Atmospheric radiation transmission signal reconstruction method

Abstract

The invention discloses an atmospheric radiation transmission signal reconstruction method, and belongs to the technical field of atmospheric optical transmission. The method comprises the following steps: constructing an atmospheric radiation transfer Monte Carlo model, and simulating photon emission, transmission, scattering and absorption processes; extracting a time broadening function h (t) through simulation under the condition of no modulation signal, and defining the time broadening function h (t) as a time point spread function of an atmospheric channel; based on the fitted error model and the time model, the optimal photon packet number N for simulation is efficiently determined by using a Bayesian optimization algorithm; and finally, realizing accurate reconstruction of the received signal y (t) by using convolution operation of the time broadening function h (t) and the original signal S (t). According to the method, the Monte Carlo simulation parameters are intelligently optimized, the calculation cost is remarkably reduced on the premise that the precision is guaranteed, and an efficient technical scheme is provided for rapid signal reconstruction in remote sensing detection and laser communication.

Description

Technical Field

[0001] This invention belongs to the field of atmospheric optical transmission technology, specifically relating to a numerical simulation method for atmospheric radiation transmission, and more particularly to a method that integrates Bayesian optimization algorithms to efficiently determine Monte Carlo simulation parameters and uses time-stretching functions for signal convolution reconstruction. Background Technology

[0002] Atmospheric radiative transfer simulation is a core foundation for fields such as remote sensing, laser communication, and climate prediction. The Monte Carlo method, through random photon tracking and statistical convergence mechanisms, can flexibly simulate the dynamic process of photon-atmosphere interaction, providing a powerful numerical analysis tool for radiative transfer in complex atmospheric environments.

[0003] However, the Monte Carlo method has a significant drawback: its computational cost is high. The accuracy of the simulation results is related to the number of photon packets tracked. N Directly related N A value that is too small will result in excessive statistical noise and unreliable results; conversely, blindly increasing the value will lead to... N While this value can improve accuracy, the computation time increases exponentially, making it difficult to meet the needs of rapid simulation in practical engineering. In existing technologies, to determine a suitable... N To obtain accurate values, researchers typically need to conduct numerous repetitive experiments, manually adjusting parameters and observing the results to balance accuracy and efficiency. This method is inefficient and relies heavily on experience.

[0004] Furthermore, existing atmospheric radiative transfer models mostly focus on improving transmission efficiency and cannot calculate the time broadening effect caused by multiple scattering of light during transmission, and lack effective means to utilize this effect for efficient signal reconstruction.

[0005] Therefore, there is an urgent need in this field to design a technical solution that can intelligently optimize the Monte Carlo simulation process, significantly reduce computational costs, and fully utilize time-spanning characteristics to achieve rapid signal reconstruction. Summary of the Invention

[0006] In view of the above-mentioned situation in the field, the present invention provides an atmospheric radiation transmission signal reconstruction method. The present invention can intelligently determine the optimal simulation parameters, significantly improve the calculation efficiency while ensuring accuracy, and realize accurate and fast reconstruction of the received signal by utilizing the time stretching effect.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for reconstructing atmospheric radiation transmission signals includes the following steps: Step 1: Construct a Monte Carlo model of atmospheric radiative transport: Establish a radiative transport model of atmospheric mixture of molecules and aerosols to simulate the emission, transport, scattering and absorption of photons, and record the energy weight and transport time of photon packets arriving at the receiver. Step 2: Simulate and extract the time-spanning function: Without adding any modulation signal to the simulated light source, run the Monte Carlo model to statistically analyze the distribution of photon packet energy over time at a specified receiver. After normalizing this distribution, the time-spanning function characterizing the atmospheric channel's time response is obtained. h ( t ); Step 3: Signal Convolution Reconstruction: Using the time-spanning function h(t) obtained in Step 2 and the original signal S(t), the received signal y(t) is reconstructed. S ( t )⊗ h ( t ).

[0008] Preferably, after step two, the following steps are performed: using a Bayesian optimization algorithm, the optimal number of photon packets N1 for simulation is determined.

[0009] Furthermore, Bayesian optimization algorithms include: Based on N Based on the error and computation time obtained from the simulation, error models and time models are established respectively. A comprehensive objective function is constructed based on the aforementioned error model and time model; A Bayesian optimizer, using a Gaussian process as a surrogate model, is employed to minimize the comprehensive objective function in order to determine the optimal number of photon packets. N 1.

[0010] In this invention, the Bayesian optimization algorithm uses the fitted error model and time model to evaluate the objective function, instead of running a time-consuming Monte Carlo simulation for each evaluation, thereby greatly improving the optimization efficiency.

[0011] Furthermore, intelligent optimization of the number of photon packets N Specifically as follows: Data acquisition and model fitting: First, in different... N Value (range 10) 5 ~10 8 Monte Carlo simulations were performed to obtain the corresponding mean absolute error ε and computation time t. Based on the error theory of the Monte Carlo method, the function model ε = a / √ N + b is used to fit the error data, where a and b are parameters determined through fitting; simultaneously, a time model is used. t =c·log 10 (N The time model is fitted to the computation time data using the formula ) + d, where c and d are parameters determined through the fitting process. If this time model affects the computation time... t i The data fitting error is high; linear interpolation can be used to adjust the computation time. t i Supplementing the data reduces simulation errors. Initialization: Based on the fitted error model and time model, the search range for the number of photon packets N is set to

[10] . 5 10 8 ]; Objective function construction: Define the comprehensive objective function f( N )=w1(ε(N) / ε max )+w2(t(N) / t max ), where w1 and w2 are weighting coefficients (e.g., w1=0.8, w2=0.2), ε max and t max The maximum value among the known data is used for normalization. Bayesian optimization loop: A Gaussian process is used as the surrogate model (a Gaussian process is a probabilistic statistical model used to establish a nonlinear mapping between input variable N and output objective function value f(N). The initial point uses a known N and the corresponding integrated objective function value. The expected improvement (EI) acquisition function is used, and its expression is: in, μ (N) and σ (N) represents the predicted mean and standard deviation of the Gaussian process at point N, respectively. f min The minimum objective function observed so far. Φ (·)and ϕ (·) represent the cumulative distribution function and probability density function of the standard normal distribution, respectively. Z =( f min - μ (N)) / ( σ (N)). This sampling function can effectively balance exploring the region of uncertainty with utilizing the current optimal solution to guide the selection of sampling points, thereby determining the next N value that can minimize the objective function as much as possible; Evaluation and Update: The newly selected N i Substitute the values ​​into the fitted error model and calculate its prediction error ε. pred Simultaneously, substitute the fitted time model into the time model and calculate its prediction calculation time t. predNext, N is calculated according to the objective function formula f(N). i The value corresponds to the comprehensive objective function value f. pred Finally, this pair of new observation data (N) i , f pred Add it to the historical observation dataset and update the parameters of the Gaussian process surrogate model to more accurately reflect the mapping relationship between N and f(N); Outputting the optimal solution: After several iterations, the Bayesian optimizer outputs the relatively optimal number of photon packets, N1, which corresponds to the point with the smallest comprehensive objective function value in the historical observation data. According to model predictions, using this N value for simulation achieves the best balance between accuracy and computation time, resulting in the lowest overall score. This method optimizes by fitting a model instead of performing a real simulation.

[0012] After step three, perform the following steps: attach the original signal to be transmitted to the analog light source. S ( t ), using the obtained optimal number of photon packets N 1. Run the Monte Carlo model to obtain the simulation output of the receiver; reconstruct the signal. y ( t After comparing and verifying with the simulation output, it is used as the prediction of the received signal under atmospheric channel conditions.

[0013] Furthermore, in step one, the model uses a combined phase function constructed from Rayleigh scattering and the Double-Henyey Greenstein (DHG) aerosol scattering phase function to characterize the scattering process in the atmosphere where molecules and aerosols mix.

[0014] Furthermore, in step one, the model divides the atmosphere into multiple uniform spherical shells, uses a Cartesian coordinate system to track the photon position, and uses a dynamic spherical polar coordinate system to track the photon direction.

[0015] Furthermore, the time stretching function h ( t The distribution of photon packet energy at a specified altitude over time is statistically analyzed and normalized, and is defined as the time-point diffusion function of the atmospheric radiation transmission channel.

[0016] Compared with the prior art, the atmospheric radiation transmission signal reconstruction method of the present invention has the following beneficial effects: 1) This invention is the first to introduce Bayesian optimization algorithm into Monte Carlo simulation of atmospheric radiative transfer. By establishing error and time models, a target function that comprehensively considers accuracy and efficiency is constructed, thereby achieving the target number of photon packets for simulation. NIts intelligent and efficient determination overcomes the drawbacks of traditional methods that rely on manual trial and error, significantly reducing computational costs and improving simulation efficiency while ensuring simulation accuracy.

[0017] 2) This invention explicitly proposes to characterize the time broadening effect as a time-point spread function of the atmospheric channel, and achieves fast and high-precision reconstruction of the received signal (coefficient of determination) through convolution operations. R (² can reach over 0.99). This invention avoids the time-consuming Monte Carlo simulation for every new signal, making real-time signal processing possible.

[0018] 3) This invention provides an efficient and reliable technical tool for predicting and correcting signal attenuation and waveform distortion in fields such as remote sensing (e.g., lidar, space optical remote sensing) and laser communication, and has broad application prospects. Attached Figure Description

[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0020] Figure 1 This is a schematic diagram of the overall process of a Bayesian-optimized atmospheric radiation transmission signal reconstruction method according to a preferred embodiment of the present invention.

[0021] Figure 2 This is a schematic diagram of the geometric relationship between the Earth, atmosphere, and point light source in a Monte Carlo model constructed according to a preferred embodiment of the present invention.

[0022] Figure 3 This is a flowchart of the photon transmission simulation using the Monte Carlo model in a preferred embodiment of the present invention.

[0023] Figure 4 This is a schematic diagram comparing the simulated output of the received signal with the convolutional reconstruction result in a preferred embodiment.

[0024] Figure 5 The time-spanning h(t) diagram output by Monte Carlo model simulation in a preferred embodiment of the present invention is shown without applying a modulation signal to the light source. Detailed Implementation

[0025] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0026] This embodiment uses the transmission of 532nm and 1550nm lasers in a satellite-to-ground link from an altitude of 1.5km to 500km as an example to provide a more detailed explanation of the technical solution of the present invention.

[0027] like Figure 1 As shown in the figure, this embodiment presents a method for reconstructing atmospheric radiative transfer signals based on Bayesian optimization. The specific steps are as follows: S1. Construct a Monte Carlo model of atmospheric radiative transport to simulate the emission, transmission, scattering, and absorption of photons; like Figure 2 As shown, the atmospheric geometry is defined as follows: A spherical shell within a 50 km altitude range, with sea level as the reference, constitutes the atmosphere. The atmosphere is divided into 50 layers at 1 km intervals, each layer considered as homogeneous. The model uses a Cartesian coordinate system to track photon positions and a dynamic polar coordinate system to track photon directions. The simulation process is as follows... Figure 3 As shown: First, initialize the photon packet, assigning it an initial optical thickness s=0, transmission direction, and energy weight w=1, etc. Check if s=0 holds. If it does, assign a new optical thickness using s=-lnξ, where ξ is a uniformly random number between [0,1]. If not, no new optical thickness is assigned. Calculate the distance D from the photon packet to the boundary along the transmission direction. Second, determine if the photon packet has exited the layer. Check if D×μt≤s holds, where μt is the extinction coefficient. If this equation holds, the photon packet moves by D and s is updated; if not, the photon packet moves by s / μt and s is reset to zero. Simultaneously, the photon packet undergoes absorption and scattering events. The absorption event is characterized by updating the energy weight w, i.e.: w = w - (μa / μt)w Where μa is the absorption coefficient; Scattering events are characterized by changes in the direction of photon packets. The deflection angle is generated by the scattering function. θ and azimuth φ Determine the new direction cosine , and (Direction in Cartesian coordinates). Let 'a' represent the direction cosine of the initial direction, then: New Direction Cosine for: in, It is perpendicular to The unit vector chosen here. Belonging to the xy plane, when the initial direction of the photon is cosine... At that time, the new directional cosine is updated as follows: Ultimately, the decision to continue or terminate transmission is made based on whether the energy weight w of the photon packet is less than the threshold and whether the photon packet has reached the target area.

[0028] S2. Without adding a modulation signal to the simulated light source, run the Monte Carlo model to statistically analyze the distribution of photon packet energy over time at the receiving end. After normalizing the distribution, obtain the time broadening function h(t) that characterizes the time response of the atmospheric channel. 2.1 Simulate and extract the time stretching function Without adding any modulation signal, a large number (e.g., 10) of signals are transmitted to the Monte Carlo model. 8 (Number) photon packets, statistically analyzing the energy of photon packets received at an altitude of 500 km over time. t The distribution of is normalized to obtain as shown below. Figure 5 The time stretching function shown h ( t This function intuitively reflects the light pulse broadening effect caused by atmospheric multiple scattering.

[0029] 2.2 Intelligent optimization of photon packet quantity N (Core optimization steps) A Bayesian optimization algorithm was used to determine the relatively optimal number of photon packets N1 for the simulation. The details are as follows: 1) Data Acquisition and Model Fitting: First, in different... N Perform Monte Carlo simulations on the values ​​to obtain the corresponding mean absolute error ε and computation time. t Based on the theoretical model ε = a / √N + b, the values ​​of parameters a and b are obtained by fitting using the nonlinear least squares method. Simultaneously, the time model t = c·log 10 We fit (N)+d to obtain parameters c and d.

[0030] 2) Initialization: Set the number of photon packets N The search range is [10 5 10 8 ].

[0031] 3) Objective function construction: Define the comprehensive objective function f( N ) = 0.8·(ε( N ) / 0.4807) + 0.2·(t( N The objective function is calculated as follows: (0.4807 / 29.734), where 0.4807 and 29.734 are the maximum values ​​of error and time in the known data, respectively, used for normalization. This objective function comprehensively considers simulation accuracy (weight 0.8) and computational efficiency (weight 0.2).

[0032] 4) Bayesian optimization loop: Using a Gaussian process as the surrogate model, the initial point is a known one. NAnd the corresponding comprehensive objective function value. Using the Expected Improvement (EI) acquisition function, determine the next acquisition function that can minimize the objective function as much as possible. N The value is calculated directly based on the fitted error model and time model. N The prediction error, computation time, and integrated objective function value corresponding to the new data points are provided without needing to run an actual simulation. The Gaussian process model is updated by adding the new data points to the observation dataset.

[0033] 5) Output the optimal solution: After Bayesian optimization, output the optimal number of photon packets. N 1. Based on model predictions, using... N Simulation with a value of 1 achieves the best balance between accuracy and computation time, resulting in the lowest overall score.

[0034] S3. Signal Convolution Reconstruction: The received signal y(t) is reconstructed by convolving the time-spanning function h(t) obtained in step two with the original signal S(t).

[0035] Specifically, a sinusoidal modulation signal is added to the analog light source. S ( t )(like Figure 4 (as shown in (a)). Using the optimal value N 1. Run the model to obtain the simulation output from the receiver. Simultaneously, apply the time stretching function obtained in step two. h ( t )and S ( t Perform convolution operation to obtain the convolution result. y ( t ).like Figure 4 As shown in (b) above, the convolutional reconstruction result y ( t The results closely match the time-consuming Monte Carlo simulation output, with a high coefficient of determination. R The result reached 0.9968, which fully verifies the effectiveness of the present invention.

[0036] This invention establishes an error model and a time model to construct an objective function that comprehensively considers accuracy and efficiency, and then combines Bayesian optimization to quickly determine the optimal solution. N This method intelligently balances computational accuracy and cost, and utilizes convolution theory to achieve rapid signal reconstruction, providing an innovative solution for accurate and efficient simulation of atmospheric radiation transmission.

[0037] In summary, this invention belongs to the field of atmospheric radiative transfer modeling technology, specifically disclosing a method for time-broadening modeling and efficient reconstruction of atmospheric radiative transfer. This invention constructs a Monte Carlo radiative transfer model of a mixed atmosphere of molecules and aerosols. By simulating multiple scattering processes, it reveals the time-broadening effect and characterizes it as a time-point spread function of the atmospheric channel. Then, it utilizes convolution operations to achieve accurate reconstruction of the received signal. To improve model efficiency, this invention integrates error analysis and adaptive sampling mechanisms, introduces Bayesian optimization and anomaly detection algorithms, and intelligently predicts the relatively optimal number of photon packets for simulation from both error and computational cost dimensions. N Apply this N The value can significantly reduce computation time while ensuring reconstruction accuracy, and is suitable for efficient simulation and reconstruction of optical signals in remote sensing and laser communication.

[0038] The above description is merely a detailed explanation of preferred embodiments and principles of the present invention. For those skilled in the art, there may be changes in specific implementation methods based on the ideas provided by the present invention, and these changes should also be considered within the scope of protection of the present invention.

Claims

1. A method for reconstructing atmospheric radiative transmission signals, characterized in that, Includes the following steps: Step 1: Construct a Monte Carlo model of atmospheric radiative transport to simulate the emission, transmission, scattering, and absorption of photons; Step 2: Without adding a modulation signal to the simulated light source, run the Monte Carlo model to statistically analyze the distribution of photon packet energy over time at the receiver. After normalizing the distribution, obtain the time broadening function h(t) that characterizes the time response of the atmospheric channel. Step 3: Convolve the time-spanning function h(t) obtained in Step 2 with the original signal S(t) to reconstruct the received signal y(t).

2. The method for reconstructing atmospheric radiation transmission signals according to claim 1, characterized in that, In step one, the Monte Carlo model uses a joint phase function constructed from Rayleigh scattering and the Double-Henyey Greenstein aerosol scattering phase function to characterize the scattering process in the atmosphere where molecules and aerosols are mixed.

3. The method for reconstructing atmospheric radiation transmission signals according to claim 1 or 2, characterized in that, In step one, the model divides the atmosphere into multiple uniform spherical shells and uses a Cartesian coordinate system to track the photon position and a dynamic spherical polar coordinate system to track the photon direction.

4. A method for reconstructing atmospheric radiation transmission signals according to claim 1 or 2, characterized in that, In step two, the time broadening function h(t) is obtained by statistically analyzing the distribution of photon packet energy over time at a specified altitude and then normalizing it. It is defined as the time-point diffusion function of the atmospheric radiation transmission channel.

5. The method for reconstructing atmospheric radiation transmission signals according to claim 4, characterized in that, After step two, the following steps are performed: The Bayesian optimization algorithm is used to determine the optimal number of photon packets N1 for simulation.

6. The method for reconstructing atmospheric radiation transmission signals according to claim 5, characterized in that, The Bayesian optimization algorithm mentioned includes: Based on N Based on the error and computation time obtained from the simulation, error models and time models are established respectively. A comprehensive objective function is constructed based on the aforementioned error model and time model; A Bayesian optimizer, using a Gaussian process as a surrogate model, is employed to minimize the comprehensive objective function in order to determine the optimal number of photon packets. N 1.

7. The method for reconstructing atmospheric radiation transmission signals according to claim 6, characterized in that, The error model is ε = a / √N+b, where a and b are parameters determined through fitting.

8. The method for reconstructing atmospheric radiation transmission signals according to claim 7, characterized in that, The time model is t =c log 10 ( N ) +d, where c and d are parameters determined by fitting.

9. The method for reconstructing atmospheric radiation transmission signals according to claim 8, characterized in that, The comprehensive objective function is f( N =w1·(ε(N) / ε max ) +w2·(t(N) / t max ), where w1 and w2 are weighting coefficients, ε max and t max This is the normalization factor.

10. The method for reconstructing atmospheric radiation transmission signals according to claim 9, characterized in that, After step three, perform the following steps: attach the original signal S(t) to be transmitted to the simulated light source, run the Monte Carlo model using the obtained optimal number of photon packets N1, and obtain the simulation output of the receiver; reconstruct the signal expression as y(t) = S(t)⊗h(t), and compare and verify it with the simulation output.

Images