct statistical analysis on conventional sounding data to obtain the normal distribution of temperature and humidity. Based on the normal distribution, the confidence interval is arranged to effectively eliminate abnormal sounding data and further narrow the empirical interval of temperature and humidity variables.

[0010] In a preferred embodiment, the NSGA2 algorithm in the genetic algorithm achieves the classification of the population by using a fast non-dominated sorting algorithm for the population, and calculates the crowding distance of the population individuals to maintain the diversity of the population. Therefore, the approximate solution obtained when the termination condition is met, the NSGA2 algorithm reduces the complexity of the non-inferior sorting genetic algorithm, has the characteristics of fast running speed and good convergence of the solution set, wherein the operation classification step of the fast non-dominated sorting algorithm is specifically to first classify the N in the population p = 0 is saved in the current set F1; then the individual i in the current set F1 is dominated by the individual set S i , all over S i Each individual p in the p =n p -1, if n p =0, then individual i is saved in set H; finally, the individual obtained in F1 is recorded as the individual in the first non-dominated layer, and H is used as the current set, and the process is repeated until the entire population is classified.

[0011] S3. Then, using the empirical formula for pressure height, the data received by the ground barometer is substituted into the pressure profile calculation formula to obtain the pressure profile;

[0012] In a preferred embodiment, a microwave radiometer is a highly sensitive instrument for passively measuring low-level microwave radiation of a target. The microwave radiometer is composed of an antenna that provides spatial resolution and collects energy, a receiver that converts the collected noise power into voltage, and a recording display. Specifically, the antenna receives a signal, which is input into the high-frequency channels of temperature and humidity respectively through a polarization divider. The intermediate frequency signal output by the high-frequency channel is then amplified by the intermediate frequency, detected by the square law detector, amplified by the low-frequency amplifier, and DC compensated. The signal is converted into a digital signal through an A / D converter and processed in the information processing and control module to invert the temperature and humidity data at various altitudes of the troposphere.

[0013] S4. Based on the inversion and optimization of the atmospheric temperature and humidity profile, a multi-objective function is constructed. During the calculation process, the distance mean square error method and the vector cosine method are used to obtain several groups of feasible solutions.

[0014] In a preferred embodiment, the mean square error method refers to the average of the square values ​​of the difference between each unit value and the average value of the sum of all unit values, which can reflect the degree of dispersion of each unit data. The specific formula is as follows:

[0015] Average Where n represents the number of this set of data, X1, X2, X3...X n分别表示 The specific value corresponding to the number of this set of data;

[0016] mean square error where σ 2 is the mean square error value; the vector cosine method refers to the use of trigonometric functions to solve vectors with direction and size. The specific formula used in the vector cosine method is in Represents two direction vectors.

[0017] S5. In several groups of feasible solutions, use Kalman filtering or Wiener filtering to smooth the humidity profiles in the feasible solutions according to the constraints;

[0018] In a preferred embodiment, the Kalman filter is an algorithm that optimally evaluates the input observation data in the form of a linear equation, filters the noise interference signal in the observation data, and restores the real data. In the computer processing process, the filtering process is performed in the form of an edited code. The code can be used as follows

[0019] {KFS->x_last=0; / / Kalman filter input

[0020] KFS->P_now=0; KFS->P_last=0.02; KFS->K=0;

[0021] KFS->Q_cov=0.005; / / process excitation noise covariance, adjustable parameters

[0022] KFS->R_cov=0.5; / / Measurement noise covariance, which is related to the nature of the instrument measurement and is adjustable

[0023] float KMFilter(KF_Struct*KFS,float z)

[0024] {KFS->P_now=KFS->P_last+KFS->Q_cov;

[0025] KFS->K=KFS->P_now / (KFS->P_now+KFS->R_cov);

[0026] KFS->out=0;}};The Wiener filter is an optimal estimation method for extracting noise interference signals from a stationary process based on the minimum mean square error. The filter is processed in the form of edit code. The code is as follows

[0027] % Input signal

[0028] A=1; % signal amplitude f=1000; % signal frequency fs=10^5; % sampling frequency t=(0:999); % sampling point

[0029] Mlag=100;%Correlation function length variable

[0030] x=A*cos(2*pi*f*t / fs);%xmean=mean(x);%xvar=var(x,1);% difference

[0031] xn=awgn(x,5); % Input Gaussian white noise with a signal-to-noise ratio of 20dB figure(1)

[0032] plot(t,xn)%Signal image; title('Input signal image')

[0033] xlabel('x-axis unit: t / s','color','b')

[0034] ylabel('y-axis unit: f / HZ','color','b')

[0035] Rxn=xcorr(xn,Mlag,'biased');% number

[0036] plot((-Mlag:Mlag),Rxn)%

[0037] title('Input signal autocorrelation function image')

[0038] [f,xi]=ksdensity(xn);% probability density, f is the probability density at the sample point xi subplot(222)

[0039] plot(xi,f)% rate density image; title('input signal probability density image')

[0040] X=fft(xn); % Fast discrete Fourier transform of signal sequence

[0041] Px=X.*conj(X) / 600;%subplot(223)

[0042] semilogy(t,Px)%Input sine wave signal sine wave signal mean.

[0043] S6. Finally, the feasible solution is optimized and the weighted average method is used to find the global optimal solution to obtain the atmospheric temperature and humidity profile.

[0044] In a preferred embodiment, the weighted average method is to use the meteorological observation values ​​measured by microwave radiometers arranged in chronological order in the past, and calculate the weighted arithmetic mean of the observation values ​​according to the number of times the variable appears in the chronological order as the weight, and use this number as a trend prediction method to predict the predicted value of the variable in the future period. The weighted average method uses the calculation formula: where x1, x2, x3...x n Indicates unit value, f1, f2, f3...f n Respectively represent the corresponding x1, x2, x3...x n The weight of .

[0045] In a preferred embodiment, the genetic algorithm is an adaptive global optimization probability search algorithm that simulates the genetic and evolutionary processes of organisms in a natural environment, and then uses computer simulation operations to spontaneously search for feasible solutions during the iterative process. It is suitable for solving some nonlinear problems or complex problems that are difficult to solve with traditional methods, and is widely used in combinatorial optimization, machine learning, signal processing and adaptive control. In the multi-objective genetic algorithm, the objectives and their corresponding weights are non-dominated, and the Pareto solution set is all mutually non-dominated vectors. The gene with the most information is obtained based on the dispersion distance of the non-dominated vectors. Therefore, the multi-objective genetic algorithm measures the distribution of discrete data information by crowding.

[0046] In a preferred embodiment, the empirical interval statistical learning is to linearly interpolate the conventional sounding data into 58 layers of profile data, and statistical analysis by month shows that the temperature and humidity (T, H) of each layer conform to the normal distribution. Substituting them into the formula Among them, the atmospheric temperature is completely normally distributed, and the relative humidity is skewed. Then the natural logarithm of the skewed distribution variable is taken and fitted into a log-normal distribution. The sounding data is taken as It can eliminate abnormal sounding data and complete statistical learning of empirical intervals.

[0047] In a preferred embodiment, the pressure profile is obtained by measuring the pressure on the ground, using the brightness temperature received by the microwave radiometer, which includes a function of atmospheric temperature, humidity, pressure, and liquid water, and calculating the pressure at each altitude layer by the pressure height formula. The specific pressure height formula can be used. Where H is the height of the troposphere in meters, P a is the pressure corresponding to H, unit hP a ,The purpose is that in the genetic algorithm, there are actually only two independent variables, namely ,atmospheric temperature and relative humidity. The pressure height formula is used to reduce the ,computational complexity and save computation time.

[0048] In a preferred embodiment, the multi-objective function is constructed by minimizing the difference between the brightness temperature of the 22GHz water vapor channel simulation and the actual observed brightness temperature, and minimizing the difference between the brightness temperature of the 58GHz oxygen channel simulation and the actual observed brightness temperature as the target data to establish a function, by The temperature and humidity profile is selected and the brightness temperature is calculated and simulated using the atmospheric microwave radiation transmission model MonoRTM. Then, the minimum temperature and humidity profile is obtained using the distance mean square error method and the vector cosine method. MonoRTM uses the Voigt line model to perform inversion calculations on the microwave radiometer.

[0049] In a preferred embodiment, the constraint condition is that in the microwave radiometer tropospheric temperature and humidity profile inversion algorithm, the absolute value of the difference between the temperature and humidity of two adjacent layers of the atmosphere is less than or equal to δ as the constraint condition, and the formula is substituted into In the calculation, several feasible solutions are obtained according to the formula restrictions, which not only retains the vertical gradient of atmospheric temperature and humidity, but also does not filter out the inverse changes of atmospheric temperature and humidity. Finally, Kalman filtering or Wiener filtering is used to smooth the humidity profile in the feasible solution.

[0050] In a preferred embodiment, the feasible solution optimization process is to manually select multiple groups of feasible solutions. According to the fact that the atmospheric temperature will not fluctuate violently in a short period of time, the vector group with large temperature change is eliminated by horizontal comparison with the profile at the previous moment, and the weighted average is calculated as the global optimal solution. The weight is the correlation coefficient γ between the current profile and the atmospheric profile at the previous moment. The specific solution formula of the correlation coefficient γ is: in, is the Pareto optimal solution for group K, yes Expectations, Y i is the atmospheric profile at the previous moment, μ Y It's Y i expectations, γ satisfies

[0051] The technical effects and advantages of the present invention are as follows:

[0052] Compared with the traditional tropospheric atmospheric temperature and humidity profile inversion method, the present invention adopts the multi-objective genetic NSGA2 algorithm. The NSGA2 algorithm establishes the effect of the tropospheric atmospheric temperature and humidity profile inversion model, thereby improving the inversion accuracy of the tropospheric atmospheric temperature and humidity profile. The NSGA2 algorithm has adaptive capabilities and can meet the characteristics of microwave radiometers used in areas with a lack of historical meteorological data. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flowchart of the tropospheric atmospheric temperature and humidity profile inversion method based on a multi-objective genetic algorithm of the present invention. DETAILED DESCRIPTION

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0055] Example 1

[0056] This embodiment provides Figure 1 The present invention provides a method for inverting tropospheric atmospheric temperature and humidity profiles based on a multi-objective genetic algorithm, which specifically includes the following steps:

[0057] S1. First, construct the objective function and set the constraints such as temperature and humidity lapse rate so that the error between the calculated and measured values ​​of the simulated brightness temperature of the water vapor channel and the oxygen channel is minimized.

[0058] In this embodiment, it is specifically explained that the objective function is to use the temperature and humidity data variables collected in the troposphere to represent the target form of the temperature and humidity profile; the temperature and humidity lapse rate is that in the troposphere, the air temperature and humidity decrease with increasing altitude, because in the tropospheric atmosphere, the ground long-wave radiation, the higher the altitude from the ground, the lower the temperature and water vapor humidity; the brightness temperature is a description of the radiation characteristics of the actual object itself, specifically, at the same wavelength, when the spectral radiation intensity of the actual object is consistent with that of a black body, the temperature of the black body is called the brightness temperature of the actual object at that wavelength, where the black body is an idealized object that can completely absorb all external electromagnetic radiation, and the constraint condition is to limit the variables that determine the construction of the objective function, and the variable condition that limits the objective function is the constraint condition.

[0059] S2. Statistical analysis of conventional sounding data is performed to obtain the normal distribution of temperature and humidity. Confidence intervals are arranged based on the normal distribution to effectively eliminate abnormal sounding data and further narrow the empirical intervals of temperature and humidity variables.

[0060] In this embodiment, the NSGA2 algorithm in the genetic algorithm is specifically described. It achieves population classification by using a fast non-dominated sorting algorithm for the population, and calculates the crowding distance of the population individuals to maintain the diversity of the population. Therefore, the approximate solution obtained when the termination condition is met, the NSGA2 algorithm reduces the complexity of the non-inferior sorting genetic algorithm, has the characteristics of fast running speed and good convergence of the solution set, wherein the operation classification step of the fast non-dominated sorting algorithm is specifically to first sort the N in the population. p = 0 is saved in the current set F1; then the individual i in the current set F1 is dominated by the individual set S i , all over S i Each individual p in the p =n p -1, if n p =0, then individual i is saved in set H; finally, the individual obtained in F1 is recorded as the individual in the first non-dominated layer, and H is used as the current set, and the process is repeated until the entire population is classified.

[0061] S3. Then, using the empirical formula for pressure height, the data received by the ground barometer is substituted into the pressure profile calculation formula to obtain the pressure profile;

[0062] In this embodiment, it is specifically described that the microwave radiometer is a highly sensitive instrument for passively measuring low-level microwave radiation from a target. The microwave radiometer is composed of an antenna that provides spatial resolution and collects energy, a receiver that converts the collected noise power into voltage, and a recording display. Specifically, the antenna receives a signal, which is input into the high-frequency channels of temperature and humidity respectively through a polarization divider. The intermediate frequency signal output by the high-frequency channel is then amplified by the intermediate frequency, detected by a square-law detector, amplified by a low-frequency amplifier, and DC-compensated. The signal is then converted into a digital signal through an A / D converter. After being processed in an information processing and control module, the temperature and humidity data at various altitudes in the troposphere are inverted.

[0063] S4. Based on the inversion and optimization of the atmospheric temperature and humidity profile, a multi-objective function is constructed. During the calculation process, the distance mean square error method and the vector cosine method are used to obtain several groups of feasible solutions.

[0064] In this embodiment, it is specifically explained that the mean square error method refers to the average of the square values ​​of the difference between each unit value and the average value of the sum of all unit values, which can reflect the degree of dispersion of each unit data. The specific formula used is as follows:

[0065] Average Where n represents the number of this set of data, X1, X2, X3...X n分别表示 The specific value corresponding to the number of this set of data;

[0066] mean square error where σ 2 is the mean square error value; the vector cosine method refers to the use of trigonometric functions to solve vectors with direction and size. The specific formula used in the vector cosine method is in Represents two direction vectors.

[0067] S5. In several groups of feasible solutions, use Kalman filtering or Wiener filtering to smooth the humidity profiles in the feasible solutions according to the constraints;

[0068] In this embodiment, it is specifically described that the Kalman filter is an algorithm that optimally evaluates the input observation data in the form of a linear equation, filters the noise interference signal in the observation data, and restores the real data filtering technology. In the computer processing process, the filtering process is performed in the form of editing code, and the code can be used as follows

[0069] {KFS->x_last=0; / / Kalman filter input

[0070] KFS->P_now=0; KFS->P_last=0.02; KFS->K=0;

[0071] KFS->Q_cov=0.005; / / process excitation noise covariance, adjustable parameters

[0072] KFS->R_cov=0.5; / / Measurement noise covariance, which is related to the nature of the instrument measurement and is adjustable

[0073] float KMFilter(KF_Struct*KFS,float z)

[0074] {KFS->P_now=KFS->P_last+KFS->Q_cov;

[0075] KFS->K=KFS->P_now / (KFS->P_now+KFS->R_cov);

[0076] KFS->out=0;}};The Wiener filter is an optimal estimation method for extracting noise interference signals from a stationary process based on the minimum mean square error. The filter is processed in the form of edit code. The code is as follows

[0077] % Input signal

[0078] A=1; % signal amplitude f=1000; % signal frequency fs=10^5; % sampling frequency t=(0:999); % sampling point

[0079] Mlag=100;%Correlation function length variable

[0080] x=A*cos(2*pi*f*t / fs);%xmean=mean(x);%xvar=var(x,1);% difference

[0081] xn=awgn(x,5); % Input Gaussian white noise with a signal-to-noise ratio of 20dB figure(1)

[0082] plot(t,xn)%Signal image; title('Input signal image')

[0083] xlabel('x-axis unit: t / s','color','b')

[0084] ylabel('y-axis unit: f / HZ','color','b')

[0085] Rxn=xcorr(xn,Mlag,'biased');% number

[0086] plot((-Mlag:Mlag),Rxn)%

[0087] title('Input signal autocorrelation function image')

[0088] [f,xi]=ksdensity(xn);% probability density, f is the probability density at the sample point xi subplot(222)

[0089] plot(xi,f)% rate density image; title('input signal probability density image')

[0090] X=fft(xn); % Fast discrete Fourier transform of signal sequence

[0091] Px=X.*conj(X) / 600;%subplot(223)

[0092] semilogy(t,Px)%Input sine wave signal sine wave signal mean.

[0093] S6. Finally, the feasible solution is optimized and the weighted average method is used to find the global optimal solution to obtain the atmospheric temperature and humidity profile.

[0094] In this embodiment, the weighted average method is specifically described as using a number of past microwave radiometer meteorological observation values ​​arranged in chronological order, using the number of times the variable appears in the chronological order as a weight, and calculating the weighted arithmetic mean of the observation values. This number is used as a trend prediction method to predict the predicted value of the variable in the future period. The weighted average method uses the calculation formula: where x1, x2, x3...x n Indicates unit value, f1, f2, f3...f n Respectively represent the corresponding x1, x2, x3...x n The weight of .

[0095] The difference between this embodiment and the prior art is that the inversion method of the tropospheric atmospheric temperature and humidity profile adopts a genetic algorithm combined with a multi-objective function to perform inversion operations. The genetic algorithm and the multi-objective function constitute a multi-objective genetic algorithm. The entire process is not available in the prior art. The constraints required for constructing the multi-objective function cannot cause oscillation of the objective function. This embodiment limits the specific algorithm or program process.

[0096] In this embodiment, it is specifically described that the genetic algorithm is an adaptive global optimization probability search algorithm that simulates the genetic and evolutionary processes of organisms in a natural environment, and then uses computer simulation operations to spontaneously search for feasible solutions during the iterative process. It is suitable for solving some nonlinear problems or complex problems that are difficult to solve with traditional methods, and is widely used in combinatorial optimization, machine learning, signal processing and adaptive control. In the multi-objective genetic algorithm, the objectives and their corresponding weights are non-dominated, and the Pareto solution set is all mutually non-dominated vectors. The gene with the most information is obtained based on the dispersion distance of the non-dominated vectors. Therefore, the multi-objective genetic algorithm measures the distribution of discrete data information by crowding.

[0097] In this embodiment, it is specifically explained that the empirical interval statistical learning is to linearly interpolate the conventional sounding data into 58 layers of profile data. The monthly statistical analysis shows that the temperature and humidity (T, H) of each layer conform to the normal distribution. Substituting them into the formula Among them, the atmospheric temperature is completely normally distributed, and the relative humidity is skewed. Then the natural logarithm of the skewed distribution variable is taken and fitted into a log-normal distribution. The sounding data is taken as It can eliminate abnormal sounding data and complete statistical learning of empirical intervals.

[0098] In this embodiment, it is specifically explained that the pressure profile is obtained based on the ground-measured pressure, using the brightness temperature received by the microwave radiometer, which includes a function of atmospheric temperature, humidity, pressure, and liquid water, and calculating the pressure at each altitude layer through the pressure height formula. The specific pressure height formula can be used. Where H is the height of the troposphere in meters, P a is the pressure corresponding to H, unit hP a ,The purpose is that in the genetic algorithm, there are actually only two independent variables, namely ,atmospheric temperature and relative humidity. The pressure height formula is used to reduce the ,computational complexity and save computation time.

[0099] In this embodiment, it is specifically described that the multi-objective function is constructed to minimize the difference between the simulated brightness temperature of the 22GHz water vapor channel and the actual observed brightness temperature, and to minimize the difference between the simulated brightness temperature of the 58GHz oxygen channel and the actual observed brightness temperature as the target data to establish a function. The temperature and humidity profile is selected and the brightness temperature is calculated and simulated using the atmospheric microwave radiation transmission model MonoRTM. Then, the minimum temperature and humidity profile is obtained using the distance mean square error method and the vector cosine method. MonoRTM uses the Voigt line model to perform inversion calculations on the microwave radiometer.

[0100] In this embodiment, it is specifically explained that the constraint condition is that in the microwave radiometer tropospheric temperature and humidity profile inversion algorithm, the absolute value of the temperature and humidity difference between two adjacent layers of the atmosphere is less than or equal to δ as the constraint condition, and the formula is substituted into In the calculation, several feasible solutions are obtained according to the formula restrictions, which not only retains the vertical gradient of atmospheric temperature and humidity, but also does not filter out the inverse changes of atmospheric temperature and humidity. Finally, Kalman filtering or Wiener filtering is used to smooth the humidity profile in the feasible solution.

[0101] In this embodiment, it is specifically explained that the feasible solution optimization process is to manually select multiple groups of feasible solutions. According to the fact that the atmospheric temperature will not fluctuate violently in a short period of time, the vector group with large temperature change is eliminated by horizontal comparison with the profile at the previous moment, and the weighted average is calculated as the global optimal solution. The weight is the correlation coefficient γ between the current profile and the atmospheric profile at the previous moment. The specific solution formula of the correlation coefficient γ is: in, is the Pareto optimal solution for group K, yes Expectations, Y i is the atmospheric profile at the previous moment, μ Y It's Y i expectations, γ satisfies

[0102] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for inverting tropospheric atmospheric temperature and humidity profiles based on a multi-objective genetic algorithm, characterized by: The specific steps include: S1. First, construct the objective function and set the temperature and humidity lapse rate constraints to minimize the error between the calculated and measured brightness temperatures of the water vapor and oxygen channels. S2. Statistical analysis of conventional sounding data is performed to obtain the normal distribution of temperature and humidity. Confidence intervals are arranged based on the normal distribution to effectively eliminate abnormal sounding data and further narrow the empirical intervals of temperature and humidity variables. S3. Then, using the empirical formula for pressure height, the data received by the ground barometer is substituted into the pressure profile calculation formula to obtain the pressure profile; S4. Based on the inversion and optimization of the atmospheric temperature and humidity profile, a multi-objective function is constructed. During the calculation process, the distance mean square error method and the vector cosine method are used to obtain several groups of feasible solutions. S5. In several groups of feasible solutions, use Kalman filtering or Wiener filtering to smooth the humidity profiles in the feasible solutions according to the constraints; S6. Finally, the feasible solution is optimized and the weighted average method is used to find the global optimal solution to obtain the atmospheric temperature and humidity profile; Empirical interval statistical learning is to linearly interpolate conventional sounding data into 58 layers of profile data. Statistical analysis by month shows that the temperature and humidity (T, H) of each layer conform to the normal distribution. Among them, the atmospheric temperature is completely normally distributed, and the relative humidity is skewed. Then the natural logarithm of the skewed distribution variable is taken and fitted into a log-normal distribution. The sounding data is taken as Ability to eliminate abnormal sounding data and complete statistical learning of empirical intervals; The multi-objective function is constructed by minimizing the difference between the brightness temperature of the 22GHz water vapor channel simulation and the actual observation brightness temperature, and minimizing the difference between the brightness temperature of the 58GHz oxygen channel simulation and the actual observation brightness temperature as the target data to establish a function. The temperature and humidity profiles were selected and the brightness temperature was calculated and simulated using the atmospheric microwave radiation transmission model MonoRTM. Then the minimum temperature and humidity profile was obtained using the distance mean square error method and the vector cosine method. The constraint condition is that in the microwave radiometer tropospheric temperature and humidity profile inversion algorithm, the absolute value of the temperature and humidity difference between two adjacent layers of the atmosphere is less than or equal to δ as the constraint condition, and it is substituted into the formula Calculate in the middle and obtain several feasible solutions according to the formula restrictions; The feasible solution optimization process is to manually select multiple feasible solutions. According to the fact that the atmospheric temperature will not fluctuate violently in a short period of time, the vector group with large temperature change is eliminated by horizontal comparison with the previous moment profile. The weighted average is calculated as the global optimal solution. The weight is the correlation coefficient γ between the current profile and the atmospheric profile at the previous moment. The specific solution formula of the correlation coefficient γ is: in, is the Pareto optimal solution for group K, yes Expectations, Y i is the atmospheric profile at the previous moment, μ Y It's Y i expectations, γ satisfies 2. The method for inverting tropospheric atmospheric temperature and humidity profiles based on a multi-objective genetic algorithm according to claim 1, characterized in that: The genetic algorithm is an adaptive global optimization probabilistic search algorithm that simulates the genetic and evolutionary processes of organisms in the natural environment. It can spontaneously search for feasible solutions during the iterative process. In the multi-objective genetic algorithm, the objectives and their corresponding weights are non-dominated. The Pareto solution set is all mutually non-dominated vectors. The gene with the most information is obtained based on the dispersion distance of the non-dominated vectors.

3. The method for inverting tropospheric atmospheric temperature and humidity profiles based on a multi-objective genetic algorithm according to claim 1, characterized in that: The pressure profile is obtained by measuring the pressure on the ground, using the brightness temperature received by the microwave radiometer, which includes a function of atmospheric temperature, humidity, pressure, and liquid water, and calculating the pressure at each altitude layer through the pressure height formula. The specific pressure height formula can be used. Where H is the height of the troposphere in meters, and Pa is the pressure corresponding to H in hPa.

Images