An airborne laser wind sounding radar and cloud precipitation detection cooperation system

Abstract

The present application relates to meteorological detection technical field, specifically, it relates to a kind of airborne laser wind measurement radar and cloud precipitation detection coordination system. Including: intelligent sensing detection unit;Particle swarm algorithm data processing unit, the particle swarm algorithm data processing unit constructs multi-physical quantity coupling dynamic calibration model;Coordination control regulation unit.The present application is aimed at airborne scene, based on wind field gradient modulation, body vibration optical axis jitter, temperature and pressure wavelength shift three major physical mechanisms, constructs multi-physical quantity coupling dynamic calibration model, quantifies error coefficient and builds high-dimensional calibration matrix;Using the improved particle swarm algorithm of fusion multi-dimension coding, dynamic inertia weight, local optimal escape factor and beibu restraints, according to laser echo signal-to-noise ratio grading frequency band, share calibration parameter completes synchronous solution, realizes multi-error unified modeling and accurate correction, synchronous completion coordination inversion and noise reduction, improves the stability and reliability of detection data solution.

Description

Technical Field

[0001] This invention relates to the field of meteorological detection technology, and more specifically, to a collaborative system for airborne laser wind radar and cloud precipitation detection. Background Technology

[0002] Airborne laser wind radar and cloud precipitation detection are core technologies in the field of aviation meteorological observation. They are widely used in key scenarios such as flight safety early warning and meteorological forecast modeling. Their core objective is to simultaneously acquire high-precision wind field structure data and characteristic information such as particle size and velocity distribution of cloud precipitation particles in an airborne environment, providing basic support for the safety of aviation operations and the application value of meteorological data.

[0003] In existing technologies, relevant patents have explored the fields of meteorological monitoring and lidar atmospheric detection. For example, Chinese patent CN202311337565.4 discloses a meteorological monitoring system and method based on multi-source data fusion. This system includes modules for acquiring meteorological environmental data, generating monitoring models, and determining meteorological information. By analyzing multi-source meteorological data and using screening to determine models, it improves the quality of data fusion and the accuracy and relevance of the meteorological information to be released. Another example is Chinese patent CN202011290229.5, which discloses an atmospheric boundary layer classification method and device based on wind-measuring lidar. This method uses the power spectrum and inversion parameters of the wind-measuring lidar, combined with information such as carrier-to-noise ratio and spectral width, to identify and classify phenomena such as clouds and precipitation within the atmospheric boundary layer. Applied to aviation safety assurance, it has the advantages of high flexibility and high spatiotemporal resolution.

[0004] Despite the design advantages of the aforementioned technical solutions, they also suffer from the following technical shortcomings: Firstly, they lack a unified calibration and accurate calculation mechanism for the coupling errors of multiple physical quantities in airborne scenarios. CN202311337565.4 only processes meteorological information through multi-source data fusion and screening models, failing to consider the coupling effects of various errors such as the modulation of the detection object by wind field gradients in the airborne environment, the shift in the detection reference caused by airframe vibration, and the drift of equipment signals due to temperature and pressure changes. A systematic error calibration model has not been established. CN202011290229.5 focuses on the classification and identification of meteorological phenomena, and the data processing is not designed to adapt to high-dimensional error parameters. The optimization scheme for solving the problem also failed to introduce high-precision benchmark data to constrain the solution boundary, resulting in limited error correction and insufficient solution accuracy and stability. Secondly, it failed to achieve coordinated linkage and dynamic adaptation of detection parameters between laser wind measurement and cloud precipitation detection: CN202311337565.4 did not establish a collaborative architecture for laser wind measurement and cloud precipitation detection, and the two types of core detection data were processed independently without a linkage inversion mechanism; CN202011290229.5 relied solely on a single lidar device to complete the task, without forming a collaboration with the cloud precipitation detection unit, and neither of them designed a parameter adjustment mechanism based on real-time data feedback, making it unable to adapt to dynamic changes in the airborne environment and affecting detection adaptability. In view of this, we propose an airborne laser wind measurement radar and cloud precipitation detection collaborative system. Summary of the Invention

[0005] The purpose of this invention is to provide a collaborative system for airborne laser wind measurement radar and cloud precipitation detection, in order to solve the problems mentioned in the background art, such as the lack of a unified calibration and accurate calculation mechanism for the coupling error of multiple physical quantities in airborne scenarios and the failure to achieve collaborative linkage and dynamic adaptation of detection parameters between laser wind measurement and cloud precipitation detection.

[0006] To address the aforementioned technical problems, the present invention aims to provide a coordinated system for airborne laser wind measurement radar and cloud precipitation detection, comprising:

[0007] The intelligent sensing and detection unit adopts a multi-modal integrated intelligent sensor to collect laser echo signals from airborne laser wind radar and information on particle size and velocity distribution of cloud precipitation, and outputs raw detection data.

[0008] The particle swarm optimization (PSO) data processing unit constructs a multi-physical quantity coupled dynamic calibration model. This model is based on three major physical mechanisms: the modulation effect of wind field gradient on the terminal velocity of precipitation particles, the optical axis jitter error caused by body vibration, and the laser wavelength drift caused by temperature and pressure changes. It constructs a high-dimensional calibration matrix and adopts an improved PSO algorithm that combines multi-dimensional particle encoding, dynamic inertial weighting, and local optimal escape factor. The particle swarm search boundary is dynamically constrained by BeiDou reference data. The original data and calibration matrix are synchronously solved in different frequency bands. The frequency bands are divided according to the signal-to-noise ratio of the laser echo. The high-frequency band is used for precipitation particle spectrum analysis, and the low-frequency band is used for wind field structure reconstruction. By sharing calibration matrix parameters, multi-physical quantity coupled calibration, data collaborative inversion, and noise reduction are completed synchronously.

[0009] The collaborative control and adjustment unit calls the improved particle swarm algorithm of the particle swarm algorithm data processing unit to construct a corresponding particle swarm algorithm optimization model, and adjusts the lidar emission frequency and the intelligent sensor sampling period according to the real-time data feedback from the intelligent sensing and detection unit.

[0010] An integrated data output unit receives the processing results from the particle swarm algorithm data processing unit and combines them with environmental adaptation information fed back by intelligent sensors to generate standardized collaborative detection data.

[0011] As a further improvement to this technical solution, the intelligent sensing and detection unit includes a laser echo acquisition module, a cloud precipitation particle sensing module, and an environment adaptation and synchronization module, wherein:

[0012] The laser echo acquisition module interfaces with the airborne laser wind radar to acquire laser echo signals and perform preliminary signal filtering.

[0013] The cloud precipitation particle sensing module collects information on the particle size and velocity distribution of cloud precipitation particles and generates particle characteristic data.

[0014] The environment adaptation synchronization module receives the filtered signal from the laser echo acquisition module and the particle characteristic data from the cloud precipitation particle sensing module, completes the time synchronization of the filtered signal and the particle characteristic data, and outputs synchronized raw detection data.

[0015] As a further improvement to this technical solution, the particle swarm algorithm data processing unit includes a multi-physical quantity calibration model construction module, an improved particle swarm algorithm operation module, and a frequency band collaborative solution module, wherein:

[0016] The multi-physical quantity calibration model construction module is based on three major physical mechanisms: the modulation effect of wind field gradient on the terminal velocity of precipitation particles, the optical axis jitter error caused by body vibration, and the laser wavelength drift caused by temperature and pressure changes. It quantifies the error influence coefficient and constructs a high-dimensional calibration matrix.

[0017] The improved particle swarm optimization algorithm operation module includes a particle encoding submodule, a parameter adjustment submodule, and a BeiDou reference constraint submodule, wherein:

[0018] The particle encoding submodule performs multi-dimensional particle encoding to adapt to the dimensional features of the high-dimensional calibration matrix;

[0019] The parameter control submodule configures dynamic inertia weights and local optimal escape factors to optimize the algorithm's convergence performance;

[0020] The BeiDou reference constraint submodule dynamically constrains the particle population search boundary using BeiDou reference data to improve solution accuracy.

[0021] The frequency band collaborative solution module divides the frequency bands according to the signal-to-noise ratio of the laser echo, performs synchronous solution of the original data and calibration matrix in the frequency bands, and completes multi-physical quantity coupling calibration, data collaborative inversion and noise reduction processing by sharing calibration matrix parameters.

[0022] As a further improvement to this technical solution, the process of quantifying the error influence coefficient and constructing a high-dimensional calibration matrix by the multi-physical quantity calibration model construction module includes the following steps;

[0023] S21.1 For the three major physical mechanisms of wind field gradient modulation effect, optical axis jitter due to body vibration, and wavelength drift due to temperature and pressure changes, the corresponding error influence coefficients are quantified respectively. ;in, This represents the error influence coefficient of the wind field gradient modulation effect on the detection data; This represents the error influence coefficient of the optical axis jitter caused by the vibration of the machine body on the detection data; This represents the error influence coefficient of wavelength drift due to temperature and pressure changes on the detection data;

[0024] S21.2, Based on the error influence coefficient Constructing a high-dimensional calibration matrix The matrix dimension is adapted to the dimension of the original data output by the intelligent sensing and detection unit, so as to realize the unified modeling of errors of multiple physical quantities.

[0025] As a further improvement to this technical solution, the particle encoding submodule performs multi-dimensional particle encoding as follows:

[0026] S22.1 Determine the particle encoding dimension, and align the particle encoding dimension with the high-dimensional calibration matrix generated by the multi-physical quantity calibration model construction module. The dimensions are completely identical;

[0027] S22.2 Constructing a multi-dimensional particle encoding vector Each encoded component is associated with a high-dimensional calibration matrix. The error impact coefficients in the model correspond one-to-one, achieving precise adaptation between the coding and calibration models.

[0028] As a further improvement to this technical solution, the process of configuring dynamic inertia weight and local optimal escape factor by the parameter control submodule includes the following steps;

[0029] S22.3, Configure dynamic inertia weights The values ​​are dynamically adjusted according to the algorithm iteration process to adapt to the performance requirements of global search and local convergence.

[0030] S22.4. Set the trigger condition for the local optimal escape factor, and after triggering, adjust the perturbation amplitude. Adjusting the particle positions helps prevent the algorithm from getting stuck in local optima.

[0031] As a further improvement to this technical solution, the process by which the Beidou reference constraint submodule dynamically constrains the particle population search boundary using Beidou reference data includes the following steps;

[0032] S22.5. Based on BeiDou reference data, obtain real-time error correction values ​​and convert them into dynamic adjustment values ​​for the particle swarm search boundary. ;

[0033] S22.6, Based on the dynamic adjustment value Update particle population search boundary .

[0034] As a further improvement to this technical solution, the process of the frequency band collaborative solution module dividing the frequency bands according to the signal-to-noise ratio of the laser echo and sharing the calibration matrix parameters to complete the solution includes the following steps;

[0035] S23.1 Calculate the laser echo signal-to-noise ratio The high-frequency band and low-frequency band are divided according to the signal-to-noise ratio threshold;

[0036] S23.2 Extracting the high-dimensional calibration matrix generated by the multi-physical quantity calibration model construction module Parameters, constructing cross-band shared parameter weights ;

[0037] S23.3, Weighting of shared parameters across frequency bands and frequency bands Simultaneously complete multi-physical quantity coupling calibration, data collaborative inversion, and noise reduction processing.

[0038] As a further improvement to this technical solution, the collaborative control and adjustment unit includes a data feedback acquisition module, an optimization model construction module, and a parameter adjustment execution module, wherein:

[0039] The data feedback acquisition module is connected to the environment adaptation and synchronization module of the intelligent sensing and detection unit to obtain the synchronized raw detection data output by it as real-time feedback data.

[0040] The optimization model construction module calls the improved particle swarm algorithm of the particle swarm algorithm data processing unit and constructs the corresponding particle swarm algorithm optimization model based on the real-time feedback data of the data feedback acquisition module.

[0041] The parameter adjustment execution module adjusts the emission frequency of the airborne lidar and the sampling period of the multimodal integrated intelligent sensor in the intelligent sensing and detection unit based on the model calculation results output by the optimization model construction module.

[0042] As a further improvement to this technical solution, the integrated data output unit includes a processing result receiving module, an environmental information fusion module, and a standardized data generation module, wherein:

[0043] The processing result receiving module is connected to the frequency band collaborative solution module and receives the results of multi-physical quantity coupling calibration, data collaborative inversion and noise reduction output by the frequency band collaborative solution module.

[0044] The environmental information fusion module obtains environmental parameter information from the raw detection data output by the intelligent sensing and detection unit, and fuses the environmental parameter information with the processing result received by the processing result receiving module.

[0045] The standardized data generation module generates standardized collaborative detection data based on the fusion results of the environmental information fusion module.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0047] 1. This invention constructs a multi-physical quantity coupled dynamic calibration model based on three major physical mechanisms: the modulation effect of wind field gradient on the terminal velocity of precipitation particles, the optical axis jitter error caused by body vibration, and the laser wavelength drift caused by temperature and pressure changes. It quantifies the error influence coefficient and builds a high-dimensional calibration matrix. At the same time, it adopts an improved particle swarm algorithm that combines multi-dimensional particle coding, dynamic inertial weight, local optimal escape factor and Beidou reference data constraints. It divides the frequency band according to the laser echo signal-to-noise ratio and shares the calibration matrix parameters to complete synchronous calculation. This realizes unified modeling and accurate correction of multi-physical quantity errors in airborne scenarios, and simultaneously completes data collaborative inversion and noise reduction processing, effectively improving the stability and reliability of detection data calculation.

[0048] 2. This invention, through the multimodal integrated design of intelligent sensing and detection units, achieves the acquisition and timing synchronization of laser echo signals and cloud precipitation particle characteristic data, establishing a collaborative working architecture for laser wind measurement and cloud precipitation detection. Simultaneously, by leveraging a collaborative control and adjustment unit to invoke an improved particle swarm optimization algorithm to construct an optimized model, it dynamically adjusts the lidar emission frequency and sensor sampling period based on real-time detection data. Then, through an integrated data output unit, it fuses the processing results with environmental adaptation information to generate standardized data. This achieves collaborative linkage between the two types of detection and dynamic adaptation of detection parameters, enhancing the system's adaptability to complex airborne environments and meeting the core requirement of high-precision collaborative detection in aviation meteorological observation. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the system framework of the present invention;

[0050] The meanings of the labels in the diagram are as follows:

[0051] 1. Intelligent sensing and detection unit; 11. Laser echo acquisition module; 12. Cloud precipitation particle sensing module; 13. Environmental adaptation and synchronization module;

[0052] 2. Particle swarm optimization (PSO) algorithm data processing unit; 21. Multi-physical quantity calibration model construction module; 22. Improved Particle swarm optimization (PSO) algorithm operation module; 23. Frequency band collaborative solution module;

[0053] 3. Collaborative control and adjustment unit; 31. Data feedback acquisition module; 32. Optimization model construction module; 33. Parameter adjustment and execution module.

[0054] 4. Integrated data output unit; 41. Processing result receiving module; 42. Environmental information fusion module; 43. Standardized data generation module. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0056] like Figure 1 As shown, this embodiment provides a collaborative system for airborne laser wind measurement radar and cloud precipitation detection, including:

[0057] Intelligent sensing and detection unit 1 adopts a multi-modal integrated intelligent sensor to collect laser echo signals from airborne laser wind radar and information on particle size and velocity distribution of cloud precipitation, and outputs raw detection data.

[0058] In this embodiment, the intelligent sensing and detection unit 1 includes a laser echo acquisition module 11, a cloud precipitation particle sensing module 12, and an environment adaptation and synchronization module 13, wherein:

[0059] The laser echo acquisition module 11 connects to the airborne laser wind radar to acquire laser echo signals and perform preliminary signal filtering.

[0060] Specifically, the laser echo acquisition module 11 is used to interface with the airborne laser wind radar to complete the acquisition and preliminary filtering of laser echo signals, providing high-quality raw echo data for subsequent data processing. The specific implementation is as follows:

[0061] Interface Adaptation Design: A standardized optical interface is rigidly connected to the echo output port of the airborne laser wind radar. The interface incorporates a sealed dustproof gasket and vibration-damping structure, adapting to air pressure fluctuations (covering ±30% of standard atmospheric pressure) and vibration conditions (≤5g acceleration) in the airborne environment, ensuring no leakage or additional interference during optical signal transmission. The interface communication protocol is consistent with the output protocol of the airborne laser wind radar, supporting automatic protocol matching without manual configuration.

[0062] Signal Acquisition Process: A built-in low-noise photoelectric conversion device is used. The response band of this device precisely matches the emitted laser band of the airborne laser wind measurement radar (e.g., 1.55μm or 10.6μm), efficiently converting the incident laser echo optical signal into a weak electrical signal. The converted electrical signal enters a programmable gain conditioning circuit, where the gain value is dynamically adjusted according to the signal amplitude—automatically increasing the gain when the signal amplitude is below a preset threshold and decreasing the gain when it is above a saturation threshold, ensuring the signal is within the effective processing range. The conditioned electrical signal is then converted from analog to digital by a high-speed analog-to-digital converter. The sampling frequency is set to 8-12 times the emission frequency of the airborne laser wind measurement radar to ensure complete capture of the echo signal's waveform characteristics. The converted digital signal is temporarily stored in a high-speed buffer built into the laser echo acquisition module 11.

[0063] Preliminary filtering is implemented using a finite impulse response (FIR) digital filtering algorithm. The filtering parameters are dynamically configured based on the core technical parameters of the airborne laser wind-measuring radar. The filtering cutoff frequency is set to 1.5 times the radar transmission frequency to retain the effective frequency components of the laser echo signal while filtering out high-frequency electromagnetic interference and environmental noise generated by the airborne electronic equipment. The filtering process is performed in real-time by the microprocessor built into the laser echo acquisition module 11, employing a "segmented buffering-batch filtering" method. This means that a preset length of digital signal (e.g., 50-200 sampling points) is buffered before the filtering operation, avoiding the influence of single-point noise on the filtering results. After filtering, the module extracts key characteristic parameters of the echo signal, including peak echo intensity, round-trip time delay, and signal duration. These are organized into raw echo data according to a "timestamp-characteristic parameter" association format and transmitted to the environment adaptation synchronization module 13 via the internal data bus.

[0064] The cloud precipitation particle sensing module 12 collects information on the particle size and velocity distribution of cloud precipitation particles and generates particle characteristic data.

[0065] Specifically, the cloud precipitation particle sensing module 12 is used to collect information on the particle size and velocity distribution of cloud precipitation particles and generate standardized particle feature data. The specific implementation is as follows:

[0066] Detection Principle and Hardware Configuration: The cloud precipitation particle sensing module 12 adopts the laser scattering detection principle, with a built-in semiconductor laser emitting component and a high-sensitivity photoelectric receiving component. The laser beam emitted by the laser emitting component is collimated by an optical lens to form a parallel detection beam with a thickness ≤2mm and a diameter ≥10mm, constituting a fixed detection area. The photoelectric receiving component is arranged at a 30°-60° angle to the laser emission direction to capture the scattered light signal generated when particles pass through the detection area. The cloud precipitation particle sensing module 12 adopts a light-shielding and sealed design to avoid interference from stray light in the airborne environment on the detection results.

[0067] Particle size and velocity acquisition logic: When cloud precipitation particles pass through the probe beam, the scattered light signal is captured by the photoelectric receiving component and converted into an electrical pulse signal. The peak amplitude of the electrical pulse is positively correlated with the particle size, and the pulse duration is directly related to the time it takes for the particle to pass through the probe area. Particle parameters are calculated as follows:

[0068] Particle size calculation: Based on the preset "particle size-scattered light intensity" calibration curve, combined with the peak amplitude of the electric pulse, the particle size range is determined, and the particle size is divided into multiple continuous intervals (such as 0.5-2mm, 2-5mm, 5-10mm, etc.) to achieve the classification and statistics of particles of different sizes.

[0069] Velocity calculation: Based on the known length of the probe beam and the duration of the electrical pulse, the instantaneous velocity of the particles is calculated using the formula "velocity = beam length / pulse duration", and the velocity value of each particle is recorded simultaneously.

[0070] Particle Feature Data Generation: The cloud precipitation particle sensing module 12 collects particle information at a fixed sampling period (the sampling period can be dynamically adjusted according to the instructions of the subsequent collaborative control and adjustment unit 3). Within each sampling period, the particle size and velocity data of all detected particles are statistically analyzed to generate particle feature data. The particle feature data includes key information such as the number of particles in each particle size range within the sampling period, the average velocity of particles in each particle size range, the maximum particle size value, and the average velocity value. The data is stored in a structured format and transmitted to the environment adaptation synchronization module 13.

[0071] The environment adaptation synchronization module 13 receives the filtered signal from the laser echo acquisition module 11 and the particle characteristic data from the cloud precipitation particle sensing module 12, completes the time synchronization of the filtered signal and the particle characteristic data, and outputs synchronized raw detection data.

[0072] Specifically, the core task of the environment adaptation synchronization module 13 is to receive data from the laser echo acquisition module 11 and the cloud precipitation particle sensing module 12, complete the time synchronization processing, and output synchronized raw detection data. The specific implementation is as follows:

[0073] Time reference synchronization: The built-in Beidou time receiver receives the standard time signal from the Beidou satellite in real time, generates a local time reference with an accuracy of ≤1μs, and sends a synchronization clock signal to the laser echo acquisition module 11 and the cloud precipitation particle sensing module 12 through the internal synchronization signal interface, ensuring that the acquisition actions of the two modules are based on the same time scale, reducing timing deviation from the source.

[0074] Data reception and buffering: The system receives raw echo data from the laser echo acquisition module 11 and particle characteristic data from the cloud precipitation particle sensing module 12 via dual data receiving interfaces. During reception, a corresponding BeiDou timestamp (accurate to microseconds) is added to each frame of data. A built-in large-capacity buffer unit can simultaneously buffer 30 seconds of continuous output data from both modules to prevent data loss.

[0075] Timing synchronization processing: To address the potential sampling period differences between the laser echo acquisition module 11 and the cloud precipitation particle sensing module 12 (e.g., the laser echo acquisition period is 1ms, and the particle sensing sampling period is 5ms), the environment adaptation synchronization module 13 employs a linear interpolation completion mechanism to achieve timing synchronization, as detailed below:

[0076] Using the module data with a shorter sampling period as the baseline time axis, extract all time nodes on that time axis;

[0077] For asynchronous data points of another module, supplementary data corresponding to each node of the reference time axis is calculated by linear interpolation based on the values ​​of the two adjacent valid data points before and after it.

[0078] Synchronization verification: Perform consistency verification on the interpolated dataset, remove abnormal data points that exceed a reasonable range, and ensure the reliability of the synchronized data.

[0079] Synchronized Raw Detection Data Output: After time-series synchronization processing, the environment adaptation synchronization module 13 integrates the raw echo data and particle feature data in timestamp order to generate synchronized raw detection data. The synchronized raw detection data adopts a unified structured format, including timestamp fields, laser echo feature fields, and particle feature fields. It is output to the particle swarm algorithm data processing unit 2 through a standardized data interface, providing time-consistent raw data support for subsequent calibration and calculation processing.

[0080] The particle swarm optimization (PSO) data processing unit 2 constructs a multi-physical quantity coupled dynamic calibration model. This model is based on three major physical mechanisms: the modulation effect of wind field gradient on the terminal velocity of precipitation particles, optical axis jitter error caused by body vibration, and laser wavelength drift caused by temperature and pressure changes. It constructs a high-dimensional calibration matrix and employs an improved PSO algorithm combining multi-dimensional particle encoding, dynamic inertial weighting, and local optimal escape factors. The model dynamically constrains the particle swarm search boundary using BeiDou reference data and performs frequency-band synchronous calculations on the original data and calibration matrix. Frequency bands are divided according to the signal-to-noise ratio of the laser echo; the high-frequency band is used for precipitation particle spectrum analysis, and the low-frequency band is used for wind field structure reconstruction. By sharing calibration matrix parameters, multi-physical quantity coupled calibration, data collaborative inversion, and noise reduction are completed synchronously. The PSO data processing unit 2 includes a multi-physical quantity calibration model construction module 21, an improved PSO algorithm operation module 22, and a frequency-band collaborative calculation module 23.

[0081] In this embodiment, the multi-physical quantity calibration model construction module 21 quantifies the error influence coefficient and constructs a high-dimensional calibration matrix based on three major physical mechanisms: the modulation effect of wind field gradient on the terminal velocity of precipitation particles, the optical axis jitter error caused by body vibration, and the laser wavelength drift caused by temperature and pressure changes. The process of the multi-physical quantity calibration model construction module 21 quantifying the error influence coefficient and constructing a high-dimensional calibration matrix includes the following steps.

[0082] S21.1 For the three major physical mechanisms of wind field gradient modulation effect, optical axis jitter due to body vibration, and wavelength drift due to temperature and pressure changes, the corresponding error influence coefficients are quantified respectively. ;in, This represents the error influence coefficient of the wind field gradient modulation effect on the detection data; This represents the error influence coefficient of the optical axis jitter caused by the vibration of the machine body on the detection data; This represents the error influence coefficient of wavelength drift due to temperature and pressure changes on the detection data;

[0083] Specifically, The formula used to characterize the influence of unit wind field gradient changes on the deviation of precipitation particle terminal velocity detection results is as follows:

[0084] ;

[0085] This represents the deviation in the terminal velocity of precipitation particles, specifically calculated as follows: ,in The measured values ​​of particle terminal velocities collected by the cloud precipitation particle sensing module 12 are as follows. The theoretical value of particle terminal velocity without wind speed gradient interference (calculated based on the classical aerodynamic particle terminal velocity formula).

[0086] This represents the change in wind field gradient, specifically calculated as follows: ,in The current wind field gradient is obtained by the laser echo acquisition module 11. This is the standard reference wind field gradient set in the wind tunnel experiment.

[0087] During the quantization process, wind tunnel simulations were used to examine the wind field gradient ranges commonly encountered in airborne flight. Every Collect a set and A total of 30 sets of valid data were collected; calculations were performed on each set of data. and Then, the least squares method was used for linear fitting, and the final result was obtained. The stable value of the value is ensured to cover the error influence law of the entire wind field gradient range.

[0088] Specifically, The formula used to characterize the influence of unit vibration acceleration on the laser optical axis offset angle, and thus quantify its interference with the accuracy of laser echo signal detection, is as follows:

[0089] ;

[0090] This represents the laser optical axis offset angle deviation (unit: rad), specifically calculated as follows: ,in The measured optical axis offset angle is obtained by a laser collimation detection device under vibration conditions. The reference angle of the optical axis under vibration-free conditions (value 0 rad).

[0091] This represents the change in vibration acceleration, specifically calculated as follows: ,in The current airborne vibration acceleration is obtained from BeiDou attitude data. The reference vibration acceleration is 0g.

[0092] During the quantization process, random vibrations of 0.1–5g were applied using an airborne vibration simulation platform (simulating different flight conditions such as takeoff, cruise, and landing). Optical axis offset angle data were continuously collected for 10 seconds under each vibration acceleration, and the values ​​under that condition were calculated. Mean; a total of 20 vibration acceleration gradients were set, and 20 sets were obtained. and Data; obtained by averaging all the data. This ensures that it is compatible with the vibration characteristics of airborne equipment under all operating conditions.

[0093] Specifically, The formula used to characterize the impact of unit temperature and pressure changes on laser wavelength shift, and thus quantify its interference with the resolution accuracy of laser echo signals, is as follows:

[0094] ;

[0095] This represents the laser wavelength offset (unit: nm), specifically calculated as follows: ,in These are the measured laser wavelength values ​​collected by the wavelength analyzer under the current temperature and pressure conditions. The reference wavelength for lasers under standard temperature and pressure (25℃, 1013.25hPa);

[0096] The change in temperature (unit: °C) is expressed as follows: ,in The current airborne ambient temperature was obtained using BeiDou meteorological auxiliary data.

[0097] This represents the change in air pressure (unit: hPa), specifically calculated as follows: ,in The current airborne ambient air pressure was obtained using BeiDou meteorological auxiliary data.

[0098] The pressure influence weighting coefficient (unitless) is used to balance the influence of temperature and pressure on wavelength drift. Through temperature and pressure environment chamber experiments, the wavelength drift data under the combined effects of temperature and pressure are fitted using the least squares method, with a value range of 0.02 to 0.04.

[0099] During the quantization process, the temperature and pressure variations of the airborne environment (temperature -40℃ to 60℃, air pressure 800 to 1200 hPa) were simulated in a temperature and pressure environment chamber. A test point was set up every 5℃ for temperature and every 50 hPa for air pressure, for a total of 60 test points. Laser wavelength data was collected at each test point, and calculations were performed. , and Substitute the values ​​into the formula, and finally take the mean of the results calculated for all test points as the final value. .

[0100] S21.2, Based on the error influence coefficient Constructing a high-dimensional calibration matrix The matrix dimension is adapted to the dimension of the original data output by the intelligent sensing and detection unit 1, so as to realize the unified modeling of errors of multiple physical quantities.

[0101] Specifically, based on the quantized error impact coefficient Construct a high-dimensional calibration matrix that is fully adapted to the dimension of the raw data output by the intelligent sensing and detection unit 1. This achieves unified modeling of errors in multiple physical quantities, with the specific formula as follows:

[0102] ;

[0103] Among them, matrix elements The calculation formula is:

[0104] ;

[0105] This represents a high-dimensional calibration matrix (unitless), with dimensions of . , The number of independent dimensions of the raw data output by the intelligent sensing and detection unit 1 (i.e., the total number of laser echo characteristic parameters and cloud precipitation particle characteristic parameters).

[0106] Represents a high-dimensional calibration matrix The Line number Column element (unitless), corresponding to the first The first original data dimension is affected by the first Calibration weights for the impact of error types;

[0107] Indicates the first The sensitivity weight (unitless) of each original data dimension to the wind field gradient modulation effect is determined, ranging from 0 to 1. Based on the industry-standard "data dimension-wind field gradient correlation analysis method" for airborne precipitation particle detection, and combined with the correlation statistics of measured data from benchmark equipment (such as DMT Cloud Droplet Probe), the Pearson correlation coefficient between the data dimension and the wind field gradient is calculated and then normalized to the 0-1 interval to determine the sensitivity weight of different dimensions (e.g., when the first data dimension is affected by the wind field gradient, the second data dimension is affected by the wind field gradient). When the original data dimension belongs to the particle velocity-related dimension, its corresponding sensitivity weight The value range is 0.7~1.0; when the first... When a raw data dimension belongs to a dimension related to laser echo intensity, its corresponding sensitivity weight The value range is 0.2 to 0.5).

[0108] Indicates the first The sensitivity weight (unitless) of each original data dimension to the vibration of the optical axis jitter of the aircraft, ranging from 0 to 1, is determined according to the "Vibration Error Contribution Assessment Process" for airborne optical detection equipment (refer to "Quantification Method of Airborne Equipment Error Sources" in "Atmospheric Detection"). The error proportion of different data dimensions affected by optical axis jitter is tested using a vibration table to simulate airborne operating conditions. After normalization, the sensitivity weight in the range of 0 to 1 is obtained (e.g., when the first data dimension is...). When a raw data dimension belongs to a laser echo-related dimension, its corresponding sensitivity weight The value range is 0.6 to 1.0; when the first... When the original data dimension belongs to the particle number-related dimension, its corresponding sensitivity weight The value range is 0.1 to 0.3).

[0109] Indicates the first The sensitivity weights (unitless) of each original data dimension to wavelength drift caused by temperature and pressure changes are determined, ranging from 0 to 1. Based on the "Test Specifications for Temperature and Pressure Response Characteristics of Laser Equipment" (refer to "Laser Wavelength Environmental Response Analysis" in *Laser Principles*), and combined with environmental adaptability test data from airborne meteorological equipment, the influence of wavelength drift on different data dimensions is statistically analyzed and normalized to the 0-1 range to determine the sensitivity weights (e.g., when the first data dimension...). When a dimension of the original data belongs to a dimension related to the laser echo wavelength, its corresponding sensitivity weight The value range is 0.7~1.0; when the first... When the original data dimension belongs to the particle size-related dimension, its corresponding sensitivity weight The value range is 0.1 to 0.4.

[0110] During the construction process, firstly, all independent dimensions of the raw data output by the intelligent sensing and detection unit 1 (such as laser echo peak intensity, echo round-trip time delay, number of particles in each particle size range, average particle velocity, etc.) are statistically analyzed to determine... The specific values ​​were then determined; subsequently, through multiple sets of airborne detection experiments, the degree to which each data dimension was affected by the three types of errors was statistically analyzed to determine... The values ​​of are taken; finally, the matrix is ​​calculated by substituting them into the formula. All elements are combined to form a complete high-dimensional calibration matrix, achieving comprehensive coverage and unified modeling of errors in multiple physical quantities.

[0111] In this embodiment, the improved particle swarm optimization algorithm operation module 22 includes a particle encoding submodule, a parameter control submodule, and a BeiDou reference constraint submodule, wherein:

[0112] The particle encoding submodule performs multi-dimensional particle encoding to adapt to the dimensional features of the high-dimensional calibration matrix; the process of performing multi-dimensional particle encoding by the particle encoding submodule includes the following steps;

[0113] S22.1 Determine the particle encoding dimension, and align the particle encoding dimension with the high-dimensional calibration matrix generated by the multi-physical quantity calibration model construction module 21. The dimensions are completely identical;

[0114] Specifically, particle encoding dimension and high-dimensional calibration matrix The dimensions are completely consistent, ensuring that the encoding can cover all dimensions. All error calibration parameters are calculated using the following formula:

[0115] ;

[0116] Represents a high-dimensional calibration matrix The dimension (unitless), i.e. ;

[0117] This represents the number of independent dimensions (unitless) of the raw data output by the intelligent sensing and detection unit 1, and the high-dimensional calibration matrix. The dimensional definitions are consistent.

[0118] During implementation, the particle encoding submodule first reads the high-dimensional calibration matrix output by the multi-physical quantity calibration model construction module 21. The dimensional information is used to directly set the particle encoding dimension as... This avoids the omission of error calibration parameters due to dimension mismatch.

[0119] S22.2 Constructing a multi-dimensional particle encoding vector Each encoded component is associated with a high-dimensional calibration matrix. The error impact coefficients in the model correspond one-to-one, achieving precise adaptation between the coding and calibration models.

[0120] Specifically, based on a defined encoding dimension, a multi-dimensional particle encoding vector is constructed. Each encoded component is associated with a high-dimensional calibration matrix. Each element corresponds one-to-one, ensuring accurate matching between the encoding and calibration model. The calculation formula is as follows:

[0121] ;

[0122] Indicates the first Multidimensional encoding vectors of individual particles (unitless). The particle swarm size is set according to the requirements of solution accuracy and efficiency, and the value ranges from 50 to 200.

[0123] High-dimensional calibration matrix for Matrix, its first Line number Column elements are denoted as ( );

[0124] Represents the encoding vector The Each component (unitless) is associated with a high-dimensional calibration matrix. The Line number Column elements One-to-one correspondence (continuous mapping according to the "row-major" rule, for example) When =3, correspond The (each component);

[0125] The range of values ​​and corresponding The physical value range is consistent, that is .

[0126] During the construction process, according to the high-dimensional calibration matrix The row priority order will be The Line number Column elements ( Mapped to encoding vector The Each component This ensures that each error calibration parameter can be accurately represented by a particle encoding vector, providing a clear parameter carrier for subsequent calculations.

[0127] The parameter control submodule configures dynamic inertia weights and local optimal escape factors to optimize the algorithm's convergence performance. The process of configuring dynamic inertia weights and local optimal escape factors by the parameter control submodule includes the following steps.

[0128] S22.3, Configure dynamic inertia weights The values ​​are dynamically adjusted according to the algorithm iteration process to adapt to the performance requirements of global search and local convergence.

[0129] Specifically, a hybrid strategy of "linear decreasing + adaptive feedback" is used to configure dynamic inertia weights. This enhances the algorithm's global search capability in the early stages of iteration and strengthens its local convergence accuracy in the later stages of iteration. The calculation formula is as follows:

[0130] ;

[0131] Among them, adaptive adjustment amount The calculation formula is:

[0132] ;

[0133] This represents the maximum value of the inertia weight (unitless), ranging from 0.8 to 0.9, ensuring global search capability in the early stages of iteration;

[0134] This represents the minimum inertia weight (unitless), ranging from 0.1 to 0.2, ensuring local convergence accuracy in the later stages of iteration;

[0135] This represents the current iteration number (unitless), with a value range of 1~ ;

[0136] This represents the maximum number of iterations of the algorithm (unitless), which is set according to the solution requirements and ranges from 100 to 500.

[0137] Indicates the first The adaptive adjustment amount (unitless) for each iteration is used to respond to real-time changes in the solution accuracy;

[0138] Indicates the first The global optimal fitness value (unitless) for the next iteration, with the fitness function set as follows: ,in The root mean square error between the solution result and the BeiDou reference data; the solution result refers to the particle encoding. High-dimensional calibration matrix obtained by direct mapping BeiDou reference data refers to the standard calibration matrix pre-stored in the system. This matrix was generated through ground wind tunnel and environmental chamber experiments and calibrations. In the formula For high-dimensional calibration matrix The number of dimensions is consistent with the number of independent dimensions in the original data.

[0139] Indicates the first The global optimal fitness value of the next iteration (unitless);

[0140] Represents a sign function, when When, take 1, when When -1 is taken, Take 0 at that time.

[0141] During the configuration process, in the early stages of iteration ( ), near This enhances the algorithm's ability to search for the global optimum; during the mid-iteration phase ( ), The iteration count decreases linearly with the number of iterations, balancing global search and local convergence; in the later stages of iteration ( ), near It focuses on precise search for local optima;

[0142] At the same time, through Real-time adjustment—if the solution accuracy improves ( If the search accuracy decreases, then the weights are fine-tuned to strengthen the current search direction; If the weight is not adjusted, the weight will be adjusted in the opposite direction to avoid search bias.

[0143] S22.4. Set the trigger condition for the local optimal escape factor, and after triggering, adjust the perturbation amplitude. Adjusting the particle positions helps prevent the algorithm from getting stuck in local optima.

[0144] Specifically, by quantifying the triggering conditions and the magnitude of dynamic perturbations, a local optimal escape factor is constructed to prevent the algorithm from getting trapped in local optima. The specific implementation is as follows:

[0145] Escape factor trigger conditions:

[0146] The algorithm is considered to be trapped in a local optimum and the escape factor is triggered when the following two conditions are met:

[0147] ;

[0148] Indicates the first The local optimal solution (unitless) of the next iteration, i.e. the optimal position searched by a single particle;

[0149] Indicates the first Local optimal solution of the next iteration (unitless);

[0150] This represents the trigger threshold (unitless), with a value range of 10~30, i.e., continuous. Triggered when the local optimum remains unchanged in the next iteration;

[0151] Indicates the first The global optimal fitness value of the next iteration (unitless);

[0152] Indicates the first Local optimal fitness value of the next iteration (unitless);

[0153] This represents the fitness difference threshold (unitless), with a value range of 0.001 to 0.01, which is triggered when the difference between the global and local optimal fitness values ​​is extremely small.

[0154] Perturbation amplitude calculation and particle position adjustment:

[0155] After the escape factor is triggered, the disturbance amplitude is calculated using the following formula. And adjust the particle positions:

[0156] ;

[0157] ;

[0158] Indicates the magnitude of particle position perturbation (unitless);

[0159] This represents the disturbance coefficient (unitless), with a value range of 0.1 to 0.3, used to control the disturbance intensity.

[0160] Indicates the first The first particle Current position in the next iteration (unitless);

[0161] Indicates the first The global optimal solution for the next iteration (unitless);

[0162] This represents a unitless random number with a value range of [-1, 1], used to randomly determine the direction of the disturbance;

[0163] Indicates the first The first particle The adjusted position after the next iteration (unitless).

[0164] During the adjustment process, the amplitude of the disturbance The distance between the particle's current position and the global optimum is positively correlated—the greater the distance, the stronger the perturbation, helping the particle quickly escape the local optimum; the closer the distance, the weaker the perturbation, preventing excessive perturbation from ruining the current search results; simultaneously, the adjusted particle position... Must meet This ensures that the particle's position is always within a physically reasonable range.

[0165] The BeiDou reference constraint submodule dynamically constrains the particle population search boundary using BeiDou reference data to improve solution accuracy. The process of the BeiDou reference constraint submodule dynamically constraining the particle population search boundary using BeiDou reference data includes the following steps.

[0166] S22.5. Based on BeiDou reference data, obtain real-time error correction values ​​and convert them into dynamic adjustment values ​​for the particle swarm search boundary. ;

[0167] Specifically, real-time error correction values ​​are obtained based on BeiDou reference data (meteorological auxiliary data and attitude data), and then converted into dynamic adjustment values ​​for the particle swarm search boundary. The calculation formula is:

[0168] ;

[0169] Indicates the first Dynamic adjustment value of particle population search boundary in the next iteration (unitless).

[0170] The temperature deviation mapping coefficient (unitless) ranges from 0.02 to 0.05 and is determined by fitting the data through a temperature and pressure environment chamber experiment.

[0171] The pressure deviation mapping coefficient (unitless) ranges from 0.01 to 0.03 and is determined by fitting the data through a temperature and pressure environment chamber experiment.

[0172] The value represents the vibration acceleration deviation mapping coefficient (unitless), ranging from 0.03 to 0.06, and is determined by fitting through airborne vibration simulation experiments.

[0173] Indicates the first The airborne ambient temperature deviation (unit: °C) at the next iteration is calculated as follows: ,in Real-time temperature provided by BeiDou reference data. =25℃ is the reference temperature;

[0174] Indicates the first The airborne environmental pressure deviation (unit: hPa) at the next iteration is calculated as follows: ,in Real-time air pressure provided for BeiDou reference data. =1013.25 hPa is the reference pressure;

[0175] Indicates the first The airborne vibration acceleration deviation (in g) at the next iteration is calculated as follows: ,in Real-time vibration acceleration provided for BeiDou reference data. =0g is the reference acceleration.

[0176] During the calculation, the BeiDou reference constraint submodule receives temperature, air pressure, and vibration acceleration data output by the BeiDou satellite in real time, updating it every 10ms. And substitute into the formula to calculate This ensures that the dynamic adjustment values ​​are synchronized with real-time changes in the airborne environment.

[0177] S22.6, Based on the dynamic adjustment value Update particle population search boundary .

[0178] Specifically, based on the dynamically adjusted value Real-time updates of particle swarm search boundaries The calculation formula is:

[0179] ;

[0180] ;

[0181] ;

[0182] Indicates the first The particle swarm search boundary (unitless) for the next iteration is an interval... ;

[0183] Indicates the first Minimum value of the search boundary in the next iteration (unitless);

[0184] Indicates the first Maximum value of the search boundary in the next iteration (unitless);

[0185] Represents the minimum value of the initial search boundary (unitless), based on a high-dimensional calibration matrix. Setting the minimum reasonable value for the parameter;

[0186] Represents the initial search boundary maximum value (unitless), based on a high-dimensional calibration matrix. Setting the maximum reasonable value for the parameter;

[0187] During the update process, the initial search boundary Covering high-dimensional calibration matrix All reasonable values ​​for the parameters; during the iteration process, when the influence of airborne environmental errors increases ( When the error effect weakens, the search boundary expands synchronously to ensure coverage of the possible optimal solution range; when the error effect weakens, When searching, the search boundary shrinks synchronously, improving search efficiency; through dynamic updates, the particle population is always searching within the interval with the highest probability of the optimal solution, balancing solution accuracy and efficiency.

[0188] In this embodiment, the frequency band collaborative solution module 23 divides the frequency bands according to the signal-to-noise ratio of the laser echo, performs synchronous solution on the original data and calibration matrix in the frequency bands, and completes multi-physical quantity coupling calibration, data collaborative inversion and noise reduction by sharing calibration matrix parameters. The process of the frequency band collaborative solution module 23 dividing the frequency bands according to the signal-to-noise ratio of the laser echo and completing the solution by sharing calibration matrix parameters includes the following steps;

[0189] S23.1 Calculate the laser echo signal-to-noise ratio The high-frequency band and low-frequency band are divided according to the signal-to-noise ratio threshold;

[0190] Specifically, laser echo signal-to-noise ratio The calculation process is as follows:

[0191] Laser echo signal-to-noise ratio The formula used to characterize the quality of laser echo signals is:

[0192] ;

[0193] Among them, the mean of background noise The calculation formula is:

[0194] ;

[0195] The peak intensity of the laser echo signal (unit: V) is extracted from the filtered signal output by the laser echo acquisition module 11.

[0196] This represents the number of noise samples (unitless), ranging from 1000 to 10000, ensuring the stability of the noise mean calculation.

[0197] Indicates the first The intensity value (unit: V) of each noise sample is obtained by collecting signal samples when there is no laser echo.

[0198] During the calculation, the frequency-band collaborative solution module 23 extracts the peak intensity of the laser echo signal every 5ms. Simultaneously, signal samples are collected and calculated in real time when there is no laser echo. Substituting into the formula, we obtain the real-time signal-to-noise ratio. .

[0199] Specifically, the frequency band allocation rules are as follows:

[0200] Set signal-to-noise ratio threshold ,according to and The high-frequency and low-frequency bands are divided according to their relative magnitudes, as follows:

[0201] High frequency band: ;

[0202] Low frequency band: ;

[0203] The signal-to-noise ratio threshold (unitless) ranges from 10 to 20. It is based on "Design and Simulation of New Generation Multibeam Cloud Measurement LiDAR on Space" (Chen Weibiao et al., Acta Optica Sinica, 2025). This paper confirms through airborne adaptation simulation and experiments that this range can meet the accuracy requirements of airborne precipitation particle detection and is compatible with frequency band processing logic.

[0204] The classification logic is as follows: high-frequency signals have less noise interference and better quality, making them suitable for precipitation particle spectrum analysis tasks with high accuracy requirements; low-frequency signals are more affected by noise, but the continuity of wind field structure-related information is strong, making them suitable for wind field structure reconstruction tasks with high integrity requirements. During the calibration process, it is necessary to ensure that the noise ratio of high-frequency data is less than 10% and the continuity rate of wind field information in low-frequency data is higher than 90%.

[0205] S23.2 Extracting the high-dimensional calibration matrix generated by the multi-physical quantity calibration model construction module 21 Parameters, constructing cross-band shared parameter weights ;

[0206] Specifically, extract the high-dimensional calibration matrix. Core parameters, constructing cross-band shared parameter weights To achieve parameter sharing and adaptation for the two types of detection missions, the calculation formula is as follows:

[0207] ;

[0208] in:

[0209] The weighting coefficients satisfy the following constraints: ;

[0210] Represents the cross-band shared parameter weight matrix (unitless), dimension, and high-dimensional calibration matrix. Consistency );

[0211] The value represents the weighting coefficient (unitless) of the relevant parameters in the precipitation particle spectrum analysis, with values ​​ranging from 0.7 to 0.9 in the high-frequency band and from 0.1 to 0.3 in the low-frequency band.

[0212] This represents the weighting coefficients (unitless) of parameters related to wind field structure reconstruction, with values ​​ranging from 0.1 to 0.3 for high-frequency bands and from 0.7 to 0.9 for low-frequency bands.

[0213] Represents the precipitation particle spectrum analytical adaptation parameter matrix (unitless), and extracts the high-dimensional calibration matrix. Parameters related to particle size and velocity calibration;

[0214] Represents the wind field structure reconstruction adaptation parameter matrix (unitless), and extracts the high-dimensional calibration matrix. Parameters related to wind field gradient and wind speed calibration.

[0215] During the construction process, the first step is to start with the high-dimensional calibration matrix. Separate from and —— Include Matrix elements with higher weights Include Matrix elements with higher weights; then, based on the current frequency band type, they are allocated... and The value of makes Enhance particle detection calibration parameters in the high-frequency band and wind field detection calibration parameters in the low-frequency band, while sharing... The core error calibration parameters are used to avoid inconsistencies in calibration standards between the two types of tasks.

[0216] S23.3, Weighting of shared parameters across frequency bands and frequency bands Simultaneously complete multi-physical quantity coupling calibration, data collaborative inversion, and noise reduction processing.

[0217] Specifically, based on the division of frequency bands and the sharing of parameter weights across frequency bands. The original data and calibration matrix are solved synchronously to complete multi-physical quantity coupled calibration, data collaborative inversion, and noise reduction. The calculation formula adopts one of the following two serial schemes (selected according to the scenario):

[0218] Firstly, it prioritizes calibration accuracy and is suitable for scenarios where the noise level in the raw data is less than 15%.

[0219] ;

[0220] Secondly, it prioritizes reducing noise interference and is suitable for scenarios where the noise content of the original data is ≥15%.

[0221] ;

[0222] Among them, the noise reduction function The db4 wavelet thresholding denoising algorithm is used, and the specific formula is as follows:

[0223] ;

[0224] Wavelet thresholding The calculation formula is:

[0225] ;

[0226] in:

[0227] Wavelet basis function: Select db4 wavelet (suitable for noise reduction requirements of laser / particle detection signals, with measured signal-to-noise ratio improvement of 12~15dB).

[0228] Wavelet decomposition layers: set to L=4 layers (adjust to L=5 layers when the original data dimension M>15).

[0229] Noise Standard Deviation The formula is calculated by statistically analyzing the "no-signal interval" of the original data. (Use the median method to estimate noise intensity to avoid interference from signal peaks);

[0230] This represents the final data after calibration, inversion, and noise reduction (unitless), with dimensions different from the original data. Consistent;

[0231] This represents the synchronized raw detection data (unitless) output by the intelligent sensing detection unit 1. dimensional vector;

[0232] This represents a noise reduction function (unitless) used to filter out residual noise from the original data.

[0233] Represents raw data A single component (unitless);

[0234] Indicates the wavelet threshold (unitless);

[0235] Indicates the length of the original data (unitless);

[0236] The sign function is used to preserve the positive or negative characteristics of data.

[0237] Furthermore, during the solution process, the high-frequency band and the low-frequency band perform the following operations in parallel:

[0238] High-frequency band solution: Particle characteristic data (such as particle size distribution, velocity data) and ( Multiplying the high proportion of data, the coupled calibration of errors in multiple physical quantities such as wind field gradient and temperature and pressure changes is completed; residual noise in the signal is filtered out by wavelet threshold denoising algorithm; based on the calibrated particle data, the accurate precipitation particle spectrum (including information such as the number of particles and average velocity in each particle size range) is obtained by inversion.

[0239] Low-frequency band solution: Laser echo data (such as echo intensity, round-trip time delay) and ( Multiply the data by the high proportion of the data to complete the calibration of errors such as body vibration and temperature and pressure changes; retain the continuity of wind field data through noise reduction algorithm; and obtain complete wind field structure data (such as wind speed, wind direction, wind field gradient and other information) based on the calibrated echo data.

[0240] Collaborative feedback: Particle velocity data retrieved from the high-frequency band is fed back to the low-frequency band in real time to correct wind field gradient calculation results; wind field structure data retrieved from the low-frequency band is fed back to the high-frequency band in real time to optimize the calibration accuracy of particle terminal velocities; through bidirectional feedback, collaborative inversion of the two types of detection data is achieved, ultimately outputting a result that balances accuracy and completeness. .

[0241] The collaborative control and adjustment unit 3 calls the improved particle swarm algorithm of the particle swarm algorithm data processing unit 2 to construct the corresponding particle swarm algorithm optimization model. Based on the real-time data feedback from the intelligent sensing and detection unit 1, it adjusts the lidar emission frequency and the intelligent sensor sampling period.

[0242] In this embodiment, the collaborative control adjustment unit 3 includes a data feedback acquisition module 31, an optimization model construction module 32, and a parameter adjustment execution module 33, wherein:

[0243] The data feedback acquisition module 31 connects to the environment adaptation synchronization module 13 of the intelligent sensing and detection unit 1 to obtain the synchronized raw detection data output by it as real-time feedback data.

[0244] Specifically, the data feedback acquisition module 31 is used to accurately acquire the real-time detection data of the intelligent sensing detection unit 1, providing high-quality input for subsequent model optimization. The specific implementation is as follows:

[0245] Interface Adaptation and Data Transmission: The data feedback acquisition module 31 is rigidly connected to the environment adaptation synchronization module 13 of the intelligent sensing and detection unit 1 via a high-speed serial bus interface. The interface adopts an anti-electromagnetic interference design and is equipped with a sealed dustproof gasket to adapt to the vibration and temperature and pressure fluctuations of the airborne environment. Data transmission adopts a frame synchronization protocol. Each frame of data includes a frame header, data payload, and checksum. The checksum uses the mature CRC32 algorithm to ensure no data loss and no mistransmission. The transmission rate is set according to the real-time feedback requirements and matches the output rate of the environment adaptation synchronization module 13.

[0246] Data Acquisition Frequency and Filtering: The acquisition frequency of the data feedback acquisition module 31 is consistent with the output frequency of the environment adaptation synchronization module 13 to avoid data asynchrony. During the acquisition process, key feedback parameters are extracted from the synchronized raw detection data, including laser echo signal-to-noise ratio, cloud precipitation particle density, particle size distribution range, and laser echo signal amplitude stability. Simultaneously, the module has a built-in data filtering mechanism to remove abnormal data points that clearly exceed the reasonable detection range, retaining valid data as real-time feedback data.

[0247] Data preprocessing: The filtered real-time feedback data undergoes standardized preprocessing to eliminate dimensional differences between parameters and ensure a balanced weighting of each parameter's influence on subsequent model calculations. The preprocessing process references industry-standardized methods, mapping each parameter to a uniform numerical range. The preprocessed dataset is then transmitted in real-time to the optimization model building module 32 via an internal data bus.

[0248] The optimization model building module 32 calls the improved particle swarm algorithm of the particle swarm algorithm data processing unit 2, and builds the corresponding particle swarm algorithm optimization model based on the real-time feedback data of the data feedback acquisition module 31.

[0249] Specifically, the optimization model building module 32 constructs the corresponding particle swarm optimization model, including the following steps:

[0250] First, the core components of the algorithm are invoked: through the system's internal standardized API interface, the improved core module of the particle swarm algorithm data processing unit 2 is invoked, including the multi-dimensional particle coding submodule, parameter adjustment submodule (dynamic inertia weight, local optimal escape factor) and Beidou reference constraint submodule under the particle swarm algorithm data processing unit 2. The configured logic of the above modules is reused to ensure that the algorithm architecture is completely consistent with the solution rules of the particle swarm algorithm data processing unit 2.

[0251] Then, receive and parse the input data: receive the standardized real-time feedback data transmitted by the data feedback acquisition module 31, specifically including four core parameters: laser echo signal-to-noise ratio, cloud precipitation particle density, particle size distribution span, and laser echo signal amplitude stability. Clarify the two core variables for this optimization: "airborne lidar emission frequency" and "sampling period of the multimodal integrated intelligent sensor in the intelligent sensing and detection unit 1".

[0252] Next, the optimization objectives and constraints are defined: "optimal detection data quality + reasonable equipment energy consumption" are the dual core optimization objectives. Data quality is comprehensively evaluated through quantitative weights (laser echo signal-to-noise ratio weight 0.4, particle detection coverage weight 0.3, signal amplitude stability weight 0.3). Energy consumption is based on the "transmission frequency-power" and "sampling period-power consumption" correlations disclosed in the airborne equipment hardware manual and is controlled within 60%-90% of the equipment's rated power consumption.

[0253] The constraints strictly follow the hardware performance parameter settings: the airborne lidar's emission frequency range is 10kHz-100kHz, which is supported by its hardware, and the intelligent sensor's sampling period range is 1ms-20ms, which is allowed by its hardware response capability.

[0254] Next, the particle coding rules are configured: the multi-dimensional particle coding mechanism of the particle swarm algorithm data processing unit 2 is reused to map the two optimization variables into two-dimensional particle coding vectors. The first component of the coding vector directly corresponds to the emission frequency of the airborne lidar, and its value range is completely consistent with the 10kHz-100kHz constraint. The second component directly corresponds to the sampling period of the smart sensor, and its value range is completely matched with the 1ms-20ms constraint, ensuring the accurate mapping between the coding and the optimization variables.

[0255] Subsequently, the algorithm's operating parameters were configured: the parameter configuration rules of the particle swarm algorithm data processing unit 2 were reused, the particle swarm size was set to 30-50 (to meet the requirements of balancing solution accuracy and efficiency), and the maximum number of iterations was set to 50-100; the dynamic inertia weight was configured according to the strategy of "linear decrease from 0.85 to 0.15 + adaptive adjustment", that is, in the early stage of iteration, the weight is close to 0.85 to strengthen the global search, and in the later stage, it is close to 0.15 to focus on local convergence, while responding to changes in solution accuracy through adaptive adjustment; the trigger condition for the local optimal escape factor was set to "no change in the local optimal solution for 15 consecutive iterations", and after triggering, the perturbation amplitude was dynamically adjusted according to the distance between the particle's current position and the global optimal solution; at the same time, the real-time environmental adjustment value output by the Beidou reference constraint submodule in the particle swarm algorithm data processing unit 2 was connected. , used to dynamically constrain the search boundary of the particle population;

[0256] Next, we define the fitness function and initiate the iterative solution: The fitness function is constructed based on the optimization objective, as shown in the following formula:

[0257] ;

[0258] in, To optimize the variable vector (airborne lidar emission frequency, smart sensor sampling period), The standardized laser echo signal-to-noise ratio, This represents the standardized particle detection coverage. To ensure the stability of the standardized signal amplitude, Standardized equipment energy consumption (values ​​range from 0 to 1, with lower values ​​indicating higher energy consumption). The smaller the value, the better; 0.7 and 0.3 are the target weight coefficients, where: the value of 0.7 is based on the mandatory requirements of the eighth edition of the CIMO Guide for airborne meteorological detection equipment, the calibration data error must meet the measurement standard of '≤0.2℃ when temperature ≥-80℃ and ≤2hPa when air pressure ≥500hPa', accuracy is the core premise of data validity, so it is given the dominant weight (0.7); the value of 0.3 is based on the constraints of airborne equipment engineering practice (the "Airborne Geophysical Flight Technical Specification" requires that data processing should not occupy too much computing power and affect the response of the whole system), and at the same time, it conforms to the conventional weight allocation logic of multi-objective optimization in the field (efficiency index weight 0.2-0.4), to ensure that the computational resource limitations of the airborne environment are adapted to meet the accuracy requirements.

[0259] Based on the parameters and fitness function configured above, and using standardized real-time feedback data as input, the iterative operation of the improved particle swarm optimization algorithm is initiated. The iteration termination condition is set as "reaching the preset maximum number of iterations" or "the change in the global optimal fitness value is less than 0.001 after 10 consecutive iterations". During the operation, the search boundary is updated in real time through the Beidou reference constraint submodule of the particle swarm optimization algorithm data processing unit 2, so that the particle swarm always searches within the range that is suitable for the current airborne environment, thereby improving the environmental adaptability and accuracy of the solution.

[0260] Finally, the optimal solution is output: after the iteration meets the termination condition, the global optimal particle position output by the algorithm is extracted. The two-dimensional coded components corresponding to this position are directly mapped to the "optimal airborne lidar emission frequency" and the "optimal smart sensor sampling period", respectively. As the final calculation result of the particle swarm optimization model, it is transmitted to the parameter adjustment execution module 33.

[0261] The parameter adjustment execution module 33 adjusts the emission frequency of the airborne lidar and the sampling period of the multimodal integrated intelligent sensor in the intelligent sensing and detection unit 1 based on the model calculation results output by the optimization model construction module 32.

[0262] Specifically, the core of the parameter adjustment execution module 33 is to precisely adjust the airborne lidar emission frequency and the intelligent sensor sampling period based on the optimal parameters output by the optimization model, ensuring smooth and stable adjustment. The specific implementation is as follows:

[0263] Signal generation: After receiving the optimal parameters output by the optimization model construction module 32, the parameter adjustment execution module 33 converts them into control signals recognizable by the device. The lidar emission frequency control signal adopts the standard pulse modulation signal format supported by the device, and the sensor sampling period control signal is transmitted through a serial communication protocol compatible with the device. The amplitude and format of the control signals conform to the hardware interface specifications.

[0264] Smooth adjustment mechanism: To avoid equipment instability caused by sudden parameter changes, the module adopts a linear smooth adjustment strategy with a reasonable adjustment step size. Each adjustment amplitude does not exceed the maximum single adjustment range allowed by the equipment hardware. If the difference between the optimal parameter and the current parameter is large, the adjustment is performed gradually in multiple steps, with the interval between each adjustment set to the time required for the equipment to achieve a stable response, until the optimal parameter value is reached.

[0265] Adjustment verification and closed-loop feedback: After parameter adjustment is completed, the parameter adjustment execution module 33 obtains the real-time feedback data after adjustment through the data feedback acquisition module 31 to verify whether the quality of the probe data meets expectations and whether the equipment energy consumption is within a reasonable range. If the expected conditions are not met, the optimization model construction module 32 is triggered to recalculate, generate new optimal parameters, and adjust them again to form closed-loop control; if the conditions are met, the current parameters are maintained, and the next round of feedback acquisition and adjustment cycle begins.

[0266] Hardware protection mechanism: The parameter adjustment execution module 33 has built-in hardware protection logic. During the adjustment process, it monitors the working status of the equipment in real time. If abnormalities such as abnormal laser radar emission power or sensor communication interruption are detected, the adjustment is stopped immediately, the parameters are restored to the stable parameters before adjustment, and an alarm signal is sent to the system main controller to ensure the safe operation of the equipment.

[0267] The integrated data output unit 4 receives the processing results from the particle swarm algorithm data processing unit 2 and combines them with the environmental adaptation information fed back by the intelligent sensor to generate standardized collaborative detection data.

[0268] In this embodiment, the integrated data output unit 4 includes a processing result receiving module 41, an environmental information fusion module 42, and a standardized data generation module 43, wherein:

[0269] The processing result receiving module 41 connects to the frequency band collaborative solution module 23 and receives the results of multi-physical quantity coupling calibration, data collaborative inversion and noise reduction output by the frequency band collaborative solution module 23.

[0270] Specifically, the core function of the processing result receiving module 41 is to stably receive the processing results of the frequency band collaborative solution module 23, ensuring complete and unbiased data transmission. The specific implementation is as follows:

[0271] Interface design: The processing result receiving module 41 is rigidly connected to the frequency band collaborative calculation module 23 through the fiber optic data bus interface. The interface supports full-duplex transmission, adapts to the electromagnetic interference and vibration characteristics of the airborne environment, and the transmission rate matches the output rate of the frequency band collaborative calculation module 23 (outputting a set of processing results every 100ms) to avoid data accumulation or loss.

[0272] Data type and format received: The system explicitly receives three core processing results from the frequency-band collaborative solution module 23: high-frequency precipitation particle spectrum data (including particle number, average velocity, and particle size distribution spectral width for each particle size range), low-frequency wind field structure data (including wind speed, wind direction, and wind field gradient), and full-band noise-reduced laser echo characteristic data (including echo peak intensity, round-trip time delay, and signal-to-noise ratio). Data transmission uses a frame structure format, with each frame containing a frame identifier, data length, data payload, and frame checksum. The checksum uses the CRC32 algorithm to ensure error-free data transmission.

[0273] Data caching and verification: The processing result receiving module 41 has a built-in circular buffer with a capacity of 100 sets of processing results to avoid data overflow. After receiving data, the integrity of the data is first verified by a checksum. If the verification fails, a retransmission request is sent to the frequency band collaborative solution module 23. After the verification passes, the data is stored according to the "timestamp + data type" classification. The timestamp is consistent with the timestamp output by the environment adaptation synchronization module 13 of the intelligent sensing detection unit 1, providing a spatiotemporal reference for subsequent fusion.

[0274] The environmental information fusion module 42 acquires environmental parameter information from the raw detection data output by the intelligent sensing and detection unit 1, and fuses the environmental parameter information with the processing result received by the processing result receiving module 41.

[0275] Specifically, the core function of the environmental information fusion module 42 is to extract environmental parameters and accurately fuse them with the processing results, ensuring that the output data contains complete detection results and environmental background information. The specific implementation is as follows:

[0276] Environmental parameter acquisition: The module connects to the environment adaptation and synchronization module 13 of the intelligent sensing and detection unit 1 through an internal data interface to extract key environmental parameters from the raw detection data, including airborne environmental temperature and pressure data (temperature, air pressure), airframe vibration parameters (vibration acceleration), and BeiDou reference auxiliary data (detection altitude, position coordinates, real-time error correction). Particle detection environment parameters (particle density) are extracted synchronously according to timestamps to ensure spatiotemporal consistency with the processing results.

[0277] Fusion rule design: The fusion strategy of "spatiotemporal alignment + attribute association" is adopted. First, the environmental parameters and processing results are accurately matched based on the timestamp (timestamp error ≤10ms) to ensure that the detection results and environmental parameters correspond in the same spatiotemporal space. Then, the association relationship is established according to the data attributes. For example, temperature and pressure data are associated with wind field structure data, vibration parameters are associated with laser echo characteristic data, and Beidou position coordinates are associated with precipitation particle spectrum data to form a complete data chain of "detection results-environmental background".

[0278] Data fusion verification: After fusion, the data undergoes logical verification to remove contradictory data (such as a large number of large-diameter precipitation particles appearing when the detection altitude is >15,000 meters). Simultaneously, the data reliability level is marked (based on BeiDou error correction). (The smaller the correction amount, the higher the credibility level). Finally, a fused dataset containing detection results, environmental parameters, and credibility level is formed.

[0279] The standardized data generation module 43 generates standardized collaborative detection data based on the fusion results of the environmental information fusion module 42.

[0280] Specifically, the core function of the standardized data generation module 43 is to transform the fused dataset into industry-standardized collaborative detection data, ensuring that the data can be used across platforms. The specific implementation is as follows:

[0281] Standardization Basis and Format: Following aviation meteorological detection industry standards (such as QX / T565-2020 "Laser Drop Spectrum Precipitation Phenomenon Instrument"), two common data formats are available for selection: one is JSON format (suitable for data transmission and software parsing), and the other is binary format (suitable for high-speed storage and hardware reading). The two formats can be selected via a built-in switch in the module.

[0282] Standardized data structure: The data structure is designed according to the categories of "core detection parameters + auxiliary environment parameters + data identification information", and the specific fields include:

[0283] Core detection parameters: key indicators of precipitation particle spectrum (effective particle size, standard deviation of particle size distribution, particle number concentration), core wind field parameters (horizontal wind speed, vertical wind speed, wind direction, wind field gradient), and laser echo normalization parameters (signal-to-noise ratio, normalized echo amplitude).

[0284] Auxiliary environmental parameters: temperature, air pressure, vibration acceleration, detection altitude, and BeiDou position coordinates;

[0285] Data identification information: data generation timestamp, credibility level, data format identifier, and verification code.

[0286] Data standardization processing: The core detection parameters are normalized (using industry-standardized methods to map to standard numerical ranges) to ensure the comparability of detection data from different devices and scenarios; a unified encoding rule is used for text-based identification information (such as data format identifiers) to avoid ambiguity; the final standardized collaborative detection data is output through multiple channels such as Ethernet interface and serial port, and also supports local data storage (compatible with SD cards, solid-state drives and other storage media) to meet the output needs of different application scenarios.

[0287] Those skilled in the art will understand that the process of implementing all or part of the steps of the above embodiments can be carried out by hardware or by a program instructing the relevant hardware.

[0288] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A collaborative system for airborne laser wind-measuring radar and cloud precipitation detection, characterized in that, include: The intelligent sensing and detection unit (1) adopts a multi-modal integrated intelligent sensor to collect the laser echo signal of the airborne laser wind radar and the particle size and velocity distribution information of cloud precipitation particles, and outputs the raw detection data. The particle swarm algorithm data processing unit (2) constructs a multi-physical quantity coupled dynamic calibration model. The multi-physical quantity coupled dynamic calibration model is based on three physical mechanisms: the modulation effect of wind field gradient on the terminal velocity of precipitation particles, the optical axis jitter error caused by body vibration, and the laser wavelength drift caused by temperature and pressure changes. It constructs a high-dimensional calibration matrix and adopts an improved particle swarm algorithm that combines multi-dimensional particle coding, dynamic inertial weight and local optimal escape factor. It dynamically constrains the particle population search boundary through Beidou reference data and performs frequency-band synchronous calculation on the original data and calibration matrix. The frequency band is divided according to the signal-to-noise ratio of the laser echo. The high-frequency band is used for precipitation particle spectrum analysis and the low-frequency band is used for wind field structure reconstruction. By sharing calibration matrix parameters, the multi-physical quantity coupled calibration, data collaborative inversion and noise reduction are completed synchronously. The particle swarm algorithm data processing unit (2) includes an improved particle swarm algorithm operation module (22). The improved particle swarm algorithm operation module (22) includes a particle coding submodule, a parameter control submodule and a Beidou reference constraint submodule, wherein: The particle encoding submodule performs multi-dimensional particle encoding to adapt to the dimensional features of the high-dimensional calibration matrix; The parameter control submodule configures dynamic inertia weights and local optimal escape factors to optimize algorithm convergence performance, including the following steps: S22.3, Configure dynamic inertia weights The values ​​are dynamically adjusted according to the algorithm iteration process to adapt to the performance requirements of global search and local convergence. S22.

4. Set the trigger condition for the local optimal escape factor, and after triggering, adjust the perturbation amplitude. Adjusting particle positions helps prevent the algorithm from getting trapped in local optima. The BeiDou reference constraint submodule dynamically constrains the particle population search boundary using BeiDou reference data to improve solution accuracy, including the following steps: S22.

5. Based on BeiDou reference data, obtain real-time error correction values ​​and convert them into dynamic adjustment values ​​for the particle swarm search boundary. ; S22.6, Based on the dynamic adjustment value Update particle population search boundary ; The collaborative control adjustment unit (3) calls the improved particle swarm algorithm of the particle swarm algorithm data processing unit (2) to construct the corresponding particle swarm algorithm optimization model, and adjusts the laser radar emission frequency and the intelligent sensor sampling period according to the real-time data feedback of the intelligent sensing detection unit (1). The integrated data output unit (4) receives the processing result of the particle swarm algorithm data processing unit (2) and generates standardized collaborative detection data by combining the environmental adaptation information fed back by the intelligent sensor.

2. The airborne laser wind radar and cloud precipitation detection collaborative system according to claim 1, characterized in that, The intelligent sensing and detection unit (1) includes a laser echo acquisition module (11), a cloud precipitation particle sensing module (12), and an environment adaptation and synchronization module (13), wherein: The laser echo acquisition module (11) is connected to the airborne laser wind radar to acquire laser echo signals and perform preliminary signal filtering. The cloud precipitation particle sensing module (12) collects information on the particle size and velocity distribution of cloud precipitation particles and generates particle characteristic data. The environment adaptation synchronization module (13) receives the filtered signal from the laser echo acquisition module (11) and the particle characteristic data from the cloud precipitation particle sensing module (12), completes the time synchronization of the filtered signal and the particle characteristic data, and outputs synchronized raw detection data.

3. The airborne laser wind radar and cloud precipitation detection collaborative system according to claim 2, characterized in that, The particle swarm optimization data processing unit (2) further includes a multi-physical quantity calibration model construction module (21) and a frequency band collaborative solution module (23), wherein: The multi-physical quantity calibration model construction module (21) is based on three major physical mechanisms: the modulation effect of wind field gradient on the terminal velocity of precipitation particles, the optical axis jitter error caused by body vibration, and the laser wavelength drift caused by temperature and pressure changes. It quantifies the error influence coefficient and constructs a high-dimensional calibration matrix. The frequency band collaborative solution module (23) divides the frequency bands according to the signal-to-noise ratio of the laser echo, performs synchronous solution of the original data and calibration matrix in the frequency bands, and completes multi-physical quantity coupling calibration, data collaborative inversion and noise reduction processing by sharing calibration matrix parameters.

4. The airborne laser wind radar and cloud precipitation detection collaborative system according to claim 3, characterized in that, The process of quantifying the error influence coefficients and constructing a high-dimensional calibration matrix by the multi-physical quantity calibration model construction module (21) includes the following steps: S21.1 For the three major physical mechanisms of wind field gradient modulation effect, optical axis jitter due to body vibration, and wavelength drift due to temperature and pressure changes, the corresponding error influence coefficients are quantified respectively. ;in, This represents the error influence coefficient of the wind field gradient modulation effect on the detection data; This represents the error influence coefficient of the optical axis jitter caused by the vibration of the machine body on the detection data; This represents the error influence coefficient of wavelength drift due to temperature and pressure changes on the detection data; S21.2, Based on the error influence coefficient Constructing a high-dimensional calibration matrix The matrix dimension is adapted to the dimension of the original data output by the intelligent sensing and detection unit (1).

5. The airborne laser wind radar and cloud precipitation detection collaborative system according to claim 4, characterized in that, The particle encoding submodule performs multi-dimensional particle encoding, which includes the following steps: S22.1 Determine the particle encoding dimension, and make the particle encoding dimension consistent with the high-dimensional calibration matrix generated by the multi-physical quantity calibration model construction module (21). The dimensions are completely identical; S22.2 Constructing a multi-dimensional particle encoding vector Each encoded component is associated with a high-dimensional calibration matrix. The error influence coefficients in the data correspond one-to-one.

6. The airborne laser wind radar and cloud precipitation detection collaborative system according to claim 5, characterized in that, The frequency band collaborative solution module (23) divides the frequency bands according to the signal-to-noise ratio of the laser echo and completes the solution process by sharing the calibration matrix parameters, including the following steps: S23.1 Calculate the laser echo signal-to-noise ratio The high-frequency band and low-frequency band are divided according to the signal-to-noise ratio threshold; S23.2 Extracting the high-dimensional calibration matrix generated by the multi-physical quantity calibration model construction module (21) Parameters, constructing cross-band shared parameter weights ; S23.3, Weighting of shared parameters across frequency bands and frequency bands Simultaneously complete multi-physical quantity coupling calibration, data collaborative inversion, and noise reduction processing.

7. The airborne laser wind radar and cloud precipitation detection collaborative system according to claim 6, characterized in that, The collaborative control and adjustment unit (3) includes a data feedback acquisition module (31), an optimization model construction module (32), and a parameter adjustment execution module (33), wherein: The data feedback acquisition module (31) connects to the environment adaptation synchronization module (13) of the intelligent sensing and detection unit (1) to obtain the synchronized raw detection data output by it as real-time feedback data. The optimization model construction module (32) calls the improved particle swarm algorithm of the particle swarm algorithm data processing unit (2) and constructs the corresponding particle swarm algorithm optimization model based on the real-time feedback data of the data feedback acquisition module (31). The parameter adjustment execution module (33) adjusts the emission frequency of the airborne lidar and the sampling period of the multimodal integrated smart sensor in the smart sensing and detection unit (1) according to the model calculation results output by the optimization model construction module (32).

8. The airborne laser wind radar and cloud precipitation detection collaborative system according to claim 7, characterized in that, The integrated data output unit (4) includes a processing result receiving module (41), an environmental information fusion module (42), and a standardized data generation module (43), wherein: The processing result receiving module (41) connects to the frequency band collaborative solution module (23) and receives the results of multi-physical quantity coupling calibration, data collaborative inversion and noise reduction output by the frequency band collaborative solution module (23); The environmental information fusion module (42) acquires the environmental parameter information in the original detection data output by the intelligent sensing and detection unit (1), and fuses the environmental parameter information with the processing result received by the processing result receiving module (41). The standardized data generation module (43) generates standardized collaborative detection data based on the fusion results of the environmental information fusion module (42).

Images