Safety warning methods, systems, electronic devices and storage media for stratospheric airships

Abstract

This invention provides a method, system, electronic device, and storage medium for early warning of stratospheric airship safety, relating to the field of airship technology. The method includes: acquiring multidimensional data of the stratospheric airship; inputting the multidimensional data into a digital twin model of the stratospheric airship to obtain first risk assessment data of the stratospheric airship; performing pattern recognition based on the multidimensional data to obtain second risk assessment data of the stratospheric airship; generating an early warning result for the stratospheric airship based on the first and second risk assessment data; and using the early warning result to provide early warning of the stratospheric airship's safety. This approach can reduce the deviation between the risk assessment result and the actual environment, and can provide early warning of sudden risks to the stratospheric airship caused by impending severe weather conditions or the coupling of multiple factors.

Description

Technical Field

[0001] This invention relates to the field of airship technology, and in particular to a method, system, electronic device and storage medium for safety early warning of stratospheric airships. Background Technology

[0002] Airships, high-altitude balloons, and other aerostats have significant advantages such as long endurance, regional loitering, high payload capacity, and high cost-effectiveness. They can remain in the stratosphere for extended periods and have shown important application potential in fields such as communication relay, high-resolution Earth observation, environmental monitoring, and scientific research.

[0003] However, the stratospheric wind field is complex and variable, especially the wind shear and jet stream phenomena during the diurnal cycle, which can cause the aerostat to deviate significantly from its intended position, and even trigger structural overload or attitude instability. Furthermore, changes in solar radiation intensity directly affect the aerostat's thermodynamic equilibrium. In addition, cloud reflection and Earth's infrared radiation also significantly impact the aerostat's thermal equilibrium and energy system. These meteorological factors are often coupled, jointly affecting the aerostat's dynamics, thermodynamics, and energy system, forming complex multiphysics coupling effects.

[0004] Currently, the safety early warning and risk assessment methods for stratospheric airships mainly follow the traditional static assessment model in the aviation field. This leads to risk assessment results that deviate from reality and makes it difficult to provide early warnings of sudden risks to stratospheric airships caused by upcoming severe weather conditions or the coupling of multiple factors. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a method, system, electronic device, and storage medium for safety early warning of stratospheric airships.

[0006] This invention provides a stratospheric airship safety early warning method, comprising: Obtain multidimensional data of the stratospheric airship; The multidimensional data is input into the digital twin model of the stratospheric airship to obtain the first risk assessment data of the stratospheric airship; Based on the multidimensional data, pattern recognition is performed to obtain the second risk assessment data for the stratospheric airship. The early warning result of the stratospheric airship is generated based on the first risk assessment data and the second risk assessment data, and the safety warning of the stratospheric airship is given based on the early warning result.

[0007] According to the present invention, a stratospheric airship safety early warning method is provided, wherein the digital twin model of the stratospheric airship includes: A basic model is used to determine the driving input of the stratospheric airship based on the multidimensional data; A state deduction model is used to determine the system state of the stratospheric airship based on the driving input of the stratospheric airship. The basic model includes: A thermodynamic model is used to simulate the thermodynamic state of the stratospheric airship's capsule based on the multidimensional data, and to determine the net buoyancy input and real-time solar radiation of the stratospheric airship. A propulsion model is used to determine the thrust, torque, and / or load power of the airship based on the multidimensional data and the built-in characteristics of the motor and propeller of the stratospheric airship. The state deduction model includes: A dynamic model is used to perform spatial motion calculations based on the net buoyancy input, the thrust, and / or the torque, and to determine the flight data of the stratospheric airship. An energy model is used to determine the state of charge of the stratospheric airship's battery based on the real-time solar radiation and the load power.

[0008] According to the present invention, a method for early warning of safety of a stratospheric airship, wherein the step of inputting the multidimensional data into a digital twin model of the stratospheric airship to obtain first risk assessment data of the stratospheric airship includes: The multidimensional data is input into the digital twin model of the stratospheric airship to obtain the state prediction value output by the digital twin model of the stratospheric airship. Real-time acquisition of state measurements, and calculation of residual sequences based on the state measurements and the state predictions; The first risk assessment data for the stratospheric airship is obtained based on the residual sequence.

[0009] According to a stratospheric airship safety early warning method provided by the present invention, after calculating the residual sequence based on the state measurement value and the state prediction value, the method further includes: The residual sequence is input into the unknown fault identification model to obtain the hypermodel early warning result output by the unknown fault identification model; The unknown fault identification model is trained based on the sample residual sequence and the sample supermodel early warning results; The unknown fault identification model is used to identify the spatiotemporal patterns of the residual sequence in order to provide a supermodel early warning for the stratospheric airship.

[0010] According to the present invention, a stratospheric airship safety early warning method is provided, wherein the step of performing pattern recognition based on the multidimensional data to obtain second risk assessment data of the stratospheric airship includes: Based on the multidimensional data, the predicted state value of the stratospheric airship under normal mode is determined, and the state measurement value of the stratospheric airship is obtained, so as to determine the fault detection result according to the predicted state value and the state measurement value. If the fault detection result indicates a fault, the multidimensional data is input into the fault mode recognition model to obtain the fault mode output by the fault mode recognition model; wherein, the fault mode recognition model is trained based on multidimensional data of faulty samples with fault mode labels. The second risk assessment data for the stratospheric airship is obtained based on the aforementioned failure modes.

[0011] According to the present invention, a stratospheric airship safety early warning method is provided, wherein the second risk assessment data includes fault detection results, the fault detection results include the probability of fault occurrence; and the supermodel early warning results include a first confidence distribution of at least two risk levels of unknown faults. The step of generating the early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data includes: The first risk assessment data is normalized to obtain the second confidence distribution of the at least two risk levels; The third confidence distribution of the at least two risk levels is obtained based on the probability of failure occurrence; The early warning results for the stratospheric airship are generated based on the first confidence distribution, the second confidence distribution, and the third confidence distribution.

[0012] According to a stratospheric airship safety early warning method provided by the present invention, the step of generating an early warning result for the stratospheric airship based on a first confidence distribution, a second confidence distribution, and a third confidence distribution includes: The uncertainty quality of the first confidence distribution, the second confidence distribution, and the third confidence distribution is determined according to a predefined basic probability allocation function. The joint confidence distribution of the at least two risk levels is synthesized based on the uncertainty quality of the first confidence distribution, the second confidence distribution, and the third confidence distribution; The early warning results for the stratospheric airship are generated based on the joint confidence distribution.

[0013] The present invention also provides a stratospheric airship safety early warning system, comprising: a data acquisition module for acquiring multidimensional data of the stratospheric airship; The first risk assessment module is used to input the multidimensional data into the digital twin model of the stratospheric airship to obtain the first risk assessment data of the stratospheric airship. The second risk assessment module is used to perform pattern recognition based on the multidimensional data to obtain the second risk assessment data of the stratospheric airship. The safety early warning module is used to generate an early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data, and to provide a safety early warning for the stratospheric airship based on the early warning result.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the stratospheric airship safety early warning method as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the stratospheric airship safety early warning method as described above.

[0016] This invention provides a method, system, electronic device, and storage medium for early warning of stratospheric airship safety. It acquires multidimensional data of the stratospheric airship, inputs this data into a digital twin model of the stratospheric airship to obtain first risk assessment data, performs pattern recognition based on the multidimensional data to obtain second risk assessment data, and generates early warning results for the stratospheric airship based on the first and second risk assessment data. This method provides early warning of stratospheric airship safety by combining the advantages of digital twin models and pattern recognition, thereby reducing the deviation between risk assessment results and the actual environment. It also provides early warning of impending severe safety conditions or sudden risks to the stratospheric airship caused by the coupling of multiple factors. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the stratospheric airship safety early warning method provided by the present invention.

[0019] Figure 2 This is a schematic diagram of the framework of the stratospheric airship safety early warning method provided by the present invention.

[0020] Figure 3 This is a schematic diagram of the stratospheric airship safety early warning system provided by the present invention.

[0021] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] Stratospheric aerostats are a new type of near-space vehicle with significant advantages such as long loiter time, large payload capacity, flexible deployment, and relatively low cost, showing broad application prospects in fields such as Earth observation, communication relay, disaster early warning, and national defense reconnaissance. However, their operating environment is extremely unique, requiring them to cope with multiple challenges including low temperature, low pressure, strong radiation, and complex wind fields. As a large-inertia flexible body, the aerostat exhibits complex dynamic response and is extremely sensitive to external disturbances. This not only increases the design difficulty of the control system but also makes it highly dependent on a continuous and stable power supply from the energy system. During long-term missions, any failure in any of these areas—such as fatigue leakage of the airbag material under extreme environments, power fluctuations in the energy system due to changes in illumination, or attitude instability caused by complex wind fields—could directly lead to mission interruption, becoming the main technical bottleneck restricting its safe and reliable operation.

[0024] Currently, risk assessment studies for stratospheric aerostats largely rely on traditional static methods from the aerospace field, such as the risk matrix method. This method assesses the severity and likelihood of risk events based on expert experience or historical data, derives a risk index through product, and classifies risk levels according to a pre-defined matrix. While it has some reference value in the early stages of a project, its inherent limitations are significant: First, the assessment is highly static and subjective, and the results rely heavily on expert experience. Moreover, once the matrix is ​​determined, it cannot be dynamically adjusted according to the real-time status of the airship and the environment. Secondly, the lack of real-time performance and predictability makes it difficult to effectively integrate and process the large amount of high-dimensional sensor data generated when the airship is stationary, resulting in a disconnect between risk assessment and actual operating status, and an inability to predict sudden or cascading risks caused by the coupling of multiple factors. Furthermore, there is insufficient consideration of the multi-physics coupling effect. The risks of airships are essentially the result of the coupling of multiple fields such as aerodynamics, thermodynamics, structure, and energy. However, existing methods usually analyze risk factors in isolation, making it difficult to describe the complex nonlinear relationships between them.

[0025] At the same time, the limitations of a single driving mode further exacerbate the difficulty of the evaluation: While pure model-driven methods are based on prior physical knowledge, it is extremely difficult to build a high-precision physical model of the entire system. Furthermore, the simplification and linearization process introduces errors and cannot capture unmodeled dynamics and novel faults. Pure data-driven methods rely on a large amount of high-quality fault data covering various operating conditions, which is difficult to obtain in practice. Furthermore, pure data processing models lack physical interpretability, and their reliability and generalization ability are questionable.

[0026] Based on the foregoing, the following combines... Figures 1 to 4 The present invention describes a stratospheric airship safety early warning method, system, electronic device, and storage medium.

[0027] Figure 1 This is a flowchart illustrating the stratospheric airship safety early warning method provided by the present invention, as shown below. Figure 1 As shown, the method includes the following steps.

[0028] Step 101: Obtain multidimensional data of the stratospheric airship.

[0029] It should be noted that there are many ways to obtain multidimensional data from stratospheric airships, such as through stratospheric airship sensors, meteorological forecast data sources, and / or sounding data sources. This embodiment does not limit the methods used.

[0030] Stratospheric aerostats may include stratospheric airships, high-altitude balloons, and tethered balloons.

[0031] Multidimensional data of stratospheric airships can include real-time or near-real-time acquired multidimensional data on the airship and its navigation environment, specifically including: The stratospheric airship's sensors measure attitude, position, velocity, vibration, electric field strength, energy data, propulsion data, pressure difference within the capsule at multiple measuring points, skin temperature at multiple measuring points, and helium temperature at multiple measuring points. Specifically, attitude may include pitch angle, roll angle, and yaw angle; position may include the longitude, latitude, and altitude of the stratospheric airship; velocity may be the three-axis velocity of the stratospheric airship; vibration may be the three-axis acceleration of the stratospheric airship; energy data may include the state of charge (SOC), voltage, temperature, load current, and discharge current of the stratospheric airship's lithium battery; propulsion data may include the motor current, speed, and temperature of the stratospheric airship. Meteorological forecast data and / or radiosonde data, including wind speed, wind direction, temperature, solar radiation intensity, atmospheric longwave radiation, Earth's infrared radiation, and cloud-reflected radiation; This includes historical flight data, ground test data, and simulation data; among which, simulation data may include normal flight data generated through simulation models and various fault simulation data.

[0032] Near real-time refers to the acceptable deviation from real-time for stratospheric airship safety early warning.

[0033] Step 102: Input the multidimensional data into the digital twin model of the stratospheric airship to obtain the first risk assessment data of the stratospheric airship.

[0034] Among them, the first risk assessment data refers to a set of dynamic indicators generated by forward extrapolating the future state of the stratospheric airship through a digital twin model, which is used to quantify the deviation between the current health state and the expected health state of the stratospheric airship.

[0035] For example, the multidimensional data may include data describing the current state of the stratospheric airship, forecast data of the stratospheric airship's navigation environment, and control commands of the stratospheric airship. In this way, the digital twin model can obtain the first risk assessment data of the stratospheric airship by coupling and solving the data based on the multidimensional data through the built-in physical mechanism model according to the real physical laws.

[0036] Understandably, the initial risk assessment data for stratospheric airships obtained through digital twin models can yield a set of dynamic indicators with physical basis. This helps to trace the specific physical mechanism model when a stratospheric airship malfunctions, and further link it to specific physical components or environmental factors, providing a basis for subsequent fault tracing and maintenance decisions.

[0037] Furthermore, digital twin models can use future environmental forecast data to perform forward extrapolation, effectively predicting sudden risks and reducing the interference of sudden risks on stratospheric airships.

[0038] Step 103: Perform pattern recognition based on the multidimensional data to obtain the second risk assessment data of the stratospheric airship.

[0039] The second risk assessment data refers to a set of quantitative indicators generated through in-depth mining of multidimensional data using big data analytics, used to identify potential failure modes.

[0040] It should be noted that there are many ways to obtain the second risk assessment data of the stratospheric airship based on the multidimensional data through pattern recognition, such as through machine learning models, spectrum analysis, autoregressive analysis, etc., and this embodiment does not limit this.

[0041] It is understandable that stratospheric airship failures are often the result of the coupling of multiple physical fields, including aerodynamics, heat, structure, and energy, and this coupling relationship is often highly nonlinear. The second risk assessment data of stratospheric airships obtained through data analysis in this step is helpful in uncovering the hidden, nonlinear correlation characteristics between multidimensional data, and in identifying cascading risks caused by complex physical field couplings that cannot be directly calculated by a single physical mechanism model.

[0042] Step 104: Generate an early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data, and conduct a safety early warning for the stratospheric airship based on the early warning result.

[0043] Among them, the early warning results of stratospheric airships refer to standardized and structured early warning products generated based on the first risk assessment data and the second risk assessment data.

[0044] In some embodiments, the early warning results of the stratospheric airship can be pushed to downstream applications via a network in the form of graphical files and / or data interfaces. These downstream applications may include systems for stratospheric airship task scheduling, intelligent trajectory planning, and integrated display.

[0045] It should be noted that there are many ways to generate standardized and structured early warning products from the first risk assessment data and the second risk assessment data, and you can choose according to the actual working conditions. This embodiment does not limit this.

[0046] The stratospheric airship safety early warning method provided in this invention acquires multidimensional data of the stratospheric airship, inputs the multidimensional data into the digital twin model of the stratospheric airship to obtain first risk assessment data of the stratospheric airship, performs pattern recognition based on the multidimensional data to obtain second risk assessment data of the stratospheric airship, and generates an early warning result of the stratospheric airship based on the first and second risk assessment data. By using the early warning result to provide stratospheric airship safety early warning, the method can combine the advantages of digital twin model and pattern recognition, thereby reducing the deviation between risk assessment results and the actual environment. At the same time, it can provide early warning of sudden risks to the stratospheric airship caused by upcoming severe weather conditions or multiple factors.

[0047] Based on the above embodiments, after acquiring multidimensional data of the stratospheric airship and its navigation environment in real time or near real time, the method further includes preprocessing the multidimensional data; wherein, the preprocessing includes cleaning, alignment and feature extraction.

[0048] For example, cleaning the multidimensional data may include data denoising and outlier processing. Specifically, moving average filtering can be used to denoise slowly varying parameters such as temperature, air pressure, and pressure difference; Kalman filtering can be used to denoise combined inertial navigation data; and median filtering can be used to denoise electric fields, currents, and voltages. Outliers in the multidimensional data can be detected and removed using a moving window Z-score method.

[0049] For example, aligning the cleaned multidimensional data may include imputing missing values ​​in the multidimensional data. Specifically, spline interpolation can be used to align data with consecutive missing values, and linear interpolation algorithms can be used to align data with other missing values ​​to obtain time-series data.

[0050] For example, feature extraction from time-series data can yield time-domain features, frequency-domain features, and time-frequency-domain features. Specifically, time-domain features can be extracted from time-series data using methods such as mean, variance, and extreme values. Frequency-domain features can be obtained from vibration data using methods such as FFT to obtain the dominant frequency component. Time-frequency-domain features can be obtained from vibration data using methods such as wavelet transform to obtain the energy distribution of vibration data.

[0051] The energy distribution of vibration data can be used for the analysis of sudden failures such as shocks.

[0052] Based on any of the above embodiments, the digital twin model of the stratospheric airship includes: A basic model is used to determine the driving input of the stratospheric airship based on the multidimensional data; A state deduction model is used to determine the system state of the stratospheric airship based on the drive input of the stratospheric airship.

[0053] Among them, the basic model and state deduction model of the digital twin model can be constructed according to the physical mechanism of the airship, so as to obtain a high-fidelity digital twin model based on the basic model and state deduction model.

[0054] Among them, the driving input refers to the intermediate physical quantity generated by the basic model after solving the multidimensional data, which is used to drive the state inference model for dynamic simulation.

[0055] Here, system state refers to the set of physical quantities that characterize the real-time behavior and internal state of a stratospheric airship at a specific moment, output by the state deduction model after solving based on the driving input. For example, system state may include the predicted state values ​​of the entire domain, such as kinematics, thermodynamics, and energy, at a specific moment.

[0056] Understandably, by dividing the basic model into a state derivation model, this embodiment constructs a logically clear, highly interpretable, multi-physics coupling-supporting, and easily scalable digital twin architecture. This reduces the complexity of the digital twin model itself while providing a clear engineering implementation path for subsequent risk tracing, fault simulation, and system upgrades.

[0057] In some embodiments, the basic model includes a thermodynamic model, which is used to simulate the thermodynamic state of the capsule of the stratospheric airship based on the multidimensional data, and to determine the net buoyancy input and real-time solar radiation of the stratospheric airship.

[0058] Among them, the thermodynamic model, also known as the thermodynamic coupling model, can simulate the thermodynamic state of the capsule during the day-night cycle, provide net buoyancy input for the dynamic model, and provide real-time solar radiation data for the energy model.

[0059] In some embodiments, the gas temperature and internal and external pressure difference can be obtained according to a thermodynamic model, and the static equilibrium of the stratospheric airship can be determined based on the gas temperature and internal and external pressure difference to assess the buoyancy loss risk and structural overpressure risk of the stratospheric airship, providing a basis for online monitoring and early warning of the stratospheric airship.

[0060] In some embodiments, the base model includes a propulsion model for determining the thrust, torque, and / or load power of the airship based on the multidimensional data and the built-in characteristics of the motor and propeller of the stratospheric airship.

[0061] The propulsion model, also known as the propulsion system model, includes the characteristics of the motors and propellers of the built-in stratospheric airship. It is understood that the built-in motor and propeller characteristics of the propulsion models of different digital twin models of stratospheric airships will vary. The specific built-in motor and propeller characteristics can be set according to actual needs; this embodiment does not impose any limitations on this.

[0062] Flight status-related data and control command-related data from multidimensional data can be input into the propulsion model to obtain the thrust and torque output by the propulsion model to the dynamic model, as well as the real-time power output to the energy model, providing a basis for propulsion efficiency assessment and energy scheduling decisions.

[0063] In some embodiments, the state deduction model includes a dynamic model, which is used to perform spatial motion calculations based on the net buoyancy input, the thrust and / or the torque to determine the flight data of the stratospheric airship.

[0064] The dynamic model, also known as the airship dynamic model or the airship six-degree-of-freedom dynamic model, can calculate all the spatial motions of the airship based on the net buoyancy input of the thermodynamic model and the thrust and torque of the propulsion model, and finally transform it into the trajectory and attitude changes of the airship.

[0065] In some embodiments, the accuracy of inertial parameters such as mass and moment of inertia of the dynamic model is greater than a set threshold. It should be noted that the specific value of the set threshold can be set according to actual needs, and this embodiment does not impose further limitations on it.

[0066] Understandably, the accuracy of inertial parameters such as mass and moment of inertia in the dynamic model exceeds a set threshold, ensuring the simulation accuracy of the dynamic model, which serves as the core of motion control and the coupling dynamics hub of the digital twin model. Based on this, the flight stability and maneuverability of the stratospheric airship can be accurately assessed according to its trajectory and attitude changes.

[0067] In some embodiments, the state deduction model includes an energy model, which is used to determine the state of charge of the stratospheric aerostat's battery based on the real-time solar radiation and the load power.

[0068] Among them, the energy model, also known as the energy system model, can output the SOC of the stratospheric airship's battery based on the real-time solar radiation data of the thermodynamic model and the load power requirements of the propulsion model. While monitoring the stratospheric airship's status in real time, it can also provide early warning of energy risks caused by changes / abrupt changes in solar radiation or load overload by proactively assessing the energy supply and demand balance.

[0069] In some embodiments, the digital twin model can be used for real-time early warning. Specifically, real-time state data, environmental forecast data, and control commands from multidimensional data can be sent to the digital twin model of the stratospheric airship to drive the digital twin model to perform forward rolling simulation, generating state prediction values ​​for the future global state, including kinematics, thermodynamics, and energy. These state prediction values ​​are then compared with actual state measurements in real time to calculate and generate a residual sequence. This quantifies the difference between the actual behavior of the stratospheric airship and the cognitive differences of the health model, obtaining the first risk assessment data, which provides a foundation for subsequent real-time fault mode identification and early warning.

[0070] In other embodiments, the digital twin model can be used for fault simulation. Specifically, real-time status data, environmental forecast data, control commands, and fault commands from multidimensional data can be sent to the digital twin model of the stratospheric airship to drive the digital twin model to perform forward rolling simulation, simulating the evolution process of the system under fault conditions, generating simulation data containing fault characteristics, providing a dataset for the training and verification of various subsequent data-driven models, and continuously improving the ability to recognize and warn of unknown risks.

[0071] For example, structural damage such as abrupt changes in mass distribution can be injected into the dynamic model to simulate various abnormal motion states; thermodynamic anomaly data can be generated by setting failure modes of the thermodynamic model such as material aging, thermal insulation failure, or gas leakage; response datasets of the propulsion model under different health states can be constructed by simulating failure modes of the propulsion model such as motor efficiency degradation, propeller damage, or power anomalies; and various energy degradation scenarios of the energy model can be simulated by injecting failure parameters such as battery capacity decay, increased internal resistance, or decreased solar panel efficiency.

[0072] Real-time status data may include the aerostat's attitude, position, velocity, and capsule pressure. Environmental forecast data may include wind speed, wind direction, temperature, and solar radiation intensity for a configurable future time period. Control commands may include fan, exhaust valve, and thrust commands. Fault commands include preset fault parameters.

[0073] One of the configurable future time periods can be 24 hours, 48 ​​hours, etc.

[0074] Based on any of the above embodiments, the step of inputting the multidimensional data into the digital twin model of the stratospheric airship to obtain the first risk assessment data of the stratospheric airship includes: The multidimensional data is input into the digital twin model of the stratospheric airship to obtain the state prediction value output by the digital twin model of the stratospheric airship. Real-time acquisition of state measurements, and calculation of residual sequences based on the state measurements and the state predictions; The first risk assessment data for the stratospheric airship is obtained based on the residual sequence.

[0075] Among them, the state prediction value refers to the set of physical quantities that are output by the digital twin model of the stratospheric airship after being solved according to the driving input, and are used to characterize the estimated real-time behavior and internal state of the stratospheric airship at the current or future specific moment.

[0076] Among them, the state measurement value refers to the set of physical quantities that are collected and acquired in real time by physical sensors deployed on the stratospheric airship, and are used to reflect the real behavior and internal state of the stratospheric airship at the same specific moment.

[0077] The residual sequence is the sequence of differences between the measured state values ​​and the predicted state values. The same specific moment refers to the set of physical quantities representing the behavior and internal state of the stratospheric aerostat at the same time, where the predicted state values ​​and the measured state values ​​are identical.

[0078] For example, the residual sequence R(t) = state measurement(t) - state prediction(t) Understandably, by inputting multidimensional data into a digital twin model to generate state prediction values ​​and comparing them with real-time acquired state measurements to construct residual sequences, it is possible to quantify the difference between the actual behavior of a stratospheric airship and the health status represented by the digital twin model. This provides a quantitative basis for real-time monitoring of the airship's operating status, identifying potential anomalies, and achieving early warning, which helps to detect and intervene in the early stages of risk occurrence and improve the safety and reliability of airship operation.

[0079] Furthermore, compared to methods relying on a single threshold or manual experience, this embodiment generates first risk assessment data based on residual sequences. This allows for a more objective and accurate reflection of the actual health status of the stratospheric airship through continuous and quantitative data comparison. By analyzing the residual sequences, subtle, gradual fault characteristics or sudden fault characteristics can be identified, improving the accuracy of risk assessment.

[0080] In processing the residual sequences to obtain the first risk assessment data, it was realized that when the system is operating normally and the model accuracy is high, the residual sequences appear as white noise sequences with a mean of zero. However, when the system experiences anomalies or faults not accurately described by the physical model, the residual sequences exhibit specific, non-random spatiotemporal patterns. These undescribed anomalies or faults refer to unmodeled dynamics, novel fault modes, etc., while the specific, non-random spatiotemporal patterns refer to trend shifts, periodic fluctuations, and energy mutations in specific frequency bands, etc.

[0081] Based on this, and according to any of the above embodiments, after calculating the residual sequence based on the state measurement value and the state prediction value, the method further includes: The residual sequence is input into the unknown fault identification model to obtain the hypermodel early warning result output by the unknown fault identification model; The unknown fault identification model is trained based on the sample residual sequence and the sample supermodel early warning results; The unknown fault identification model is used to identify the spatiotemporal patterns of the residual sequence in order to provide a supermodel early warning for the stratospheric airship.

[0082] Among them, the supermodel early warning result refers to the judgment signal output after in-depth analysis of the residual sequence based on the unknown fault identification model, which is used to characterize whether the current operating state of the stratospheric airship is an unknown fault that exceeds the description range of the preset physical model.

[0083] A feasible training scheme for an initial unknown fault model may include: The sample residual sequence is input into the initial unknown fault model to obtain the hypermodel warning result output by the initial unknown fault model. Then, based on the hypermodel warning result and the sample hypermodel warning result, the loss function value is calculated. Finally, based on the loss function value, the model parameters of the initial unknown fault model are updated. The above input and calculation processes are iteratively executed until the loss function converges or the preset number of iterations is reached, thus obtaining the unknown fault identification model. The preset number of iterations can be set as needed and is not specifically limited here.

[0084] The initial unknown fault model can be a 1D-CNN model.

[0085] It is understood that the unknown fault identification model in this embodiment can effectively capture abnormal features not described by the physical model, such as trend shifts and frequency band energy mutations, by performing deep temporal pattern recognition on the residual sequence. Based on the spatiotemporal pattern characteristics of the residual sequence under abnormal conditions, it can provide super-model early warning for unknown faults of stratospheric airships, make up for the inherent defects of the risk assessment method based on pure physical models, significantly enhance the ability to perceive new and complex faults, and improve the operational robustness and safety of the system from a data-driven perspective.

[0086] Based on any of the above embodiments, the step of performing pattern recognition based on the multidimensional data to obtain the second risk assessment data of the stratospheric airship includes: Based on the multidimensional data, the predicted state value of the stratospheric airship under normal mode is determined, and the state measurement value of the stratospheric airship is obtained, so as to determine the fault detection result according to the predicted state value and the state measurement value. If the fault detection result indicates a fault, the multidimensional data is input into the fault mode recognition model to obtain the fault mode output by the fault mode recognition model; wherein, the fault mode recognition model is trained based on the faulty second sample multidimensional data with fault mode labels. The second risk assessment data for the stratospheric airship is obtained based on the aforementioned failure modes.

[0087] It should be noted that there are many ways to determine the predicted state of the stratospheric airship in normal mode based on the multidimensional data, such as through machine learning models, built-in mapping tables, etc., and this embodiment does not limit this method.

[0088] For example, an initial prediction model can be trained unsupervised based on unlabeled, fault-free sample multidimensional data to obtain a state prediction model. In this way, the state prediction model can analyze the received multidimensional data based on the learned dynamic patterns of the stratospheric airship's multidimensional data under normal operating conditions, and output the predicted state value of the stratospheric airship under normal operating conditions.

[0089] Specifically, multidimensional data can be input into the state prediction model. The state prediction model autonomously selects the corresponding first sub-layer to process the speed, drive current, and temperature data of each motor, thereby achieving state prediction of the stratospheric airship's motors. The state prediction model autonomously selects the corresponding second sub-layer to process the total voltage, total current, SOC, temperature, bus voltage, and solar panel output power of the energy storage battery, thereby achieving state prediction of the stratospheric airship's energy system. Alternatively, a first state prediction model can be pre-trained based on sample speed, sample drive current, sample temperature data of each motor and state sample values ​​under normal mode, and a second state prediction model can be trained based on sample total voltage, sample total current, sample SOC, sample temperature, sample bus voltage, sample solar panel output power, and state sample values ​​under normal mode of the energy storage battery, etc.

[0090] It should be noted that the prediction error can be calculated and determined based on the predicted and measured state values. When the prediction error increases significantly, it can be determined that the system behavior of the stratospheric airship deviates from the normal mode, the fault detection result is a fault, and an overall abnormal alarm is triggered. When the prediction error is within a preset range, it can be determined that the system behavior of the stratospheric airship is in the normal mode, and the fault detection result is no fault.

[0091] The training process for the fault pattern recognition model is basically the same as that for the unknown fault recognition model, and will not be repeated here. The initial fault recognition model can be a 1D-CNN.

[0092] Failure modes may include decreased thruster efficiency, partial shading / damage to solar panels, minor skin damage / leakage, and sensor drift / deviation.

[0093] It should be noted that the fault data obtained by using real fault data of stratospheric airships, also known as historical fault data, and / or fault data obtained by fault simulation using digital twin models, can be used to obtain multidimensional data of faulty samples with fault mode labels.

[0094] In some embodiments, the fault data obtained from fault simulation based on the digital twin model mainly consists of multidimensional data of faulty samples with fault mode labels.

[0095] In some embodiments, the remaining lifespan of the energy storage battery can also be predicted based on a hybrid network of 1D-CNN and LSTM. For example, 1D-CNN can be used to extract local features from real-time data of the energy storage battery, these local features can be input into an LSTM to capture long-term dependencies, and the predicted remaining lifespan of the energy storage battery can be output through a fully connected layer.

[0096] Based on any of the above embodiments, the second risk assessment data includes fault detection results, which include the probability of fault occurrence; the supermodel early warning results include a first confidence distribution of at least two risk levels of unknown faults.

[0097] For example, the predicted error can be calculated as an anomaly score based on the predicted state value and the measured state value. Alternatively, the predicted error can be converted into a scalar value using a pre-trained Gaussian mixture model to obtain the anomaly score. The fault mode recognition model output can include various fault modes and their probability distributions. The probability of fault occurrence can be obtained based on the anomaly score, various fault modes, and their probability distributions, thus yielding the second risk assessment data for the stratospheric aerostat.

[0098] Similar to the fault mode recognition model, the unknown fault recognition model can output the first confidence distribution of the existence of unknown faults and at least two risk levels, thus obtaining the supermodel early warning result.

[0099] Specifically, at least two risk levels include each risk level in a preset risk level set. For example, the preset risk level set may include four risk levels: no risk, low risk, medium risk, and high risk.

[0100] In some embodiments, generating the early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data includes: The first risk assessment data is normalized to obtain the second confidence distribution of the at least two risk levels; The third confidence distribution of the at least two risk levels is obtained based on the probability of failure occurrence; The early warning results for the stratospheric airship are generated based on the first confidence distribution, the second confidence distribution, and the third confidence distribution.

[0101] For example, when the first risk assessment data includes the prediction error between the predicted state value and the measured state value, the prediction error can be normalized to obtain a second confidence distribution with at least two risk levels. A third confidence distribution with at least two risk levels can be obtained by combining the aforementioned anomaly scores and various failure modes and their probability distributions.

[0102] Understandably, by normalizing the primary risk assessment data into a standardized confidence distribution, data of different types and scales can be comprehensively evaluated within the same framework, resulting in more robust and accurate early warning results.

[0103] Based on any of the above embodiments, generating the early warning result for the stratospheric airship according to the first confidence distribution, the second confidence distribution, and the third confidence distribution includes: The uncertainty quality of the first confidence distribution, the second confidence distribution, and the third confidence distribution is determined according to a predefined basic probability allocation function. The joint confidence distribution of the at least two risk levels is synthesized based on the uncertainty quality of the first confidence distribution, the second confidence distribution, and the third confidence distribution; The early warning results for the stratospheric airship are generated based on the joint confidence distribution.

[0104] For example, by defining the Basic Probability Assignment (BPA) function and modeling the first confidence distribution, the second confidence distribution, and the third confidence distribution respectively, the Dempster combination rule can be used for synthesis to obtain the joint confidence distribution of each risk level in at least two risk levels.

[0105] Based on this, the comprehensive risk index of the stratospheric airship can be calculated according to the corresponding utility value pre-assigned to each of the at least two risk levels, and the early warning result of the stratospheric airship can be generated based on the comprehensive risk index.

[0106] For example, the utility value for no risk can be 0, the utility value for low risk can be 0.3, the utility value for medium risk can be 0.7, and the utility value for high risk can be 1.0.

[0107] For example, the comprehensive risk index of a stratospheric airship can be calculated using the following formula: Overall risk index = ∑(joint confidence level of risk level × utility value of risk level) Understandably, by predefining the basic probability allocation function to determine the uncertainty quality of the first, second, and third confidence distributions respectively, the unreliability of each information source, such as sensor noise, model error, and ambiguity of unknown faults, can be explicitly modeled as weights. This effectively avoids the bias that may be introduced by simple weighted averaging or voting fusion, allowing the subsequent synthesis process to reasonably weigh the credibility of each source, thus generating a robust joint confidence distribution even in the event of conflicting or incomplete risk assessment data.

[0108] Figure 2This is a schematic diagram of the architecture of the stratospheric airship safety early warning method provided by the present invention, as shown below. Figure 2 As shown, in order to illustrate the function of the stratospheric airship safety early warning method provided in this embodiment, a specific example is provided below.

[0109] First, multidimensional data of the stratospheric airship can be acquired and preprocessed; the preprocessing includes cleaning, alignment and feature extraction.

[0110] The preprocessed multidimensional data is then input into the digital twin model and the AI ​​analysis engine, respectively. The digital twin model can predict the state of the stratospheric airship under real-time early warning and can also receive fault injections and generate simulation data containing fault characteristics under fault simulation. The AI ​​analysis engine can determine the state prediction value of the stratospheric airship under normal mode based on the multi-dimensional data, and perform anomaly detection to determine the fault detection result based on the state prediction value and the state measurement value of the stratospheric airship. When the fault detection result is that there is a fault, the fault can be classified according to the fault mode recognition model, output the specific fault mode, and perform fault diagnosis by combining the fault detection result and the fault mode.

[0111] Among them, the fault data obtained by fault simulation using real fault data of stratospheric airships, also known as historical fault data, and / or digital twin models, can be used to obtain multidimensional data of faulty samples with fault mode labels; the fault mode recognition model is obtained by training the model based on the multidimensional data of faulty samples with fault mode labels.

[0112] After obtaining simulation data containing fault characteristics based on the digital twin model, a residual sequence can be calculated. The residual sequence is then input into the unknown fault identification model to obtain the supermodel early warning result output by the unknown fault identification model, and residual pattern recognition is performed.

[0113] The first risk assessment data is normalized to obtain the second confidence distribution of the at least two risk levels; the third confidence distribution of the at least two risk levels is obtained based on the failure occurrence probability; the early warning result of the stratospheric airship is generated by combining the first confidence distribution of the at least two risk levels of the unknown failure in the supermodel early warning result, and the confidence fusion processing is performed to obtain the joint confidence, and the comprehensive risk index of the stratospheric airship is calculated based on the joint confidence.

[0114] In some embodiments, the comprehensive risk index of the stratospheric airship during each stage of flight, such as takeoff, cruise, hovering, and maneuvering, can be calculated in advance based on historical data. An initial static threshold is then set for each stage based on the distribution of the comprehensive risk index. Furthermore, a sliding window can be used to calculate the mean and variance of the recent comprehensive risk index, and the threshold can be dynamically adjusted based on the mean and variance. Specifically, if the recent comprehensive risk index fluctuates significantly, the threshold can be appropriately increased to avoid frequent false alarms; if the recent comprehensive risk index fluctuates less and the risk is lower, the threshold can be appropriately decreased to increase sensitivity.

[0115] Then, based on factors such as flight phase, environmental severity, system health status, and mission criticality level, the initial static threshold for each phase is adaptively adjusted to obtain the phase threshold, thus completing the dynamic threshold judgment.

[0116] Furthermore, early warning levels can be classified based on a comparison of the comprehensive risk index and dynamic thresholds, providing a foundation for tiered early warning systems. Simultaneously, decision-making suggestions can be generated based on a preset rule base, incorporating risk source location information. These suggestions include specific countermeasures such as course adjustments, achieving closed-loop intelligent management of perception, assessment, early warning, and decision-making.

[0117] The warning levels can be classified into four levels. Specifically, Level 1 can be a comprehensive risk index below the stage threshold T1, indicating that the system is operating smoothly; Level 2 can be a comprehensive risk index between the stage thresholds T2 and T3, indicating that there are parameter deviations or slight anomalies, and close monitoring is recommended; Level 3 can be a comprehensive risk index between the stage thresholds T2 and T3, indicating that key parameters have deteriorated or residuals are significantly abnormal, predicting that the future state will exceed the limit, and intervention is recommended; Level 4 can be a comprehensive risk index above the stage threshold T3, indicating that a serious failure is about to occur or has already occurred, and safety strategies such as forced landing and jettisoning are recommended to be implemented immediately.

[0118] The stratospheric airship safety early warning device provided by the present invention is described below. The stratospheric airship safety early warning device described below and the stratospheric airship safety early warning method described above can be referred to in correspondence.

[0119] Figure 3 This is a schematic diagram of the structure of the stratospheric airship safety early warning device provided by the present invention, as shown below. Figure 3 As shown, the device includes: Data acquisition module 310 acquires multidimensional data of the stratospheric airship; The first risk assessment module 320 is used to input the multidimensional data into the digital twin model of the stratospheric airship to obtain the first risk assessment data of the stratospheric airship. The second risk assessment module 330 is used to perform pattern recognition based on the multidimensional data to obtain the second risk assessment data of the stratospheric airship. The safety early warning module 340 is used to generate an early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data, and to provide a safety early warning for the stratospheric airship based on the early warning result.

[0120] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a stratospheric airship safety early warning method. The method includes: acquiring multidimensional data of the stratospheric airship; inputting the multidimensional data into a digital twin model of the stratospheric airship to obtain first risk assessment data of the stratospheric airship; performing pattern recognition based on the multidimensional data to obtain second risk assessment data of the stratospheric airship; generating an early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data; and conducting a safety early warning for the stratospheric airship based on the early warning result.

[0121] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the stratospheric airship safety early warning method provided by the above methods. The method includes: acquiring multidimensional data of the stratospheric airship; inputting the multidimensional data into a digital twin model of the stratospheric airship to obtain first risk assessment data of the stratospheric airship; performing pattern recognition based on the multidimensional data to obtain second risk assessment data of the stratospheric airship; generating an early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data; and providing a safety early warning for the stratospheric airship based on the early warning result.

[0123] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a stratospheric airship safety early warning method provided by the methods described above. This method includes: acquiring multidimensional data of the stratospheric airship; inputting the multidimensional data into a digital twin model of the stratospheric airship to obtain first risk assessment data of the stratospheric airship; performing pattern recognition based on the multidimensional data to obtain second risk assessment data of the stratospheric airship; generating an early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data; and providing a safety early warning for the stratospheric airship based on the early warning result.

[0124] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0125] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A stratospheric float safety warning method, characterized by, include: Obtain multidimensional data of the stratospheric airship; The multidimensional data is input into the digital twin model of the stratospheric airship to obtain the first risk assessment data of the stratospheric airship; Based on the multidimensional data, pattern recognition is performed to obtain the second risk assessment data for the stratospheric airship. The early warning result of the stratospheric airship is generated based on the first risk assessment data and the second risk assessment data, and the safety warning of the stratospheric airship is given based on the early warning result.

2. The stratospheric floatation vehicle safety warning method of claim 1, wherein, The digital twin model of the stratospheric airship includes: A basic model is used to determine the driving input of the stratospheric airship based on the multidimensional data; A state deduction model is used to determine the system state of the stratospheric airship based on the driving input of the stratospheric airship. The basic model includes: A thermodynamic model is used to simulate the thermodynamic state of the stratospheric airship's capsule based on the multidimensional data, and to determine the net buoyancy input and real-time solar radiation of the stratospheric airship. A propulsion model is used to determine the thrust, torque, and / or load power of the airship based on the multidimensional data and the built-in characteristics of the motor and propeller of the stratospheric airship. The state deduction model includes: A dynamic model is used to perform spatial motion calculations based on the net buoyancy input, the thrust, and / or the torque, and to determine the flight data of the stratospheric airship. An energy model is used to determine the state of charge of the stratospheric airship's battery based on the real-time solar radiation and the load power.

3. The stratospheric floatation vehicle safety warning method of claim 1, wherein, The process of inputting the multidimensional data into the digital twin model of the stratospheric airship to obtain the first risk assessment data of the stratospheric airship includes: The multidimensional data is input into the digital twin model of the stratospheric airship to obtain the state prediction value output by the digital twin model of the stratospheric airship. Real-time acquisition of state measurements, and calculation of residual sequences based on the state measurements and the state predictions; The first risk assessment data for the stratospheric airship is obtained based on the residual sequence.

4. The stratospheric floatation vehicle safety warning method of claim 3, wherein, After calculating the residual sequence based on the state measurements and the state predictions, the method further includes: The residual sequence is input into the unknown fault identification model to obtain the hypermodel early warning result output by the unknown fault identification model; The unknown fault identification model is trained based on the sample residual sequence and the sample supermodel early warning results; The unknown fault identification model is used to identify the spatiotemporal patterns of the residual sequence in order to provide a supermodel early warning for the stratospheric airship.

5. The stratospheric floatation vehicle safety warning method of claim 1, wherein, The second risk assessment data for the stratospheric airship, obtained by pattern recognition based on the multidimensional data, includes: Based on the multidimensional data, the predicted state value of the stratospheric airship under normal mode is determined, and the state measurement value of the stratospheric airship is obtained, so as to determine the fault detection result according to the predicted state value and the state measurement value. If the fault detection result indicates a fault, the multidimensional data is input into the fault mode recognition model to obtain the fault mode output by the fault mode recognition model; wherein, the fault mode recognition model is trained based on multidimensional data of faulty samples with fault mode labels. The second risk assessment data for the stratospheric airship is obtained based on the aforementioned failure modes.

6. The stratospheric floatation vehicle safety warning method of claim 1, wherein, The second risk assessment data includes fault detection results, which include the probability of fault occurrence; The hypermodel early warning results include the first confidence distribution of at least two risk levels for unknown faults; The step of generating the early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data includes: The first risk assessment data is normalized to obtain the second confidence distribution of the at least two risk levels; The third confidence distribution of the at least two risk levels is obtained based on the probability of failure occurrence; The early warning results for the stratospheric airship are generated based on the first confidence distribution, the second confidence distribution, and the third confidence distribution.

7. The stratospheric floatation vehicle safety warning method of claim 6, wherein, The step of generating the early warning result for the stratospheric airship based on the first confidence distribution, the second confidence distribution, and the third confidence distribution includes: The uncertainty quality of the first confidence distribution, the second confidence distribution, and the third confidence distribution is determined according to a predefined basic probability allocation function. The joint confidence distribution of the at least two risk levels is synthesized based on the uncertainty quality of the first confidence distribution, the second confidence distribution, and the third confidence distribution; The early warning results for the stratospheric airship are generated based on the joint confidence distribution.

8. A stratospheric float safety warning system characterized by, include: The data acquisition module acquires multidimensional data of the stratospheric airship; The first risk assessment module is used to input the multidimensional data into the digital twin model of the stratospheric airship to obtain the first risk assessment data of the stratospheric airship. The second risk assessment module is used to perform pattern recognition based on the multidimensional data to obtain the second risk assessment data of the stratospheric airship. The safety early warning module is used to generate an early warning result for the stratospheric airship based on the first risk assessment data and the second risk assessment data, and to provide a safety early warning for the stratospheric airship based on the early warning result.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the stratospheric airship safety early warning method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by the processor, it implements the stratospheric airship safety early warning method as described in any one of claims 1 to 7.

Images