Adaptive optical remote control interaction method for quantum communication

Abstract

The invention discloses a self-adaptive optical remote control interaction method for quantum communication, and relates to the technical field of quantum communication, and the method comprises the following steps: S1, system initialization and video stream generation; the CPU and the FPGA are integrated on one board card, the FPGA has lower control delay and power consumption compared with a GPU when processing calculation with high repeatability, such as Hartmann calculation and centroid calculation, so that the correction frequency of the adaptive optical system is improved, the stability and integrity of light beams in quantum communication are enhanced, and meanwhile, the accuracy of the correction frequency of the adaptive optical system is improved. The CPU + FPGA hardware architecture retains the advantages of a CPU von's structure, introduces the characteristics of no instruction, low delay and reprogramming of the FPGA, overcomes the problem of complex instruction stream control logic of the CPU and the GPU, finally achieves the purposes of improving the correction frequency of light beams and reducing correction delay in quantum communication, reduces the power consumption of equipment, and improves the reliability of the equipment. And the reliability and efficiency of the quantum communication system in practical application are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of quantum communication, and in particular to an adaptive optical remote control interaction method for quantum communication. Background Art

[0002] In the field of modern communications, quantum communication has attracted much attention for its ultra-high security and has become an important direction for the development of communication technology. The core of quantum communication lies in the transmission of quantum states, and the integrity and stability of light beams as carriers of quantum states play a decisive role in the quality and reliability of quantum communication. In the process of quantum communication, light beams need to propagate in the atmospheric environment. However, the atmosphere is not a uniform and stable medium. Factors such as atmospheric turbulence can cause the wavefront of light waves to be distorted, that is, wavefront distortion. This wavefront distortion can cause the light beam to have phase changes, intensity fluctuations and other problems, which seriously affect the accurate transmission of quantum states, and thus reduce the performance of quantum communication. Adaptive optical correction systems have emerged to solve the problem of distortion of light wavefronts by atmospheric turbulence. Its working principle is mainly to first detect the distortion of the wavefront, and then use deformable mirrors to correct the wavefront in real time.

[0003] In existing adaptive optics systems, CPUs and GPUs are often used to implement adaptive algorithms to control AO correction. For example, a hardware architecture of an adaptive optics real-time controller based on a multi-core CPU and GPU uses a multi-core CPU and GPU to complete the slope calculation, wavefront restoration, and wavefront control of the sub-aperture of the Shack-Hartmann wavefront sensor in parallel. Specifically, using a general multi-core CPU and GPU as the computing platform, a CPU core transfers part of the sub-aperture image data to the GPU through the PCIE bus, and the GPU completes the wavefront calculation task of this part of the sub-aperture, while the remaining CPU cores complete the remaining sub-aperture wavefront calculation tasks in parallel.

[0004] However, there are many problems with the above scheme. Since the CPU and GPU belong to the von Neumann structure, the instruction decoding and execution share memory, and the execution unit can execute any instruction, which makes the instruction flow control logic extremely complex. The complex instruction flow control makes it difficult to reduce the control delay and cannot meet the demand for rapid beam correction in quantum communication. The GPU needs to read the off-chip DRAM frequently during operation, which not only increases the delay of data transmission, but also leads to higher power consumption, which is not conducive to the long-term stable operation of quantum communication equipment. Therefore, it is necessary to propose an adaptive optical remote control interaction method for quantum communication to solve the problems in the prior art. Summary of the invention

[0005] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide an adaptive optical remote control interaction method for quantum communication, which can improve the integrity and stability of the light beam during quantum communication, reduce control delay, and reduce device power consumption by optimizing the hardware architecture and algorithm flow, thereby improving the overall performance of quantum communication.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: an adaptive optical remote control interaction method for quantum communication, the method comprising the following steps:

[0007] S1. System initialization and video stream generation: After the device is turned on, the CPU and FPGA integrated on a board are initialized. The FPGA obtains image data from the light source detection camera, performs cache processing and generates a video stream, which is transmitted to the Linux system of the CPU. The initialization of the FPGA includes configuring the internal cache size and data transmission format according to the accuracy requirements of quantum communication;

[0008] S2, light source detection and adaptive optical correction start: The Linux system starts the light source detection program after receiving the video stream. When the light source is detected, the FPGA performs the following operations:

[0009] S201, Hartmann calculation: Based on the data collected by the Hartmann wavefront sensor, the sensor divides the beam wavefront into multiple sub-apertures. For each sub-aperture, the slope is determined by calculating the offset of the center of mass of the light spot relative to the reference position. The coordinates of the center of mass of the light spot in the two-dimensional plane are defined as (x c ,y c ), the reference position coordinate is (x r ,y r ), the subaperture size parameter is d, then the slope S x =(x c -x r ) / d,S y =(y c -y r ) / d, during the calculation process, the spot signal is weak and the noise interference is large, so an adaptive weighted filtering algorithm is used to assign different weights according to the distance between the pixel point and the center of the spot for filtering;

[0010] S202, centroid calculation: Based on the light spot calculated by Hartmann, the centroid coordinates (x c ,y c ), the calculation method is Where (x i ,y i ) is the coordinate of the i-th pixel in the spot, p iis the intensity value of the pixel, n is the total number of pixels in the spot, the spot shape is irregular and partially blocked, and the morphological optimization algorithm based on deep learning is used to train the neural network to identify the effective area of ​​the spot and optimize the shape before calculating the centroid coordinates;

[0011] S203, control matrix generation and transmission: Based on the slope calculated by Hartmann and the center of mass coordinates calculated by the center of mass, the control matrix of the deformable mirror and the tilt mirror is generated by a multi-parameter fitting algorithm, and the slope deviation is defined as ΔS x , ΔS y , the center of mass coordinate deviation is Δx c , Δy c , control matrix element M ij =f(ΔS x ,ΔS y ,Δx c ,Δy c ), where f is a complex function relationship. After the control matrix is ​​generated, it is transmitted to the Linux system and the beam correction stage through the PCIE bus. During the transmission process, an error correction coding algorithm is used to ensure data accuracy;

[0012] S3, beam correction: After receiving the control matrix in the beam correction stage, the deformable mirror and the tilt mirror are controlled. For the deformable mirror, a voltage signal is applied through a high-voltage amplifier according to the element value of the control matrix to deform the mirror surface. The relationship between the deformation degree and the voltage value is determined according to the physical model. For the tilt mirror, the tilt angle is adjusted through the motor drive system according to the angle information in the control matrix. The motor drive system adopts an intelligent control algorithm.

[0013] S4. Continuous repair of quantum communication: In the process of quantum communication from the sending end to the receiving end, steps S2 and S3 are continuously repeated. If abnormalities occur in the energy loss of the light beam and the fidelity of the quantum state, the frequency of the Hartmann calculation and the center of mass calculation are adjusted according to the monitoring and evaluation algorithm, and the light source and optical components are checked. At the same time, the parameters of the adaptive optical system are dynamically adjusted according to the environmental adaptation algorithm.

[0014] Furthermore, in step S1, the initialization process of the CPU includes loading a Linux operating system specially optimized for quantum communication. The kernel of the operating system is customized and has the ability to efficiently schedule quantum communication data processing. Its scheduling strategy dynamically allocates computing resources based on quantum state priority and data real-time requirements, and automatically detects the communication interface status with FPGA and other hardware devices at startup. If a communication interface abnormality is detected, an interface repair algorithm is used for automatic repair. The algorithm locates the fault point by sending a specific detection signal sequence and analyzing the feedback signal. Specifically, the feedback signal is spectrally analyzed, the characteristic frequency is extracted and compared with a preset fault frequency library, and then targeted repair operations are performed according to the fault type. At the same time, the repair process and result information are recorded, and the repair record is stored in a specific non-volatile storage area, and the storage capacity is managed in a cyclic overwrite manner.

[0015] Furthermore, in step S201, in the Hartmann calculation, the adaptive weighted filtering algorithm is specifically as follows: the distance between the pixel point (x, y) and the center of the light spot (x0, y0) is defined as Assign weight w(r)=1 / (1+αr β +γr δ ), where α, β, γ and δ are parameters determined according to the noise characteristics of the quantum communication environment, and the calculation formula of the filtered pixel value p′(x, y) is:

[0016] N(x,y) is a set of neighborhood pixels centered at (x,y), and the size of the neighborhood is dynamically adjusted according to the spot size and noise distribution. This algorithm effectively removes noise and retains the spot detail information based on the characteristics of spot noise distribution in quantum communication. At the same time, the filtering effect is monitored in real time during the execution of the algorithm. If the centroid offset of the spot after filtering exceeds the preset threshold, the weight parameters are automatically adjusted to optimize the filtering performance.

[0017] Furthermore, in step S202, centroid calculation, the deep learning-based morphological optimization algorithm includes the following steps:

[0018] Construct a deep learning model, which takes the spot image as input and outputs the optimized spot shape. The model structure includes convolution layer, pooling layer and fully connected layer. The kernel function of the convolution layer adopts the function form optimized for the edge features of the spot, the pooling layer adopts an adaptive step size strategy, and the fully connected layer adopts a neuron activation function.

[0019] A large number of spot images with different shapes and occlusion conditions are used for training, and the loss function is used in the training process. Where S i is the original spot shape, S′ i is the optimized spot shape, W jis the model weight, λ and μ are regularization parameters, C l is the statistical characteristics of the feature map of the middle layer of the model;

[0020] After the training is completed, the actual collected spot image is input into the model to obtain the optimized spot shape, and then the center of mass coordinates are calculated. This algorithm adapts to the complex changes in the spot shape, and continuously collects new spot samples for incremental learning during the use of the model to continuously improve the model performance.

[0021] Furthermore, in step S203, control matrix generation and transmission, the multi-parameter fitting algorithm is specifically: define where a1-a n and b1-b m 、p k ,q k 、r k 、s k In order to determine the parameters based on the optical characteristics and correction requirements of the quantum communication system, the algorithm is optimized through a large amount of experimental data and simulation calculations. The algorithm comprehensively considers various combinations of slope and center of mass coordinate deviation to generate a more accurate control matrix. At the same time, an adaptive parameter adjustment mechanism is introduced into the algorithm to dynamically adjust some parameters to adapt to different atmospheric disturbances based on the real-time monitoring results of the beam quality during quantum communication.

[0022] Furthermore, in step S203, in the generation and transmission of the control matrix, the error correction coding check algorithm includes: grouping and coding the control matrix, adding a check bit to each group of data, the value of the check bit is generated according to the data bit and the coding rule, and the data bit is defined as D i , the check digit is P j , P j =g(D1,D2,…,D n ), where g is a function relationship. The check bit is recalculated at the receiving end based on the received data and compared with the transmitted check bit. If there is any inconsistency, an error location and correction algorithm is used. This algorithm determines the error location and corrects it by analyzing the logical relationship between the check bit difference and the data bit. Specifically, an error location discrimination matrix is ​​constructed, and the matrix element value is calculated according to the check bit comparison result to determine the error location. In addition, the data is encrypted during the transmission process. The encryption algorithm uses an encryption method based on quantum key distribution.

[0023] Furthermore, in step S3, in the deformable mirror control of beam correction, the physical model is: define the control matrix element value as M ij , the high voltage amplifier output voltage is V out , the deformation displacement of the deformable mirror surface is d,

[0024]

[0025] where k1-k 13 The parameters are determined according to the material characteristics and optical properties of the deformable mirror, and are obtained through experimental measurement and theoretical calculation. This model accurately describes the relationship between the control matrix element value and the deformation of the deformable mirror surface, and monitors the deformation of the mirror surface in real time during the operation of the deformable mirror, and fine-tunes the physical model parameters based on the feedback information.

[0026] Furthermore, in step S3, in the control of the tilt mirror for beam correction, the intelligent control algorithm includes: the motor drive system calculates the number of motor rotation steps according to the angle information in the control matrix, defines the target angle of the tilt mirror as θ, the number of motor rotation steps as N, N=h(θ), wherein h is a piecewise function relationship, and different calculation coefficients are used according to different angle ranges. During the rotation of the motor, the actual rotation angle of the motor is monitored in real time. If the deviation from the target angle exceeds a preset threshold, a dynamic adjustment strategy is adopted to adjust the motor drive voltage and pulse frequency according to the size and direction of the deviation. The adjustment method is to adopt a fuzzy logic control algorithm to determine the voltage and frequency adjustment amount according to the deviation and the deviation change rate. At the same time, the motor operating status is monitored in real time, including temperature and current parameters. If an abnormality is detected, the motor drive parameters are adjusted in time and an alarm is issued.

[0027] Furthermore, in step S4, in the abnormal processing of quantum communication continuous repair, the monitoring and evaluation algorithm is: define the beam energy loss rate as η, the quantum state fidelity as F, and calculate the evaluation value Among them, α, β, γ and δ are parameters determined according to the characteristics of the quantum communication system. Whether the frequency of Hartmann calculation and center of mass calculation needs to be adjusted is determined according to the size of the evaluation value. If E exceeds the preset threshold, the calculation frequency is increased. At the same time, the environmental adaptation algorithm adjusts the correction parameters of the adaptive optical system in real time according to the atmospheric turbulence intensity, temperature and humidity environmental parameters. The adjustment rules are based on the mapping relationship established by a large amount of experimental data and machine learning models, and detailed exception information and processing procedures are recorded during the exception handling process.

[0028] Compared with the prior art, the adaptive optical remote control interaction method for quantum communication has the following beneficial effects:

[0029] The present invention integrates a CPU and an FPGA on a single board. When processing highly repetitive calculations, such as Hartmann calculations and centroid calculations, the FPGA has lower control delay and power consumption than a GPU, thereby increasing the correction frequency of the adaptive optical system, thereby enhancing the stability and integrity of the light beam in quantum communication. At the same time, the CPU+FPGA hardware architecture retains the advantages of the CPU Feng structure, and introduces the characteristics of FPGA without instructions, low latency, and reprogramming, thereby overcoming the problem of complex control logic of the CPU and GPU instruction flow, ultimately achieving the purpose of increasing the correction frequency of the light beam in quantum communication and reducing the correction delay, and reducing the power consumption of the device, thereby improving the reliability and efficiency of the quantum communication system in practical applications.

[0030] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0032] Figure 1 is a flow chart of an adaptive optical remote control interaction method for quantum communication;

[0033] Figure 2 A timing diagram of an adaptive optical remote-controlled interaction method for quantum communication. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] Embodiment 1

[0036] In an actual quantum communication scenario, a system including a quantum communication transmitter, a receiver, and an intermediate transmission link is constructed, and an adaptive optical correction device based on the present invention is deployed in the transmission link.

[0037] After the device is turned on, the system is initialized first. The CPU and FPGA integrated on a board start their respective initialization processes. The FPGA completes the configuration of the internal cache size and the setting of the data transmission format based on the requirements of the quantum communication system for beam processing accuracy, and prepares to receive image data from the light source detection camera. The CPU loads the Linux operating system specially optimized for quantum communication. Its kernel has efficient scheduling capabilities for quantum communication data processing, and can dynamically allocate computing resources according to quantum state priority and data real-time requirements. At the same time, the CPU automatically detects the communication interface status with the FPGA and other hardware devices (such as high-voltage amplifiers, deformable mirrors, tilt mirrors, etc.) when it starts.

[0038] Light source detection and adaptive optical correction are started. After completing the image cache, the FPGA generates a video stream and transmits it to the Linux system of the CPU. The Linux system starts the light source detection program and uses the light source detection camera to collect light source information.

[0039] When a light source is detected, the FPGA begins performing critical adaptive optics correction calculations.

[0040] Based on the data collected by the Hartmann wavefront sensor, the sensor divides the beam wavefront into multiple sub-apertures. For each sub-aperture, the slope is determined by calculating the offset of the center of mass of the light spot relative to the reference position. The coordinates of the center of mass of the light spot in the two-dimensional plane are set as (x c ,y c ), the reference position coordinate is (x r ,y r ), the subaperture size parameter is d, then the slope S x =(x c -x r ) / dS y =(y c -y r ) / d.

[0041] During the calculation process, if the spot signal is weak or the noise interference is large, the newly created adaptive weighted filtering algorithm is used, and the distance between the pixel point (x, y) and the spot center (x0, y0) is set to r = √(x-x0) 2 +(y-y0) 2 , assign weight w(r)=1 / (1+αr β +γr δ )(where α, β, γ and δ are parameters determined according to the noise characteristics of the quantum communication environment), the pixel value after filtering

[0042] N(x,y) is a set of neighborhood pixels centered at (x,y), and the neighborhood size is dynamically adjusted according to the spot size and noise distribution to improve the accuracy of slope calculation.

[0043] Based on the light spot calculated by Hartmann, the centroid coordinates (x c ,y c ), the calculation method is (where (x i ,y i ) is the coordinate p of the i-th pixel in the spot i is the intensity value of the pixel, n is the total number of pixels in the spot), if the spot shape is irregular or partially blocked, a morphological optimization algorithm based on deep learning is used.

[0044] First, a deep learning model is constructed. The model takes the spot image as input and outputs the optimized spot shape. The model structure includes convolution layer, pooling layer and fully connected layer. The kernel function of the convolution layer adopts the function form optimized for the edge features of the spot. The pooling layer adopts an adaptive step size strategy. The fully connected layer adopts a newly created neuron activation function. It is trained through a large number of spot images with different shapes and occlusion conditions. The newly created loss function is used in the training process. (Where S i is the original spot shape, S′ i is the optimized spot shape, W j is the model weight, λ and μ are regularization parameters, C l is the statistical feature of the model's middle layer feature map). After the training is completed, the actually collected spot image is input into the model to obtain the optimized spot shape, and then the center of mass coordinates are calculated to improve the accuracy of the center of mass calculation.

[0045] Control matrix generation and transmission: Based on the slope calculated by Hartmann and the center of mass coordinates calculated by the center of mass, the control matrix of the deformable mirror and the tilt mirror is generated by a newly created multi-parameter fitting algorithm. The slope deviation is ΔS x , ΔS y , the center of mass coordinate deviation is Δx c , Δy c , control matrix elements

[0046] (where a1-a n and b1-b m 、p k ,q k 、r k 、s kThe algorithm is able to comprehensively consider various combinations of slope and center of mass coordinate deviation to generate a more accurate control matrix.

[0047] After the control matrix is ​​generated, it is transmitted to the Linux system and the beam correction stage through the PCIE bus. During the transmission process, the newly created error correction coding algorithm is used to group the control matrix and add a check bit to each group of data. The value of the check bit is generated according to the data bit and the newly created coding rule. Let the data bit be D i , the check digit is P j , P j =g(D1,D2,…,D n ), (where g is a newly created function relationship), the receiving end recalculates the check bit according to the received data and compares it with the transmitted check bit. If they are inconsistent, a newly created error location and correction algorithm is used to determine the error location and correct it by analyzing the logical relationship between the check bit difference and the data bit. At the same time, the data is encrypted during the transmission process. The encryption algorithm uses an encryption method based on quantum key distribution to ensure the security of data transmission.

[0048] After receiving the control matrix, the beam correction stage controls the deformable mirror and the tilt mirror.

[0049] For a deformable mirror, according to the control matrix element value M ij The mirror is deformed by applying a voltage signal through a high-voltage amplifier. Suppose the output voltage of the high-voltage amplifier is V o ut, the deformation displacement of the deformable mirror is d, (where k1-k 13 The model can accurately describe the relationship between the control matrix element value and the deformable mirror surface deformation, achieve more precise beam wavefront phase adjustment, monitor the mirror surface deformation in real time during the operation of the deformable mirror, and fine-tune the physical model parameters according to the feedback information to compensate for the influence of factors such as material aging on deformation control.

[0050] For the tilt mirror, the tilt angle is adjusted by the motor drive system according to the angle information in the control matrix. The motor drive system calculates the number of motor rotation steps according to the angle information in the control matrix. The target angle of the tilt mirror is θ, and the number of motor rotation steps is N, N = h (θ) (where h is a newly created piecewise function relationship, and different calculation coefficients are used according to different angle ranges). During the rotation of the motor, the actual rotation angle of the motor is monitored in real time. If the deviation from the target angle exceeds the preset threshold, a newly created dynamic adjustment strategy is adopted to adjust the motor drive voltage and pulse frequency according to the size and direction of the deviation. The adjustment method is to use a fuzzy logic control algorithm to determine the voltage and frequency adjustment amount according to the deviation and the deviation change rate, so that the tilt mirror can be quickly and accurately adjusted to the target angle, thereby improving the accuracy and speed of the correction of the beam propagation direction. At the same time, the motor operation status is monitored in real time, including parameters such as temperature and current. If an abnormality is detected, the motor drive parameters are adjusted in time or an alarm is issued.

[0051] During the entire communication process from the sending end to the receiving end of quantum communication, the above adaptive optical correction process is continuously repeated, and the system monitors the state of the light beam in real time. If the energy loss of the light beam or the fidelity of the quantum state is abnormal, it is judged according to the newly created monitoring and evaluation algorithm. Suppose the energy loss rate of the light beam is η i The quantum state fidelity is F, and the evaluation value is calculated (where α, β, γ and δ are parameters determined according to the characteristics of the quantum communication system). The evaluation value is used to determine whether the frequency of the Hartmann calculation and the center of mass calculation needs to be adjusted. If E exceeds the preset threshold, the calculation frequency is increased. At the same time, the newly created environmental adaptation algorithm adjusts the correction parameters of the adaptive optical system in real time according to environmental parameters such as atmospheric turbulence intensity, temperature, and humidity, such as the sub-aperture division method in the Hartmann calculation and the parameters in the control matrix generation algorithm. The adjustment rules are based on a mapping relationship established based on a large amount of experimental data and machine learning models to adapt to different quantum communication environments and ensure that the light beam is always in a good correction state.

[0052] Through the application of this embodiment, in the process of quantum communication, the interference of factors such as atmospheric turbulence on the light beam is effectively overcome. First, due to the low latency and low power consumption characteristics of FPGA when processing key computing tasks, the correction frequency of the adaptive optical system is significantly improved, and the wavefront distortion can be corrected in time, thereby greatly enhancing the stability and integrity of the light beam, which ensures the accuracy of the quantum state during transmission, reduces the bit error rate, and improves the reliability of quantum communication. Secondly, the CPU+FPGA hardware architecture and the application of various new algorithms overcome the problems of complex instruction flow control logic and high power consumption in traditional technologies. The entire system can operate efficiently and stably, adapt to the complex and changeable quantum communication environment, and provide strong technical support for the large-scale practical application of quantum communication technology, and promote the further development of quantum communication technology in actual scenarios. At the same time, mechanisms such as data encryption and error correction ensure the accuracy and security of control signal transmission, and further improve the overall performance of the quantum communication system.

[0053] Embodiment 2

[0054] In a satellite quantum communication scenario, a communication link is established between a ground station and a quantum communication satellite, and the adaptive optical correction system of the present invention is integrated into the receiving device of the ground station.

[0055] After the device is turned on, the system is initialized and started, and the CPU of the ground station loads the customized Linux operating system. The system is optimized for the characteristics of satellite quantum communication and has efficient data processing and task scheduling capabilities. It can give priority to processing data related to key information of quantum states. At the same time, the CPU detects the connection status with the FPGA and other hardware (such as optical antennas for receiving satellite signals, signal amplifiers, deformable mirrors and tilt mirrors related to beam correction, etc.). According to the expected signal strength and accuracy requirements of satellite communication, the FPGA configures the internal cache size and data transmission format, and prepares to receive the light signal from the optical antenna and convert it into image data for processing.

[0056] The optical antenna receives quantum communication signal light from the satellite, the FPGA converts it into image data, caches and generates a video stream for transmission to the CPU's Linux system, and the Linux system starts the signal detection program and begins to analyze the signal features in the video stream to determine whether a valid quantum communication signal has been successfully captured.

[0057] When a valid signal is detected, the FPGA performs adaptive optical correction operations.

[0058] The Hartmann wavefront sensor is used to divide the received beam wavefront into multiple sub-apertures, and the offset of the center of mass of the light spot in each sub-aperture relative to the reference position is calculated to determine the slope. The coordinates of the center of mass of the light spot (x c ,y c) and the reference coordinate (x r ,y r ), subaperture size parameter d, then the slope S x =(x c -x r ) / d,S y =(y c -y r ) / d. If the signal is weak or the noise caused by atmospheric interference affects the calculation, an adaptive weighted filtering algorithm is used. The distance between the pixel point (x, y) and the center of the light spot (x0, y0) is set to Weight w(r) = 1 / (1+αr β +γr δ )(α, β, γ and δ are determined according to the noise characteristics of the satellite communication environment), the pixel value after filtering

[0059] N(x,y) is a dynamically adjusted set of neighborhood pixels to improve the accuracy of slope calculation.

[0060] Based on the light spot calculated by Hartmann, calculate the center of mass coordinates (x c ,y c ), the formula is is the pixel coordinate of the light spot, p i is the pixel intensity, n is the total number of spot pixels). If the light spot is irregular in shape due to cloud cover, a morphological optimization algorithm based on deep learning is used.

[0061] Build a deep learning model that takes a spot image as input and outputs an optimized shape. It includes specially designed convolutional layers, pooling layers, and fully connected layers. The loss function is used during training.

[0062] (Where S i is the original spot shape, S′ i is the optimized spot shape, W j is the model weight, λ and μ are regularization parameters, C l is the statistical feature of the feature map of the middle layer of the model). After training, the actual spot image is input to obtain the optimized shape and then the center of mass coordinates are calculated to improve the calculation accuracy.

[0063] Calculate the coordinate deviation ΔS based on the Hartmann calculation slope and center of mass x , ΔS y , Δx c , Δy c , the control matrix M is generated by multi-parameter fitting algorithm ij ,

[0064] (a1-a n Parameters such as the above are determined according to the characteristics of the satellite communication optical system). After being generated, they are transmitted to the CPU and the beam correction link via the PCIE bus. The transmission adopts the error correction coding check algorithm, and the packet coding adds the check bit P j =g(D1,D2,…,D n )(g is a newly created function relationship), the receiving end compares the check bits, and uses the error location correction algorithm to process them when they are inconsistent. At the same time, quantum encryption is used to ensure transmission security.

[0065] After receiving the control matrix, the beam correction part controls the deformable mirror according to the control matrix element M ij , high voltage amplifier output voltage V out The deformable mirror surface is deformed and displaced by d, and the relationship is where k1-k 13 The deformation is monitored in real time during the process and the model parameters are fine-tuned to adapt to environmental changes, depending on the material and optical properties of the variable-open mirror.

[0066] For the tilt mirror, based on the angle information of the control matrix, the motor drive system calculates the number of motor rotation steps N, N = h(θ) (h is a piecewise function with a coefficient determined according to the angle range), monitors the actual angle during rotation, and uses a fuzzy logic control algorithm to adjust the motor drive voltage and pulse frequency when the deviation exceeds the threshold. The adjustment amount is determined based on the deviation and the rate of change. At the same time, the motor temperature, current, etc. are monitored, and parameters are adjusted or an alarm is issued when an abnormality occurs.

[0067] During the entire process of satellite quantum communication, the above adaptive optical correction process is continuously repeated, and the system monitors the beam performance in real time. If the beam energy loss is abnormal or the quantum state fidelity decreases, the energy loss rate is defined as η and the fidelity F is calculated according to the monitoring and evaluation algorithm. (parameters such as α are determined according to system characteristics). When the threshold is exceeded, the frequency of Hartmann and center of mass calculations is increased. The newly created environmental adaptation algorithm adjusts the correction parameters according to atmospheric turbulence, temperature, humidity, etc., such as Hartmann calculation sub-aperture division, control matrix generation parameters, etc., based on the mapping relationship between experimental data and machine learning models to ensure that the light beam is in a good correction state.

[0068] In the satellite quantum communication scenario, this embodiment has significant advantages. First, the efficient computing characteristics of FPGA ensure that the adaptive optical system can quickly respond to the wavefront distortion of satellite signals caused by changes in the atmosphere and space environment, and correct the light beam in time, greatly improving the stability and reliability of the quantum communication link. This enables quantum communication between satellites and ground stations to maintain high accuracy in complex space environments, effectively reducing signal interruptions and bit errors. Secondly, the CPU+FPGA architecture combined with innovative algorithms solves the high latency and high power consumption problems of traditional technologies when processing complex satellite communication signals. The system can adjust in real time according to factors such as changes in satellite orbits and changes in atmospheric conditions to ensure the continued stability of quantum communication. The data encryption and error correction mechanism effectively ensures the security and accuracy of the control signal transmitted between the satellite and the ground, preventing the signal from being stolen or tampered with, further enhancing the overall performance of the satellite quantum communication system, and laying a solid foundation for the construction and application of future satellite quantum communication networks.

[0069] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the same elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. An adaptive optical remote control interaction method for quantum communication, characterized in that: The method comprises the following steps: S1. System initialization and video stream generation: After the device is turned on, the CPU and FPGA integrated on a board are initialized. The FPGA obtains image data from the light source detection camera, performs cache processing and generates a video stream, which is transmitted to the Linux system of the CPU. The initialization of the FPGA includes configuring the internal cache size and data transmission format according to the accuracy requirements of quantum communication; S2, light source detection and adaptive optical correction start: The Linux system starts the light source detection program after receiving the video stream. When the light source is detected, the FPGA performs the following operations: S201, Hartmann calculation: Based on the data collected by the Hartmann wavefront sensor, the sensor divides the beam wavefront into multiple sub-apertures. For each sub-aperture, the slope is determined by calculating the offset of the center of mass of the light spot relative to the reference position. The coordinates of the center of mass of the light spot in the two-dimensional plane are defined as (x c ,y c ), the reference position coordinate is (x r ,y r ), the subaperture size parameter is d, then the slope S x =(x c -x r ) / d,S y =(y c -y r ) / d, during the calculation process, the spot signal is weak and the noise interference is large, so an adaptive weighted filtering algorithm is used to assign different weights according to the distance between the pixel point and the center of the spot for filtering; S202, centroid calculation: Based on the light spot calculated by Hartmann, the centroid coordinates (x c ,y c ), the calculation method is Where (x i ,y i ) is the coordinate of the i-th pixel in the spot, p i is the intensity value of the pixel, n is the total number of pixels in the spot, the spot shape is irregular and partially blocked, and the morphological optimization algorithm based on deep learning is used to train the neural network to identify the effective area of ​​the spot and optimize the shape before calculating the centroid coordinates; S203, control matrix generation and transmission: Based on the slope calculated by Hartmann and the center of mass coordinates calculated by the center of mass, the control matrix of the deformable mirror and the tilt mirror is generated by a multi-parameter fitting algorithm, and the slope deviation is defined as ΔS x , ΔS y , the center of mass coordinate deviation is Δx c , Δy c , control matrix element M ij =f(ΔS x ,ΔS y ,Δx c ,Δy c ), where f is a complex function relationship. After the control matrix is ​​generated, it is transmitted to the Linux system and the beam correction stage through the PCIE bus. During the transmission process, an error correction coding algorithm is used to ensure data accuracy; S3, beam correction: After receiving the control matrix in the beam correction stage, the deformable mirror and the tilt mirror are controlled. For the deformable mirror, a voltage signal is applied through a high-voltage amplifier according to the element value of the control matrix to deform the mirror surface. The relationship between the deformation degree and the voltage value is determined according to the physical model. For the tilt mirror, the tilt angle is adjusted through the motor drive system according to the angle information in the control matrix. The motor drive system adopts an intelligent control algorithm. S4. Continuous repair of quantum communication: In the process of quantum communication from the sending end to the receiving end, steps S2 and S3 are continuously repeated. If abnormalities occur in the energy loss of the light beam and the fidelity of the quantum state, the frequency of the Hartmann calculation and the center of mass calculation are adjusted according to the monitoring and evaluation algorithm, and the light source and optical components are checked. At the same time, the parameters of the adaptive optical system are dynamically adjusted according to the environmental adaptation algorithm.

2. The adaptive optical remote control interaction method for quantum communication according to claim 1, characterized in that: In step S1, the initialization process of the CPU includes loading a Linux operating system specially optimized for quantum communication. The kernel of the operating system has been customized and has the ability to efficiently schedule quantum communication data processing. Its scheduling strategy dynamically allocates computing resources based on quantum state priority and data real-time requirements, and automatically detects the communication interface status with FPGA and other hardware devices at startup. If a communication interface abnormality is detected, an interface repair algorithm is used for automatic repair. The algorithm locates the fault point by sending a specific detection signal sequence and analyzing the feedback signal. Specifically, the feedback signal is spectrally analyzed, the characteristic frequency is extracted and compared with a preset fault frequency library, and then targeted repair operations are performed according to the fault type. At the same time, the repair process and result information are recorded, and the repair record is stored in a specific non-volatile storage area, and the storage capacity is managed in a cyclic overwrite manner.

3. The adaptive optical remote control interaction method for quantum communication according to claim 1, characterized in that: In step S201, Hartmann calculation, the adaptive weighted filtering algorithm is specifically as follows: define the distance between the pixel point (x, y) and the center of the light spot (x0, y0) as Assign weight w(r)=1 / (1+αr β +γr δ ), where α, β, γ and δ are parameters determined according to the noise characteristics of the quantum communication environment, and the calculation formula of the filtered pixel value p′(x, y) is: It is a set of neighborhood pixels centered at (x, y), and the neighborhood size is dynamically adjusted according to the spot size and noise distribution. This algorithm effectively removes noise and retains the spot detail information based on the characteristics of spot noise distribution in quantum communication. At the same time, the filtering effect is monitored in real time during the execution of the algorithm. If the centroid offset of the spot after filtering exceeds the preset threshold, the weight parameters are automatically adjusted to optimize the filtering performance.

4. The adaptive optical remote control interaction method for quantum communication according to claim 1, characterized in that: In step S202, centroid calculation, the morphological optimization algorithm based on deep learning includes the following steps: Construct a deep learning model, which takes the spot image as input and outputs the optimized spot shape. The model structure includes convolution layer, pooling layer and fully connected layer. The kernel function of the convolution layer adopts the function form optimized for the edge features of the spot, the pooling layer adopts an adaptive step size strategy, and the fully connected layer adopts a neuron activation function. A large number of spot images with different shapes and occlusion conditions are used for training, and the loss function is used in the training process. Where S i is the original spot shape, S′ i is the optimized spot shape, W j is the model weight, λ and μ are regularization parameters, C l is the statistical characteristics of the feature map of the middle layer of the model; After the training is completed, the actual collected spot image is input into the model to obtain the optimized spot shape, and then the center of mass coordinates are calculated. This algorithm adapts to the complex changes in the spot shape, and continuously collects new spot samples for incremental learning during the use of the model to continuously improve the model performance.

5. The adaptive optical remote control interaction method for quantum communication according to claim 1, characterized in that: In step S203, control matrix generation and transmission, the multi-parameter fitting algorithm is specifically as follows: define where a1-a n and b1-b m 、p k ,q k 、r k 、s k In order to determine the parameters based on the optical characteristics and correction requirements of the quantum communication system, the algorithm is optimized through a large amount of experimental data and simulation calculations. The algorithm comprehensively considers various combinations of slope and center of mass coordinate deviation to generate a more accurate control matrix. At the same time, an adaptive parameter adjustment mechanism is introduced into the algorithm to dynamically adjust some parameters to adapt to different atmospheric disturbances based on the real-time monitoring results of the beam quality during quantum communication.

6. The adaptive optical remote control interaction method for quantum communication according to claim 1, characterized in that: In step S203, control matrix generation and transmission, the error correction coding check algorithm includes: group coding the control matrix, adding a check bit to each group of data, the value of the check bit is generated according to the data bit and the coding rule, and the data bit is defined as D i , the check digit is P j , P j =g(D1,D2,…,D n ), where g is a function relationship. The check bit is recalculated at the receiving end based on the received data and compared with the transmitted check bit. If there is any inconsistency, an error location and correction algorithm is used. This algorithm determines the error location and corrects it by analyzing the logical relationship between the check bit difference and the data bit. Specifically, an error location discrimination matrix is ​​constructed, and the matrix element value is calculated according to the check bit comparison result to determine the error location. In addition, the data is encrypted during the transmission process. The encryption algorithm uses an encryption method based on quantum key distribution.

7. The adaptive optical remote control interaction method for quantum communication according to claim 1, characterized in that: In step S3, the deformable mirror control of beam correction, the physical model is: define the control matrix element value M ij , the high voltage amplifier output voltage is V out , the deformation displacement of the deformable mirror surface is d, where k1-k 13 The parameters are determined according to the material characteristics and optical properties of the deformable mirror, and are obtained through experimental measurement and theoretical calculation. This model accurately describes the relationship between the control matrix element value and the deformation of the deformable mirror surface, and monitors the deformation of the mirror surface in real time during the operation of the deformable mirror, and fine-tunes the physical model parameters based on the feedback information.

8. The adaptive optical remote control interaction method for quantum communication according to claim 1, characterized in that: In step S3, the control of the tilt mirror for beam correction, the intelligent control algorithm includes: the motor drive system calculates the number of motor rotation steps according to the angle information in the control matrix, defines the target angle of the tilt mirror as θ, the number of motor rotation steps as N, N=h(θ), wherein h is a piecewise function relationship, and different calculation coefficients are used according to different angle ranges. During the rotation of the motor, the actual rotation angle of the motor is monitored in real time. If the deviation from the target angle exceeds a preset threshold, a dynamic adjustment strategy is adopted to adjust the motor drive voltage and pulse frequency according to the size and direction of the deviation. The adjustment method is to adopt a fuzzy logic control algorithm to determine the voltage and frequency adjustment amount according to the deviation and the deviation change rate. At the same time, the motor operating status is monitored in real time, including temperature and current parameters. If an abnormality is detected, the motor drive parameters are adjusted in time and an alarm is issued.

9. The adaptive optical remote control interaction method for quantum communication according to claim 1, characterized in that: In step S4, the abnormal processing of quantum communication continuous repair, the monitoring and evaluation algorithm is: define the beam energy loss rate as η, the quantum state fidelity as F, and calculate the evaluation value Among them, α, β, γ and δ are parameters determined according to the characteristics of the quantum communication system. Whether the frequency of Hartmann calculation and center of mass calculation needs to be adjusted is determined according to the size of the evaluation value. If E exceeds the preset threshold, the calculation frequency is increased. At the same time, the environmental adaptation algorithm adjusts the correction parameters of the adaptive optical system in real time according to the atmospheric turbulence intensity, temperature and humidity environmental parameters. The adjustment rules are based on the mapping relationship established by a large amount of experimental data and machine learning models, and detailed exception information and processing procedures are recorded during the exception handling process.

Images