Single-photon laser radar data generation method and system for power transmission line in icing state

Abstract

The invention relates to a single-photon laser radar data generation method and system for a power transmission line in an icing state. The method comprises the following steps: constructing geometric information of the power transmission line and different icing states thereof; based on a Monte Carlo sampling method, simulating a transmission process of photons reaching an ice-coated power transmission line in a fog-free scene, and obtaining single photon laser radar data in the fog-free scene; on the basis of a Monte Carlo sampling method and a Mie scattering theory, simulating a propagation process of photons to a power transmission line in an icing state in a fog scene, and obtaining single photon laser radar data in the fog scene; and the single-photon laser radar data in the fog-free scene and the single-photon laser radar data in the fog scene are encoded and then are linearly superposed to generate single-photon laser radar photon data under different fog concentration conditions. According to the invention, the single-photon laser radar data of the power transmission line in the icing state can be effectively generated.

Description

Technical Field

[0001] This invention relates to the field of data generation technology, and in particular to a method for generating single-photon lidar data for iced power transmission lines. Background Technology

[0002] Single-photon lidar, as a novel active detection method, has high temporal resolution and high sensitivity. It can obtain target depth information by measuring the flight time of a single photon, providing technical support for the identification and assessment of icing conditions on power transmission lines.

[0003] However, the availability of raw data from single-photon lidar on power transmission lines under icing conditions is extremely limited, which significantly restricts the research and optimization of related algorithms. Furthermore, existing single-photon lidar data generation methods rarely consider the photon propagation characteristics in foggy environments and the impact of fog backscattered photons on the actual detected photons, resulting in data lacking high fidelity and physical consistency.

[0004] Existing single-photon lidar data generation methods are typically only applicable to general scenarios and cannot meet the requirements of data authenticity, physical consistency, and adaptability to complex scenarios in power transmission line icing detection. Therefore, to improve the rationality, accuracy, and practicality of detection algorithms in different scenarios, it is urgent to design a high-quality, high-fidelity data generation method based on single-photon lidar active imaging technology for different scenarios. This would enhance the reliability and applicability of the detection algorithm in various scenarios and provide a data foundation for subsequent identification of power transmission line icing conditions. Summary of the Invention

[0005] Therefore, the technical problem to be solved by the present invention is to overcome the fact that the existing single-photon lidar data generation methods are usually only applicable to general scenarios and are difficult to meet the requirements of data authenticity, physical consistency and adaptability to complex scenarios in power transmission line icing detection.

[0006] To address the aforementioned technical problems, this invention provides a method for generating single-photon lidar data for iced power transmission lines, comprising:

[0007] Step S1: Construct the geometric information of the transmission line and its different icing states;

[0008] Step S2: Based on the Monte Carlo sampling method, simulate the propagation process of photons reaching the icy power line in a fog-free scene to obtain single-photon lidar data in a fog-free scene;

[0009] Step S3: Based on the Monte Carlo sampling method and the Mie scattering theory, simulate the propagation process of photons reaching the icy power line in a fog scene to obtain single-photon lidar data in the fog scene;

[0010] Step S4: Encode the single-photon lidar data in the fog-free scene and the single-photon lidar data in the fog scene, and then linearly superimpose them to generate single-photon lidar photon data under different fog concentration conditions.

[0011] In one embodiment of the present invention, the different icing states in step S1 include uniform concentric icing, asymmetric eccentric icing, and rough irregular icing, wherein,

[0012] The uniform concentric circular icing: the icing thickness is fixed, and the radius of the transmission line after icing is... ,in, The radius of the transmission line. The thickness of the ice layer covering the power transmission line;

[0013] The asymmetric eccentric icing: the icing thickness changes with the azimuth angle of the transmission line cross-section, and is modeled using an angle function, expressed as: ,in, For average ice thickness, Indicates the direction of the incoming wind. Indicates the degree of asymmetry. The azimuth angle of the cross-section of the transmission line;

[0014] The rough and irregular icing: The surface roughness of the icing is simulated by superimposing a high-frequency perturbation term on the reference icing radius, expressed as: ,in, Represents the noise function. For the disturbance amplitude, For scale parameters, The azimuth angle of the cross-section of the transmission line.

[0015] In one embodiment of the present invention, step S2, based on the Monte Carlo sampling method, simulates the propagation and return process of photons reaching an icy power line in a fog-free scene. The method for obtaining single-photon lidar data in a fog-free scene includes:

[0016] First, the path of freedom of a photon in the current fog-free environment is calculated, where the path of freedom is the distance the photon travels along the current direction. Then, within the range of the current travel distance, it is determined whether the photon will hit the power line. If it hits the power line, the photon's direction of motion is updated according to the law of specular reflection, and the current travel distance is recorded. If it does not hit the power line, the photon is considered lost. Finally, for photons that hit the power line, a new path of freedom is calculated, and the photon moves along the updated direction of motion. It is then determined whether the photon is received by the detector. If it is received by the detector, the current photon is recorded as a valid photon event; otherwise, it is an invalid photon event, and the motion of the next photon is simulated. Finally, all valid photon events received by the detector constitute the data generated by the single-photon lidar in the fog-free scene.

[0017] In one embodiment of the present invention, step S3, which constructs a physical modeling method for the propagation process of photons based on Mie scattering theory, includes:

[0018] Based on Mie scattering theory, a physical model is constructed to represent the backscattering during photon propagation, the multiple collisions between photons and medium particles in foggy scenes, and energy attenuation, as follows:

[0019] ;

[0020] in, The scattering efficiency factor. This is the absorption efficiency factor. Extinction efficiency factor For scattering cross section, For absorption cross section, For extinction cross section, For droplet size factor, The diameter of the droplets is [missing information]. The wavelength of the laser. and The Mie scattering coefficient is... Let the order of the series expansion be . This indicates taking the real part of a complex number.

[0021] In one embodiment of the present invention, after each collision between a photon and a medium particle in a foggy scene, the azimuth angle and scattering angle of the photon are remodeled, specifically as follows:

[0022] By the azimuth angle of the photon in Uniform sampling within the range enables modeling of the photon azimuth angle after each collision;

[0023] The following formula guarantees the photon scattering angle. The statistical distribution conforms to physical laws, enabling modeling of the photon scattering angle after each collision:

[0024] ;

[0025] in, Here is a parameter used to control the directionality of scattering, and its value range is... , This is the attenuation coefficient.

[0026] In one embodiment of the present invention, when step S2 acquires single-photon lidar data in a fog-free scene and step S3 acquires single-photon lidar data in a foggy scene, the method further includes simulating multi-surface interaction between photons and the adhesive layer of the transmission line and its outer icing layer to generate multi-surface echo signals that conform to actual physical characteristics, as shown below:

[0027] ;

[0028] ;

[0029] ;

[0030] in, For the total echo photon information, For noisy photon information, This represents a multi-peak echo signal formed on the time axis. This represents the number of surfaces on which photons can interact. For photons and the first The number of photons returned by each surface interaction. This indicates the photon travels from the laser source to the... The propagation distance of each interactive surface It is the speed of light.

[0031] In one embodiment of the present invention, the method for generating single-photon lidar photon data under different fog concentration conditions by encoding single-photon lidar data in a fog-free scene and single-photon lidar data in a fog scene and then linearly superimposing them in step S4 includes:

[0032] Single-photon lidar data in a fog-free scene is processed using a first variational autoencoder. Encode the data and output the mean parameters of the Gaussian distribution. and variance parameter And based on the reparameterization sampling method of the standard Gaussian distribution, the corresponding latent variables are generated. ;

[0033] Single-photon lidar data in foggy scenes is processed using a second variational autoencoder. Encode and output gamma distribution parameters and The corresponding latent variables are generated through the reparameterization sampling method of the gamma distribution. ;

[0034] Generate hidden variables and latent variables Using linear superposition as the input to the decoder, the decoded output is single-photon lidar data in foggy scenes. .

[0035] To address the aforementioned technical problems, this invention provides a single-photon lidar data generation system for iced power transmission lines, comprising:

[0036] Building blocks: Used to construct the geometric information of transmission lines and their different icing conditions;

[0037] First data acquisition module: used to simulate the propagation process of photons reaching icy power lines in a fog-free scene based on the Monte Carlo sampling method, and to acquire single-photon lidar data in a fog-free scene;

[0038] The second data acquisition module is used to simulate the propagation process of photons reaching icy power lines in fog scenes based on the Monte Carlo sampling method and Mie scattering theory, and to acquire single-photon lidar data in fog scenes.

[0039] Data generation module: This module encodes and linearly superimposes single-photon lidar data from fog-free and foggy scenarios to generate single-photon lidar photon data under different fog concentration conditions.

[0040] To solve the above-mentioned technical problems, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the single-photon lidar data generation method for iced power transmission lines as described above.

[0041] To address the aforementioned technical problems, the present invention provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the single-photon lidar data generation method for iced power transmission lines as described above.

[0042] Compared with the prior art, the above-described technical solution of the present invention has the following advantages:

[0043] The single-photon lidar data generation method for iced power transmission lines described in this invention combines parameterized geometric modeling and Monte Carlo photon propagation simulation to generate high-fidelity single-photon lidar photon data in both fog-free and foggy scenarios. Furthermore, considering the high time cost of pixel-by-pixel simulation, this invention also proposes a single-photon lidar data generation method for foggy scenarios based on a dual-encoder VAE architecture. This not only overcomes the bottleneck of low computational efficiency in traditional pixel-by-pixel simulation but also strictly adheres to the physical distribution laws of photon propagation during the generation process. Finally, by recording the time it takes for photons to return to the detector after interacting with multiple surfaces and superimposing this data on the time axis, multi-peak echo single-photon lidar data is generated, enabling single-photon lidar data generation in multi-surface scenarios. Attached Figure Description

[0044] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0045] Figure 1 This is a flowchart of the method of the present invention;

[0046] Figure 2 This is a schematic diagram of the geometric modeling results of the transmission line under the uniform concentric circle icing state in an embodiment of the present invention;

[0047] Figure 3 This is a histogram of photon distribution in the single-photon lidar data generated in a fog-free scene according to an embodiment of the present invention;

[0048] Figure 4 This is a histogram of photon distribution in the single-photon lidar data generated in a foggy scene according to an embodiment of the present invention;

[0049] Figure 5 This is a schematic diagram of the single-photon lidar data generation method in a fog scene based on a dual-encoder VAE in an embodiment of the present invention. Detailed Implementation

[0050] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0051] Example 1

[0052] Reference Figure 1 As shown, this invention relates to a method for generating single-photon lidar data for transmission lines under icing conditions, comprising:

[0053] Step S1: Construct the geometric information of the transmission line and its different icing states;

[0054] Step S2: Based on the Monte Carlo sampling method, simulate the propagation process of photons reaching the icy power line in a fog-free scene to obtain single-photon lidar data in a fog-free scene;

[0055] Step S3: Based on the Monte Carlo sampling method and the Mie scattering theory, simulate the propagation process of photons reaching the icy power line in a fog scene to obtain single-photon lidar data in the fog scene;

[0056] Step S4: Encode the single-photon lidar data in the fog-free scene and the single-photon lidar data in the fog scene, and then linearly superimpose them to generate single-photon lidar photon data under different fog concentration conditions.

[0057] The following is a detailed description of this embodiment:

[0058] Existing single-photon lidar data generation methods are typically limited to general scenarios and cannot meet the requirements of data authenticity, physical consistency, and adaptability to complex scenarios in power transmission line icing detection. This invention proposes a data generation method based on single-photon lidar active imaging technology for power transmission line icing detection in different scenarios. By combining parameterizable geometric modeling and Monte Carlo photon propagation simulation, high-fidelity single-photon lidar photon data can be generated in both fog-free and foggy scenarios. Furthermore, considering the high time cost of pixel-by-pixel simulation, this invention also proposes a single-photon lidar data generation method in foggy scenarios based on a dual-encoder VAE architecture. This not only overcomes the bottleneck of low computational efficiency in traditional pixel-by-pixel simulation but also strictly adheres to the physical distribution laws of photon propagation during the generation process. Finally, by recording the time it takes for photons to return to the detector after interacting with multiple surfaces and superimposing this data on the time axis, multi-peak echo single-photon lidar data is generated, enabling single-photon lidar data generation in multi-surface scenarios.

[0059] (1) Parametric geometric model of transmission line

[0060] Existing geometric modeling methods for power transmission lines typically only consider single geometric shapes or simple icing conditions, making it difficult to meet the simulation requirements of high-fidelity single-photon lidar data generation for complex icing states. This invention proposes a parameterizable geometric modeling method for power transmission line icing detection. By combining chain-line modeling with descriptions of multiple icing states, it achieves accurate characterization of the geometric shape of power transmission lines under different natural environments.

[0061] Specifically: First, a parameterizable geometric model of the transmission line is established, taking into account icing conditions such as... Figure 2 As shown, taking a uniform concentric circle ice-covered state as an example, Figure 2The innermost layer of dots represents the main structure of the transmission line, while the area between the outer and innermost dots constitutes a cylindrical ice layer covering the transmission line. The spatial morphology of the transmission line is modeled using a chain-like structure, with the span direction denoted as [missing information]. The axis, perpendicular to the direction, is denoted as axis.

[0062]

[0063] in, The span position is At that time, the vertical height of the transmission line, It is a hyperbolic cosine function. , For horizontal tension, For linear density, It is the acceleration due to gravity. and The displacement constant is given. The catenary model can be solved based on the span length, boundary height, and tension. And the offset, and obtain a series of center points through discrete sampling. The obtained center point is used to determine the axial position of the transmission line in three-dimensional space and serves as the reference skeleton for subsequent icing geometry modeling. Transmission lines are typically modeled as cylinders with a radius set to... However, under natural environmental conditions, the icing state of transmission lines can vary. This embodiment considers three categories in its geometric modeling to address the different icing conditions that may occur on transmission lines.

[0064] The first type is the uniform concentric circle icing state. This state is mainly applicable to scenarios with low wind speeds or relatively uniform environmental conditions, where the icing is uniformly deposited on the surface of the transmission line. In specific modeling, it is assumed that the outer covering thickness of the transmission line is... The outer radius of the ice layer after it is covered with ice is: ,in, The radius of the transmission line. This model represents the thickness of the ice layer covering the transmission line. It simplifies calculations and allows for the rapid acquisition of the geometric parameters of the transmission line under uniform icing conditions.

[0065] The second type is asymmetric eccentric icing. This condition often occurs in natural environments with strong winds or winds of unidirectional direction, where the icing is thicker on the windward side and thinner on the leeward side. To accurately characterize its distribution characteristics, the icing thickness of the transmission line under the current conditions is modeled using an angle function: ,in, For average ice thickness, Indicates the direction of the incoming wind. Indicates the degree of asymmetry. The azimuth angle of the cross-section of the transmission line.

[0066] The third type is the rough and irregular icing state. This state often occurs in environments with frequent temperature changes and alternating icing and melting, resulting in irregular roughness on the surface of the transmission line. Therefore, high-frequency perturbation terms are superimposed on the reference shape to simulate the roughness of the iced surface: .in, Represents the noise function. For the disturbance amplitude, For scale parameters, This represents the azimuth angle of the transmission line's cross-section. This allows for the creation of irregular icing patterns, thereby improving the model's ability to acquire training data in complex environments.

[0067] Through the above-mentioned parameterizable geometric modeling method, this invention can accurately depict the spatial morphology and various icing states of power transmission lines under different natural environments. This not only improves the physical realism of power transmission line icing data generation, but also provides a reliable geometric basis for single-photon lidar simulation, and provides clear depth information for subsequent single-photon lidar data generation in fog-free and foggy scenarios.

[0068] (2) Data generation method for single-photon lidar in fog-free scenarios

[0069] In fog-free scenarios, this invention proposes a single-photon lidar data generation method for detecting icing on power transmission lines. By employing Monte Carlo sampling to simulate the photon propagation process in fog-free conditions, and combining the physical characteristics, temporal response, optical reflection laws, and atmospheric photon propagation models of single-photon lidar, photon data in active imaging mode under fog-free conditions is generated. Based on the depth information obtained from the constructed parameterizable geometric model of the power transmission line, clear target object location information is provided for the generation of active imaging data by single-photon lidar. The simulation process and principles will be explained in detail below.

[0070] First, the parameters required for single-photon lidar simulation are initialized. These parameters include laser pulse parameters, receiver parameters, atmospheric environment parameters, and target object parameters. The laser pulse parameters include: resolution (fps), wavelength, transport distance, time interval (bin), vertical theta, horizontal phi, and initial photon emission position. Receiver parameters include: detector aperture. Atmospheric environment parameters include: smoke visibility. Target object parameters include: reflectivity and surface roughness. After initializing all required parameters, the photon emission direction at each pixel location in each frame is determined based on the set vertical theta, horizontal phi, and corresponding resolution information.

[0071] Since photons experience energy attenuation during atmospheric propagation, the following steps are taken: First, the free path of the photon in the current fog-free environment is calculated, where the free path is the distance the photon travels along the current direction. Then, within the current travel distance, it is determined whether the photon will hit the target object (power line), i.e., an icy power line. If it hits the power line, the photon's direction of motion is updated according to the law of specular reflection, and the current travel distance is recorded. If it does not hit the power line, the photon is considered lost. Finally, for photons that hit the power line, a new free path is calculated, and the photon moves along the updated direction. It is then determined whether the photon is received by the detector. If it is received, the current photon is recorded as a valid photon event; otherwise, it is considered an invalid photon event, and the simulation of the next photon's motion begins. Ultimately, all valid photon events received by the detector are used as data generated by the single-photon lidar in a fog-free scene.

[0072] (3) Method for generating single-photon lidar data in foggy scenarios

[0073] The data generation methods for single-photon lidar in foggy and fog-free scenarios are similar, both based on Monte Carlo simulation of photon propagation in the atmosphere. The difference lies in the scattering effect introduced by the fog environment. Specifically, in foggy scenarios, the environment contains a large number of medium particles, and emitted photons are highly susceptible to collisions with these particles during propagation. Each collision not only causes photon energy attenuation but may also result in absorption or dissipation, while simultaneously altering the photon's propagation direction. Therefore, this invention, building upon the data generation method for fog-free scenarios, further introduces the atmospheric particle scattering effect. Based on Mie scattering theory, it comprehensively models photon backscattering, multiple collisions, and energy attenuation, thereby more realistically simulating the photon propagation and detection process in fog and obtaining high-fidelity generated data for single-photon lidar in foggy scenarios.

[0074] Specifically, it is known that the scattering characteristics of photons in fog satisfy the Mie scattering theory, and the absorption cross section can be accurately calculated based on this theory. scattering cross section Extinction section Scattering efficiency factor Absorption efficiency factor and extinction efficiency factor The calculation formula is as follows:

[0075]

[0076] In the formula, For droplet size factor, The diameter of the droplets is [missing information]. The wavelength of the laser. and The Mie scattering coefficient is... Let the order of the series expansion be . To represent the real part of a complex number, we can then determine the dissipation and absorption rates of photons propagating in fog. Furthermore, as fog concentration increases, the free path decreases, increasing the probability of photons colliding per unit distance. Therefore, in the simulation, each photon undergoes multiple collision calculations before being received by the detector. After each collision between a photon and a medium particle, the scattering angle of the photon is randomly sampled according to the Henyey–Greenstein phase function. To ensure that the statistical distribution of the scattering angle conforms to physical laws, the calculation formula is as follows:

[0077]

[0078] in, Controlling the directionality of scattering. When When the value approaches 1, scattering is mainly concentrated in front. As the angle approaches 0, it tends to backscatter more. And the azimuth angle of the photon... Then from Uniform sampling is performed within the range to update the photon's motion direction. Simultaneously, it determines whether the current photon has been absorbed or dissipated; if absorbed or dissipated, its subsequent trajectory is no longer monitored. Finally, by simulating the trajectory of each emitted photon at each pixel location, the photon information received by the detector constitutes the photon data for a single-photon lidar in a foggy scene. This invention comprehensively considers multiple scattering, absorption, and collisions during photon propagation in fog, ensuring that the generated single-photon lidar photon data conforms to physical laws and strictly follows the physical process of photon propagation, thereby effectively guaranteeing the authenticity and reliability of the simulation data.

[0079] (4) A method for generating single-photon lidar data in fog scenes based on a dual-encoder VAE architecture

[0080] Existing single-photon lidar data generation methods, whether based on Monte Carlo sampling or modeling the distribution characteristics of fog backscattered photons and target echo photons, all require pixel-by-pixel simulation. This pixel-by-pixel simulation method is not only computationally intensive, but its time cost also increases exponentially with scene complexity, severely limiting data generation efficiency. Currently, common data generation frameworks mainly include Variational Auto-Encoder (VAE), Generative Adversarial Network (GAN), and Diffusion Model. Among them, the Diffusion Model, through additative Gaussian noise and Markov chain training mechanisms, is suitable for modeling continuous Gaussian distributed data. GANs rely on the adversarial game between the generator and discriminator to improve the realism of generated samples, but due to the lack of direct modeling of physical constraints, they have certain limitations in simulating the impact of backscattered photons on target echo photon detection in fog scenes. Considering that backscattered photons in fog physically follow a gamma distribution while target echo photons follow a Gaussian distribution, it is difficult to accurately capture the differences in statistical properties between the two types of photons if only a general generation model is used.

[0081] To address the current challenges, this invention proposes a method for generating single-photon lidar data in foggy scenes based on a dual-encoder VAE architecture. The method utilizes dual encoders to encode single-photon lidar photon data in both fog-free and foggy scenes. Latent variables in the corresponding latent space are obtained through gamma reparameterization and Gaussian reparameterization techniques, respectively. These latent variables are then linearly fused and used as the input to the decoder. Finally, by minimizing the reconstruction error between the generated and labeled images, and the KL divergence between the encoder latent variable distribution and the prior distribution, a generative model suitable for single-photon lidar data in foggy scenes is constructed.

[0082] During training, the model uses real-world fog-prone single-photon lidar data as supervisory labels to calculate the reconstruction error between the decoder's output and the actual observation data. Simultaneously, to ensure the latent space retains its probabilistic generation characteristics, KL divergence constraints are applied to the latent variables of both the clean signal encoder and the fog noise encoder, making their distributions approximate a pre-defined prior distribution. This training strategy enables the decoder to learn the combined mapping relationship between fog-free photon signals and fog noise in the latent space, while maintaining the continuity and structure of the latent space. This ensures that the generated foggy single-photon lidar data not only conforms to physical laws but also possesses diversity and controllability. In the actual inference phase, only fog-free single-photon lidar data needs to be input. By randomly sampling from the latent variable distribution of the fog noise encoder, single-photon lidar data with different concentrations and random characteristics in fog-prone scenarios can be generated, thus meeting the data requirements for single-photon lidar in various fog concentration scenarios. The training objective function is:

[0083]

[0084] in, For loss function, Let the mean squared error loss function be . This refers to single-photon lidar data generated by the decoder in a foggy environment. The data is from observations in real foggy scenes. Kullback-Leibler divergence is used to constrain the latent space distribution to be consistent with the prior distribution. For clean signal encoders of single-photon lidar data in fog-free scenes Generate latent variables The approximate posterior distribution, To and These represent the prior probabilities of the latent variables in a clean environment and a foggy environment, respectively. For fog noise encoders, single-photon lidar data in foggy scenes Generate latent variables The approximate posterior distribution is obtained. The MSE term optimizes the decoder, making its generated foggy single-photon data approximate the actual observation data. The KL divergence term constrains the two encoders, ensuring the latent variable distribution maintains its probabilistic generation characteristics. Therefore, this loss function enables joint training of the two encoders and decoder, ensuring the generation process follows physical laws while possessing diversity and controllability.

[0085] This invention proposes a method for generating single-photon lidar data in foggy scenes based on a dual-encoder VAE architecture. This method not only breaks through the bottleneck of low efficiency in traditional pixel-by-pixel simulation calculations, but also strictly follows the physical distribution law of photon propagation during the generation process.

[0086] (5) Single-photon lidar data generation method in multi-surface scenarios

[0087] The multi-surface nature of power transmission lines stems from the ice covering their surfaces. Some laser beams can penetrate the ice to detect the power line surface, while others directly detect the ice covering the power line and then return. Specifically, existing single-photon lidar data generation methods often only consider the echo from a single target, failing to accurately reflect the multi-peak echo characteristics formed by photons interacting with multiple surfaces in complex scenarios. This invention proposes a single-photon lidar data generation method for multi-surface scenarios, simulating the multi-surface interaction between photons and the adhesive layer and the ice layer covering the power transmission line, thereby generating echo signals that more closely resemble actual physical characteristics. Specifically, for each emitted photon, this invention calculates the time it takes for the photon to return to the detector, i.e., the time-of-flight (ToF), calculated using the following formula:

[0088]

[0089] in, This indicates the photon travels from the emitter to the [missing information - likely a specific point or location]. The propagation distance of each interactive surface Let be the speed of light. When a photon interacts with multiple surfaces during its propagation, each interaction generates a corresponding return signal. Let be the distance between the photon and the first surface. The number of returned photons from each surface interaction is The multi-peak echo signal formed on the time axis can then be represented as:

[0090]

[0091] in, This represents the number of surfaces on which photons may interact. Each peak corresponds to a detection event generated by a photon interacting with a specific surface, and the peak shape characteristics in the photon distribution histogram can further reflect physical information such as surface roughness and reflectivity. The detector records the time it takes for photons to return to the detector after interacting with each surface, and these returning photon detection events are superimposed on the time axis to form a multi-peaked echo sequence. This sequence can more comprehensively reflect the physical laws of photon propagation and interaction in complex scenes. To enhance the realism of the simulated data, this invention further introduces the effects of detector noise, photon timing jitter, and environmental background noise during the generation process. The final multi-surface target echo signal is:

[0092]

[0093] in, For the total echo photon information, This represents information about noisy photons. Through the above design, this invention can not only simulate the multi-peak echo characteristics of photons in multi-surface scenarios, but also comprehensively consider actual detectors and environmental factors, thereby generating more physically realistic single-photon lidar echo signals, providing reliable data support for imaging and recognition in multi-surface scenarios.

[0094] The method for generating single-photon lidar data in fog-free scenarios based on Monte Carlo simulation is as follows:

[0095] First, the physical and environmental parameters of the single-photon lidar are initialized. The photon emission direction is determined based on the initialized pitch angle, scattering angle, and resolution. Next, the following simulation process is performed for each emitted photon: First, the photon's path of freedom in the current atmospheric environment is calculated, i.e., the distance the photon travels along the current direction. Then, within the current travel distance, it is determined whether the photon will hit a target. If it hits a target, the photon's motion direction is updated according to the specular reflection law, and the current travel distance is recorded. If it does not hit a target, the photon is considered lost. Finally, for photons that hit a target, a new path of freedom is calculated, and the photon moves along the updated motion direction to determine if it is received by the detector. If it is received, the current photon is recorded as a valid detection photon event; otherwise, the motion of the next photon is simulated. Finally, after simulating each emitted photon for each pixel according to the above process, the photon information received by the detector is the data generated by the single-photon lidar in a fog-free scene. Figure 3 To simulate and obtain the photon information distribution histogram of a single-photon lidar in a fog-free scene based on the depth information obtained from the current modeling.

[0096] The data generation method for single-photon lidar in fog scenes based on Monte Carlo simulation is as follows:

[0097] First, the physical and environmental parameters of the single-photon lidar are initialized. The photon emission direction is determined based on the initialized elevation angle, scattering angle, and resolution. Next, the following simulation process is performed for each emitted photon: First, the free path of the current photon is calculated, i.e., the distance the photon travels along the corresponding direction of motion. Then, it is determined whether the photon hits the target object. If it hits, the photon's direction of motion is updated according to the specular reflection law, and the total distance traveled is recorded (the photon's flight time can be obtained by dividing the total distance by the photon's speed, and a histogram can be constructed based on the photon's flight time). If it misses, it is determined whether the photon is received by the detector. If it is received, it is recorded as a valid photon event, and the total flight time of the photon is calculated based on the total distance traveled and the speed of light. If it is not received by the detector, the scattering angle is randomly sampled in the Henyey–Greenstein phase function, and the scattering angle is determined based on the speed of light. The azimuth angle is uniformly sampled within the range to update the photon's motion direction, and the total distance the photon has traveled is recorded. This simulation process is repeated. If the current photon has dissipated or been absorbed, it indicates that the current photon is lost, and there is no need to track its subsequent trajectory. Finally, after simulating each emitted photon for each pixel according to the above process, the photon information received by the detector is the data generated by the single-photon lidar in a foggy scene. Figure 4 To simulate and obtain the photon data distribution histogram of a single-photon lidar in a fog scene based on the depth information obtained from the current modeling.

[0098] The method for generating single-photon lidar data in fog scenes based on a dual-encoder VAE architecture is as follows:

[0099] Specifically, a fog chamber was constructed to collect multiple paired sets of data: fog-free data, pure fog noise data, and foggy data, for model training and testing. All data were in 3D histogram format. ,in Indicates pixel size, Indicates the number of frames. Indicates the number of channels. The dual-encoder VAE network architecture is as follows: Figure 5 As shown, two independent encoders were designed, one of which was used to process single-photon lidar data in fog-free scenes. Encode the data and output the mean parameters of the Gaussian distribution. and variance parameter And based on the reparameterization sampling method of the standard Gaussian distribution, the corresponding latent variables are generated. Another encoder is used to process single-photon lidar data in pure fog scenarios. Encode and output gamma distribution parameters and The corresponding latent variables are generated through the reparameterization sampling method of the gamma distribution. The generated latent variables are then normalized. The two latent variables are combined linearly and used as input to the decoder to output single-photon lidar data in foggy scenes. Then, in the generative inference stage, single-photon lidar data from fog-free scenarios is input into the model, and the corresponding latent variables are obtained through the corresponding encoder. Then, random sampling is performed from the latent variable distribution learned by the fog noise encoder to obtain the latent variables corresponding to the fog scene. .Will and After being superimposed in the hidden space, the data can be used as input to the decoder to generate single-photon lidar photon data that meets different fog concentration conditions and has fog backscattering characteristics.

[0100] Example 2

[0101] This embodiment provides a single-photon lidar data generation system for transmission lines under icing conditions, including:

[0102] Building blocks: Used to construct the geometric information of transmission lines and their different icing conditions;

[0103] Building blocks: Used to construct the geometric information of transmission lines and their different icing conditions;

[0104] First data acquisition module: used to simulate the propagation process of photons reaching icy power lines in a fog-free scene based on the Monte Carlo sampling method, and to acquire single-photon lidar data in a fog-free scene;

[0105] The second data acquisition module is used to simulate the propagation process of photons reaching icy power lines in fog scenes based on the Monte Carlo sampling method and Mie scattering theory, and to acquire single-photon lidar data in fog scenes.

[0106] Data generation module: This module encodes and linearly superimposes single-photon lidar data from fog-free and foggy scenarios to generate single-photon lidar photon data under different fog concentration conditions.

[0107] Example 3

[0108] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the single-photon lidar data generation method for iced power transmission lines described in Embodiment 1.

[0109] Example 4

[0110] This embodiment provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the steps of the single-photon lidar data generation method for iced power transmission lines described in Embodiment 1.

[0111] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0112] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0113] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0114] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0115] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for generating single-photon lidar data for iced power transmission lines, characterized in that, include: Step S1: Construct the geometric information of the transmission line and its different icing states; Step S2: Based on the Monte Carlo sampling method, simulate the propagation process of photons reaching the icy power line in a fog-free scene to obtain single-photon lidar data in a fog-free scene; Step S3: Based on the Monte Carlo sampling method and the Mie scattering theory, simulate the propagation process of photons reaching the icy power line in a fog scene to obtain single-photon lidar data in the fog scene; Step S4: Encode the single-photon lidar data in the fog-free scene and the single-photon lidar data in the fog scene, and then linearly superimpose them to generate single-photon lidar photon data under different fog concentration conditions.

2. The method for generating single-photon lidar data for iced power transmission lines according to claim 1, characterized in that: The different icing states in step S1 include uniform concentric icing, asymmetric eccentric icing, and rough, irregular icing. The uniform concentric circular icing: the icing thickness is fixed, and the radius of the transmission line after icing is... ,in, The radius of the transmission line. The thickness of the ice layer covering the power transmission line; The asymmetric eccentric icing: the icing thickness changes with the azimuth angle of the transmission line cross-section, and is modeled using an angle function, expressed as: ,in, For average ice thickness, Indicates the direction of the incoming wind. Indicates the degree of asymmetry. The azimuth angle of the cross-section of the transmission line; The rough and irregular icing: The surface roughness of the icing is simulated by superimposing a high-frequency perturbation term on the reference icing radius, expressed as: ,in, Represents the noise function. For the disturbance amplitude, For scale parameters, The azimuth angle of the cross-section of the transmission line.

3. The method for generating single-photon lidar data for iced power transmission lines according to claim 1, characterized in that: Step S2, based on the Monte Carlo sampling method, simulates the propagation and return process of photons to an icy power line in a fog-free scene. The method for obtaining single-photon lidar data in a fog-free scene includes: First, the path of freedom of a photon in the current fog-free environment is calculated, where the path of freedom is the distance the photon travels along the current direction. Then, within the range of the current travel distance, it is determined whether the photon will hit the power line. If it hits the power line, the photon's direction of motion is updated according to the law of specular reflection, and the current travel distance is recorded. If it does not hit the power line, the photon is considered lost. Finally, for photons that hit the power line, a new path of freedom is calculated, and the photon moves along the updated direction of motion. It is then determined whether the photon is received by the detector. If it is received by the detector, the current photon is recorded as a valid photon event; otherwise, it is an invalid photon event, and the motion of the next photon is simulated. Finally, all valid photon events received by the detector constitute the data generated by the single-photon lidar in the fog-free scene.

4. The method for generating single-photon lidar data for iced power transmission lines according to claim 1, characterized in that: Step S3, which constructs a physical modeling method for the photon propagation process based on Mie scattering theory, includes: Based on Mie scattering theory, a physical model is constructed to represent the backscattering during photon propagation, the multiple collisions between photons and medium particles in foggy scenes, and energy attenuation, as follows: ； in, The scattering efficiency factor. This is the absorption efficiency factor. Extinction efficiency factor For scattering cross section, For absorption cross section, For extinction cross section, For droplet size factor, The diameter of the droplets is [missing information]. The wavelength of the laser. and The Mie scattering coefficient is... Let the order of the series expansion be . This indicates taking the real part of a complex number.

5. The method for generating single-photon lidar data for iced power transmission lines according to claim 4, characterized in that: In foggy scenes, after each collision between a photon and medium particles, the azimuth and scattering angles of the photon are remodeled, specifically as follows: By the azimuth angle of the photon in Uniform sampling within the range enables modeling of the photon azimuth angle after each collision; The following formula guarantees the photon scattering angle. The statistical distribution conforms to physical laws, enabling modeling of the photon scattering angle after each collision: ； in, Here is a parameter used to control the directionality of scattering, and its value range is... , This is the attenuation coefficient.

6. The method for generating single-photon lidar data for iced power transmission lines according to claim 1, characterized in that: When acquiring single-photon lidar data in a fog-free scene in step S2 and acquiring single-photon lidar data in a foggy scene in step S3, the process also includes simulating multi-surface interactions between photons and the adhesive layer of the transmission line and its outer icing layer to generate multi-surface echo signals that conform to actual physical characteristics, as shown below: ； ； ； in, For the total echo photon information, For noisy photon information, This represents a multi-peak echo signal formed on the time axis. This represents the number of surfaces on which photons can interact. For photons and the first The number of photons returned by each surface interaction. This indicates the photon travels from the laser source to the... The propagation distance of each interactive surface It is the speed of light.

7. The method for generating single-photon lidar data for iced power transmission lines according to claim 1, characterized in that: The method for generating single-photon lidar photon data under different fog concentration conditions by encoding and then linearly superimposing single-photon lidar data in fog-free and fog-filled scenarios in step S4 includes: Single-photon lidar data in a fog-free scene is processed using a first variational autoencoder. Encode the data and output the mean parameters of the Gaussian distribution. and variance parameter And based on the reparameterization sampling method of the standard Gaussian distribution, the corresponding latent variables are generated. ; Single-photon lidar data in foggy scenes is processed using a second variational autoencoder. Encode and output gamma distribution parameters and The corresponding latent variables are generated through the reparameterization sampling method of the gamma distribution. ; Generate hidden variables and latent variables Using linear superposition as the input to the decoder, the decoded output is single-photon lidar data in foggy scenes. .

8. A single-photon lidar data generation system for iced power transmission lines, characterized in that, include: Building blocks: Used to construct the geometric information of transmission lines and their different icing conditions; First data acquisition module: used to simulate the propagation process of photons reaching icy power lines in a fog-free scene based on the Monte Carlo sampling method, and to acquire single-photon lidar data in a fog-free scene; The second data acquisition module is used to simulate the propagation process of photons reaching icy power lines in fog scenes based on the Monte Carlo sampling method and Mie scattering theory, and to acquire single-photon lidar data in fog scenes. Data generation module: This module encodes and linearly superimposes single-photon lidar data from fog-free and foggy scenarios to generate single-photon lidar photon data under different fog concentration conditions.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the steps of the single-photon lidar data generation method for iced power transmission lines as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the single-photon lidar data generation method for iced power transmission lines as described in any one of claims 1 to 7.

Images