Unmanned aerial vehicle take-off and landing field predictive maintenance method and system based on health state driving

Abstract

The invention belongs to the technical field of low-altitude economy, and provides an unmanned aerial vehicle take-off and landing field predictive maintenance method based on health state driving, and the method comprises the steps: collecting the operation data of a take-off and landing platform, a new energy supply facility and a communication navigation facility in real time through a multi-source sensor, and synchronizing the operation data to a digital twin model; performing security risk assessment on each facility based on a plurality of preset risk identification models, calculating sub-health degrees, and obtaining a comprehensive health degree index through dynamic weighted fusion; fitting a life cycle curve for the historical health degree data by using a polynomial regression model, and predicting a future health state; and according to a comparison result of a predicted value and a multi-level alarm threshold value, automatically triggering graded early warning and scheduling operation and maintenance resources to execute maintenance operation, thereby realizing full-closed-loop intelligent operation and maintenance from perception, evaluation and prediction to autonomous restoration. According to the method, the change from passive response to active prediction of take-off and landing field operation and maintenance is realized, the operation safety and the operation and maintenance efficiency are remarkably improved, and the accident rate and manual intervention dependence are reduced.

Description

Technical Field

[0001] This application belongs to the field of low-altitude economic technology, and in particular relates to a predictive maintenance method, system, terminal equipment and storage medium for UAV take-off and landing sites based on health status. Background Technology

[0002] Low-altitude drone take-off and landing sites, as the cornerstone of the low-altitude economy, are similar to stations and ports in ground transportation. They are the physical infrastructure for low-altitude flight activities and are crucial to the development of the low-altitude economy. The healthy and stable operation of drone take-off and landing site infrastructure is fundamental to the safe operation of the site. However, the health of this infrastructure is often affected by various factors such as severe weather conditions, satellite signal interference, and structural aging and damage, leading to challenges in take-off and landing efficiency and safety. For example, extreme weather conditions such as heavy rain and lightning can severely impact drone take-off and landing, even causing safety accidents; human-caused or intentional interference with satellite navigation signals can lead to drone crashes during take-off and landing; battery thermal runaway during charging of new energy vehicles can cause violent combustion or even explosion; damaged or uneven surfaces on the take-off and landing platform can cause drones to sway, tip over, or even overturn during take-off or landing. Currently, the maintenance of existing drone take-off and landing sites mainly relies on scheduled regular inspections or reactive repairs after malfunctions occur.

[0003] However, existing methods suffer from the problem that the operation and maintenance mode of take-off and landing sites remains stuck in the post-event processing and passive response stage, failing to achieve predictive maintenance and proactive prevention, resulting in low overall operational efficiency and management level of UAV take-off and landing sites. Summary of the Invention

[0004] This application provides a method, system, terminal device, and storage medium for predictive maintenance of UAV take-off and landing sites based on health status. It can solve the problem that existing methods are stuck in the post-event processing and passive response stage of take-off and landing site operation and maintenance, and cannot realize predictive maintenance and proactive prevention, resulting in low overall operational efficiency and management level of UAV take-off and landing sites.

[0005] In a first aspect, embodiments of this application provide a predictive maintenance method for UAV take-off and landing sites based on health status, comprising: S1, collecting real-time operational status data corresponding to each infrastructure, and driving a digital twin model to be updated synchronously based on the real-time operational status data; the infrastructure includes UAV take-off and landing platforms, new energy supply facilities, and communication and navigation facilities; S2, based on the real-time operational status data, using multiple preset risk identification models to identify safety risks of each infrastructure respectively, obtaining multiple risk assessment results, and calculating multiple sub-health levels based on the multiple risk assessment results; S3, based on the multiple sub-health levels, using dynamic... S4. Based on historical comprehensive health data sequences, a multinomial regression model is used to fit the life cycle curve of the health status of the UAV take-off and landing site, and the predicted comprehensive health index is obtained according to the life cycle curve; S5. The predicted comprehensive health index is compared with the preset multi-level alarm threshold. When the prediction result reaches the corresponding threshold, the corresponding level of early warning mechanism is automatically triggered, and the operation and maintenance execution resources in the take-off and landing site are scheduled to perform maintenance or emergency response operations according to the predefined operation and maintenance strategy. At the same time, the execution result is fed back to the digital twin model to realize predictive maintenance of the take-off and landing site.

[0006] In one possible implementation of the first aspect, S1, collecting real-time operational status data corresponding to each infrastructure, and driving the digital twin model to be synchronously updated based on the real-time operational status data, specifically includes:

[0007] By collecting surface status data of the UAV take-off and landing platform, operational parameter data of the new energy supply facility, and communication quality data of the communication and navigation facility through multi-source sensors deployed in the UAV take-off and landing site, real-time operational status data of each infrastructure is formed.

[0008] Outlier detection, missing data filling, and data format standardization are performed on real-time operating status data to obtain standardized status data for subsequent analysis.

[0009] The digital twin model is updated by synchronizing data based on standardized state data. A periodic synchronization strategy is used for normal state data, and a priority synchronization strategy is used for abnormal state data to improve the timeliness of response to abnormal states in the digital twin model.

[0010] The synchronized updated digital twin model will be used as a virtual mapping of the physical entities of the UAV take-off and landing site for subsequent health status assessment, prediction and operation and maintenance decisions.

[0011] Optionally, in another possible implementation of the first aspect, the preset risk identification model is an image semantic segmentation model based on a fully convolutional neural network; S2 above, based on real-time operational status data, uses multiple preset risk identification models to identify security risks of each infrastructure, obtains multiple risk assessment results, and calculates multiple sub-health scores based on the multiple risk assessment results, specifically including:

[0012] Real-time image data of the UAV take-off and landing platform is acquired by optical cameras and thermal imaging cameras, and the real-time image data is input into an image semantic segmentation model based on a fully convolutional neural network.

[0013] The image semantic segmentation model is used to perform pixel-level semantic segmentation on real-time image data to identify the percentage of water, snow and ice areas on the drone take-off and landing platform, i.e., the coverage rate, and to identify the length, width and location of cracks on the drone take-off and landing platform.

[0014] Based on coverage, the primary risk assessment result is determined to be the health of the drone take-off and landing platform under slippery conditions. ;

[0015] Based on the three preset ranges corresponding to the crack width, the two classification regions corresponding to the crack location, and the three preset ranges corresponding to the crack length, the second risk assessment result is determined as the health status of the UAV take-off and landing platform under damage conditions, according to the crack length, crack width, and crack location of the UAV take-off and landing platform. ;

[0016] Based on health Health And with preset weights, a weighted fusion algorithm is used to determine the sub-health of the drone take-off and landing platform. .

[0017] Optionally, in another possible implementation of the first aspect, the preset risk identification model is a CNN-BiLSTM neural network model; S2 above, based on real-time operational status data, uses multiple preset risk identification models to identify security risks of each infrastructure, obtains multiple risk assessment results, and calculates multiple sub-health scores based on the multiple risk assessment results, specifically including:

[0018] Collect real-time operating status data of new energy supply facilities. The real-time operating status data includes at least temperature parameters, voltage parameters, current parameters, and combustible gas concentration parameters, and construct the real-time operating status data into time series data.

[0019] The time series data is input into the CNN-BiLSTM neural network model, which outputs a risk assessment result, namely the probability of the accident occurring, specifically:

[0020] Time series data is input into a CNN network to extract features from the time series data in order to obtain the spatial correlation features between various operating parameters;

[0021] Spatial correlation features are input into a BiLSTM network, and forward and backward long short-term memory networks are used to perform temporal modeling of spatial correlation features in order to extract time-related features of the operating status of new energy supply facilities.

[0022] Input time-related features into a fully connected layer and output the probability of an accident.

[0023] Mapping the probability of accidents to the sub-health of renewable energy supply facilities .

[0024] Optionally, in another possible implementation of the first aspect, the preset risk identification model is a BP deep learning neural network; S2 above, based on real-time operational status data, uses multiple preset risk identification models to identify security risks of each infrastructure, obtains multiple risk assessment results, and calculates multiple sub-health scores based on the multiple risk assessment results, specifically including:

[0025] Collect real-time operational status data of communication and navigation facilities. The real-time operational status data shall include at least the signal strength, signal-to-noise ratio, number of tracked satellites, transmission delay, data packet loss rate and bit error rate parameters of satellite communication, 5G-A communication and ADS-B communication.

[0026] The real-time operational status data of communication and navigation facilities is normalized and then used to construct a feature vector, which is then input into a trained BP deep learning neural network. The trained BP deep learning neural network is trained based on the monitoring status data of communication and navigation facilities.

[0027] The trained BP deep learning neural network is used to perform nonlinear mapping on the feature vectors, and the risk assessment results of communication and navigation facilities are output.

[0028] Based on the risk assessment results of communication and navigation facilities, the operational status of communication and navigation facilities is mapped to the sub-health of communication and navigation facilities. .

[0029] Optionally, in another possible implementation of the first aspect, S3 above, based on multiple sub-health levels, uses a dynamic weighted fusion algorithm to calculate the comprehensive health index of the UAV take-off and landing field, specifically including:

[0030] Based on the sub-health of the drone take-off and landing platform Sub-health of new energy supply facilities and the sub-health of communication and navigation facilities Weighted calculations were performed to obtain the comprehensive health index H of the drone take-off and landing site:

[0031]

[0032] in, , , They are respectively , as well as The corresponding dynamic weighting coefficients; , , For the first The dynamic weighting coefficients corresponding to the health of each infrastructure sub-item for Weighting coefficients in the dimension of risk severity. for The weighting coefficient for the facility's impact range, , , ;in, The calculation formula is:

[0033]

[0034] in, The assignment relationship is as follows: when the influence range of the drone take-off and landing platform is to affect all drones, The value is 0.5; when the impact range of the new energy supply facility is limited to affecting new energy drones, The value is 0.3; when the impact range of communication and navigation facilities is limited to affecting communication but can be switched to a backup link, It is 0.2.

[0035] Optionally, in another possible implementation of the first aspect, S4 above, based on historical comprehensive health data sequences, uses a multinomial regression model to fit the lifecycle curve of the health status of the UAV take-off and landing field, and obtains the predicted comprehensive health index based on the lifecycle curve, specifically including:

[0036] Obtain historical comprehensive health data sequences of drone take-off and landing sites And normalize the historical comprehensive health data sequence;

[0037] Based on the normalized historical comprehensive health data series, a multinomial regression model is constructed:

[0038]

[0039] in, It is the first Health prediction values ​​at each time point These are the regression coefficients;

[0040] Define a loss function to quantify the difference between the predicted and actual health scores:

[0041]

[0042] in, This represents the actual health level.

[0043] Using the least squares method, with the goal of minimizing the loss function, we find the parameter combination that minimizes the squared error loss through matrix operations. :

[0044]

[0045] in, , It is a matrix The transpose of the matrix, It is a matrix The inverse matrix, It is a comprehensive health index of the drone take-off and landing site. ;

[0046] Based on the combination of substituted parameters The multinomial regression model is used to obtain the life cycle curve, and the predicted comprehensive health index is obtained based on the life cycle curve.

[0047] Secondly, embodiments of this application provide a health-state-driven predictive maintenance system for UAV take-off and landing sites, comprising: a status perception module, used to collect real-time operational status data corresponding to each infrastructure, and drive a digital twin model to be updated synchronously based on the real-time operational status data; the infrastructure includes UAV take-off and landing platforms, new energy supply facilities, and communication and navigation facilities; and a health assessment and prediction module, used to identify safety risks of each infrastructure based on the real-time operational status data using multiple preset risk identification models, obtain multiple risk assessment results, and calculate multiple sub-health levels based on the multiple risk assessment results; and to dynamically weight the multiple sub-health levels. The system employs a fusion algorithm to calculate the comprehensive health index of the UAV take-off and landing site; a lifecycle curve analysis module, which uses a multinomial regression model to fit the lifecycle curve of the UAV take-off and landing site's health status based on historical comprehensive health data sequences, and obtains the predicted comprehensive health index based on the lifecycle curve; and an early warning and self-repair module, which compares the predicted comprehensive health index with preset multi-level alarm thresholds. When the prediction result reaches the corresponding threshold, the module automatically triggers the corresponding level of early warning mechanism and schedules maintenance or emergency response operations within the take-off and landing site according to predefined operation and maintenance strategies. Simultaneously, the module feeds back the execution results to the digital twin model, enabling predictive maintenance of the take-off and landing site.

[0048] Thirdly, embodiments of this application provide a terminal 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 aforementioned health-state-driven predictive maintenance method for UAV take-off and landing sites.

[0049] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the aforementioned health-state-driven predictive maintenance method for UAV take-off and landing sites.

[0050] Beneficial Effects: This application first collects real-time operational status data for each infrastructure, and drives the digital twin model to be updated synchronously based on the real-time operational status data. Then, based on the real-time operational status data, multiple preset risk identification models are used to identify security risks for each infrastructure, obtaining multiple risk assessment results. Multiple sub-health levels are calculated based on the multiple risk assessment results. Next, based on the multiple sub-health levels, a dynamic weighted fusion algorithm is used to calculate the comprehensive health index of the UAV take-off and landing field. Based on the historical comprehensive health data sequence, a multinomial regression model is used to fit the life cycle curve of the health status of the UAV take-off and landing field, and the predicted comprehensive health index is obtained based on the life cycle curve. Finally, the predicted comprehensive health index is compared with the preset multi-level alarm thresholds. When the prediction result reaches the corresponding threshold, the corresponding level of early warning mechanism is automatically triggered, and the operation and maintenance execution resources in the take-off and landing field are scheduled to perform maintenance or emergency response operations according to the predefined operation and maintenance strategy. At the same time, the execution results are fed back to the digital twin model to realize predictive maintenance of the take-off and landing field. This application has achieved a fundamental shift in the maintenance mode of UAV take-off and landing sites from passive response to proactive prediction by constructing an intelligent closed-loop operation and maintenance system that integrates state perception, dynamic assessment, trend prediction and autonomous repair, which significantly improves operational safety and maintenance efficiency. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating a health-state-driven predictive maintenance method for UAV take-off and landing sites provided in an embodiment of this application.

[0053] Figure 2 This is a fitting diagram of the life cycle curve of a UAV take-off and landing site provided in an embodiment of this application;

[0054] Figure 3 This is a schematic diagram of the structure of a health-state-driven predictive maintenance system for UAV take-off and landing sites provided in an embodiment of this application;

[0055] Figure 4 This is a schematic diagram of the structure of the terminal device provided in the embodiments of this application. Detailed Implementation

[0056] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0057] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0058] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0059] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0060] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0061] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0062] The following description, with reference to the accompanying drawings, details a health-state-driven predictive maintenance method, system, terminal equipment, and storage medium for UAV take-off and landing sites provided in this application.

[0063] Figure 1 The diagram shows a flowchart of a health-state-driven predictive maintenance method for UAV take-off and landing sites provided in an embodiment of this application.

[0064] like Figure 1 As shown, this health-state-driven predictive maintenance method for UAV takeoff and landing sites includes the following steps:

[0065] S1. Collect real-time operational status data for each infrastructure component and drive the digital twin model to update synchronously based on the real-time operational status data; the infrastructure includes UAV take-off and landing platforms, new energy supply facilities, and communication and navigation facilities.

[0066] Furthermore, in this embodiment of the application, step S1 includes:

[0067] By collecting surface status data of the UAV take-off and landing platform, operational parameter data of the new energy supply facility, and communication quality data of the communication and navigation facility through multi-source sensors deployed in the UAV take-off and landing site, real-time operational status data of each infrastructure is formed.

[0068] Outlier detection, missing data filling, and data format standardization are performed on real-time operating status data to obtain standardized status data for subsequent analysis.

[0069] The digital twin model is updated by synchronizing data based on standardized state data. A periodic synchronization strategy is used for normal state data, and a priority synchronization strategy is used for abnormal state data to improve the timeliness of response to abnormal states in the digital twin model.

[0070] The synchronized updated digital twin model will be used as a virtual mapping of the physical entities of the UAV take-off and landing site for subsequent health status assessment, prediction and operation and maintenance decisions.

[0071] S2. Based on real-time operational status data, multiple preset risk identification models are used to identify security risks in each infrastructure, obtain multiple risk assessment results, and calculate multiple sub-health scores based on the multiple risk assessment results.

[0072] Furthermore, in this embodiment, when the preset risk identification model is an image semantic segmentation model based on a fully convolutional neural network, the above step S2 includes:

[0073] Real-time image data of the UAV take-off and landing platform is acquired by optical cameras and thermal imaging cameras, and the real-time image data is input into an image semantic segmentation model based on a fully convolutional neural network.

[0074] The image semantic segmentation model is used to perform pixel-level semantic segmentation on real-time image data to identify the percentage of water, snow and ice areas on the drone take-off and landing platform, i.e., the coverage rate, and to identify the length, width and location of cracks on the drone take-off and landing platform.

[0075] Based on coverage, the primary risk assessment result is determined to be the health of the drone take-off and landing platform under slippery conditions. ;

[0076] Based on the three preset ranges corresponding to the crack width, the two classification regions corresponding to the crack location, and the three preset ranges corresponding to the crack length, the second risk assessment result is determined as the health status of the UAV take-off and landing platform under damage conditions, according to the crack length, crack width, and crack location of the UAV take-off and landing platform. ;

[0077] Based on health Health And with preset weights, a weighted fusion algorithm is used to determine the sub-health of the drone take-off and landing platform. .

[0078] Furthermore, in this embodiment of the application, when the preset risk identification model is a CNN-BiLSTM neural network model, the above step S2 includes:

[0079] Collect real-time operating status data of new energy supply facilities. The real-time operating status data includes at least temperature parameters, voltage parameters, current parameters, and combustible gas concentration parameters, and construct the real-time operating status data into time series data.

[0080] The time series data is input into the CNN-BiLSTM neural network model, which outputs a risk assessment result, namely the probability of the accident occurring, specifically:

[0081] Time series data is input into a CNN network to extract features from the time series data in order to obtain the spatial correlation features between various operating parameters;

[0082] Spatial correlation features are input into a BiLSTM network, and forward and backward long short-term memory networks are used to perform temporal modeling of spatial correlation features in order to extract time-related features of the operating status of new energy supply facilities.

[0083] Input time-related features into a fully connected layer and output the probability of an accident.

[0084] Mapping the probability of accidents to the sub-health of renewable energy supply facilities .

[0085] Furthermore, in this embodiment of the application, when the preset risk identification model is a BP deep learning neural network, the above step S2 includes:

[0086] Collect real-time operational status data of communication and navigation facilities. The real-time operational status data shall include at least the signal strength, signal-to-noise ratio, number of tracked satellites, transmission delay, data packet loss rate and bit error rate parameters of satellite communication, 5G-A communication and ADS-B communication.

[0087] The real-time operational status data of communication and navigation facilities is normalized and then used to construct a feature vector, which is then input into a trained BP deep learning neural network. The trained BP deep learning neural network is trained based on the monitoring status data of communication and navigation facilities.

[0088] The trained BP deep learning neural network is used to perform nonlinear mapping on the feature vectors, and the risk assessment results of communication and navigation facilities are output.

[0089] Based on the risk assessment results of communication and navigation facilities, the operational status of communication and navigation facilities is mapped to the sub-health of communication and navigation facilities. .

[0090] S3. Based on multiple sub-health scores, a dynamic weighted fusion algorithm is used to calculate the comprehensive health index of the UAV take-off and landing site;

[0091] Furthermore, in this embodiment of the application, step S3 includes:

[0092] Based on the sub-health of the drone take-off and landing platform Sub-health of new energy supply facilities and the sub-health of communication and navigation facilities Weighted calculations were performed to obtain the comprehensive health index H of the drone take-off and landing site:

[0093]

[0094] in, , , They are respectively , as well as The corresponding dynamic weighting coefficients; , , For the first The dynamic weighting coefficients corresponding to the health of each infrastructure sub-item for Weighting coefficients in the dimension of risk severity. for The weighting coefficient for the facility's impact range, , , ;in, The calculation formula is:

[0095]

[0096] in, The assignment relationship is as follows: when the influence range of the drone take-off and landing platform is to affect all drones, The value is 0.5; when the impact range of the new energy supply facility is limited to affecting new energy drones, The value is 0.3; when the impact range of communication and navigation facilities is limited to affecting communication but can be switched to a backup link, It is 0.2.

[0097] S4. Based on the historical comprehensive health data series, use a multinomial regression model to fit the life cycle curve of the health status of the UAV take-off and landing field, and obtain the predicted comprehensive health index based on the life cycle curve.

[0098] Furthermore, in this embodiment of the application, step S4 includes:

[0099] Obtain historical comprehensive health data sequences of drone take-off and landing sites And normalize the historical comprehensive health data sequence;

[0100] Based on the normalized historical comprehensive health data series, a multinomial regression model is constructed:

[0101]

[0102] in, It is the first Health prediction values ​​at each time point These are the regression coefficients;

[0103] Define a loss function to quantify the difference between the predicted and actual health scores:

[0104]

[0105] in, This represents the actual health level.

[0106] Using the least squares method, with the goal of minimizing the loss function, we find the parameter combination that minimizes the squared error loss through matrix operations. :

[0107]

[0108] in, , It is a matrix The transpose of the matrix, It is a matrix The inverse matrix, It is a comprehensive health index of the drone take-off and landing site. ;

[0109] Based on the combination of substituted parameters The multinomial regression model is used to obtain the life cycle curve, and the predicted comprehensive health index is obtained based on the life cycle curve.

[0110] S5. Compare the predicted comprehensive health index with the preset multi-level alarm thresholds. When the prediction result reaches the corresponding threshold, the corresponding level of early warning mechanism is automatically triggered. According to the predefined operation and maintenance strategy, the operation and maintenance execution resources in the take-off and landing field are scheduled to perform maintenance or emergency response operations. At the same time, the execution results are fed back to the digital twin model to realize predictive maintenance of the take-off and landing field.

[0111] This application provides a predictive maintenance method for UAV take-off and landing sites based on health status. First, real-time operational status data for each infrastructure component is collected, and a digital twin model is synchronously updated based on this data. Then, based on the real-time operational status data, multiple preset risk identification models are used to identify safety risks for each infrastructure component, obtaining multiple risk assessment results. Multiple sub-health levels are calculated based on these risk assessment results. Next, a dynamic weighted fusion algorithm is used to calculate the comprehensive health index of the UAV take-off and landing site based on these sub-health levels. Based on historical comprehensive health data sequences, a multinomial regression model is used to fit the lifecycle curve of the UAV take-off and landing site's health status, and a predicted comprehensive health index is obtained from the lifecycle curve. Finally, the predicted comprehensive health index is compared with preset multi-level alarm thresholds. When the predicted result reaches the corresponding threshold, the corresponding level of early warning mechanism is automatically triggered, and maintenance or emergency response operations are performed according to predefined operation and maintenance strategies. Simultaneously, the execution results are fed back to the digital twin model, achieving predictive maintenance of the take-off and landing site. This application has achieved a fundamental shift in the maintenance mode of UAV take-off and landing sites from passive response to proactive prediction by constructing an intelligent closed-loop operation and maintenance system that integrates state perception, dynamic assessment, trend prediction and autonomous repair, which significantly improves operational safety and maintenance efficiency.

[0112] The technical solution provided in this application will be illustrated below with another embodiment.

[0113] Step 1: Sensing the operational status of the takeoff and landing field;

[0114] Step 1.1: Data Collection and Preprocessing. Data is collected from IoT sensing terminals such as takeoff and landing field cameras, meteorological monitoring sensors, drone nests, charging / hydrogen refueling / battery swapping equipment, combustible gas monitoring sensors, and satellite signal monitoring equipment. Box plots are used to detect outliers in the dataset and fill in missing values ​​to obtain a multimodal monitoring dataset of the takeoff and landing field infrastructure operation status.

[0115] Step 1.2: Digital Twin 3D Modeling. By combining data sources such as BIM, CIM, 3DGIS, IoT, and Video, operational status data of important infrastructure such as UAV take-off and landing platforms, hangars, charging piles, and communication facilities are collected. Through high-precision 3D modeling and real-time data fusion technology, a virtual digital space that is a complete mirror image of the physical entity of the UAV take-off and landing site is constructed.

[0116] The digital twin model of the takeoff and landing field adopts a strategy of real-time transmission while giving priority to abnormal data. That is, regular data (such as temperature and signal strength) is synchronized every 50Hz, while abnormal data (such as excessive combustible gas concentration or sudden change in crack length) triggers priority synchronization. Edge computing preprocessing reduces transmission delay and ensures that the synchronization delay is ≤100ms. In case of synchronization failure, the backup communication link is automatically switched (such as 5G-A to satellite communication).

[0117] Step 2: Health assessment of takeoff and landing site infrastructure;

[0118] Step 2.1: Health assessment of the drone take-off and landing platform.

[0119] Damage to drone takeoff and landing platforms, such as water accumulation, snow accumulation, ice, and cracks, poses a safety threat to drone takeoff and landing. Water and snow accumulation can cause short circuits in charging contacts or communication interfaces on the platform, potentially damaging platform equipment and causing irreversible electrical damage to the drone itself. When drone wheels come into contact with a slippery platform containing water, snow, or ice, friction is drastically reduced, easily leading to skidding, veering, or even tipping over. Cracks in the takeoff and landing platform can jam the drone's landing gear, causing structural damage. Uneven surfaces such as cracks and rust can prevent the support structure from landing smoothly at the same time, causing violent shaking or even tipping over, damaging the propellers and gimbal.

[0120] Thermal imaging and optical cameras are used to collect image data of the take-off and landing platform. A semantic segmentation model based on a fully convolutional neural network (such as FCN-8s) is constructed, and pixel-level fine-grained inference is achieved by performing dense label prediction on each pixel.

[0121] Based on the monitored coverage of water, snow, and ice accumulation on the drone landing platform, the health status of the drone landing platform under slippery conditions is designed. ( The evaluation criteria are shown in Table 1.

[0122] Table 1

[0123] Based on the identified crack length and width of the UAV take-off and landing platform, and considering the location of the crack, the health of the UAV take-off and landing platform under the damaged state is assessed. The takeoff and landing platform is divided into a core area and a non-core area. The core area is the load-bearing location of the UAV's conventional takeoff and landing pivot, while the non-core area is located in the platform's non-core load-bearing zone and edge area, far from the conventional takeoff and landing pivot. The health of the UAV takeoff and landing platform under design damage conditions is also considered. ( The evaluation criteria are shown in Table 2.

[0124] Table 2

[0125] Overall health of the drone take-off and landing platform under slippery conditions and the health of the drone take-off and landing platform under damaged conditions. The overall health status of the drone take-off and landing platform is obtained:

[0126]

[0127] in, , These are the weighting coefficients. .

[0128] Step 2.2: Health assessment of drone refueling infrastructure;

[0129] New energy drones use clean energy sources such as liquid hydrogen, methanol, and liquid ammonia. During charging at drone landing sites, leaks or operational errors could lead to serious casualties and environmental pollution. This study collects data on temperature, voltage, current, and hydrogen, methane, and propane concentrations from charging / hydrogen refueling / battery swapping facilities. A fire and explosion risk monitoring model for these facilities is established using a CNN-BILSTM neural network to assess the fire and explosion risks. The overall health of the new energy supply infrastructure is then evaluated based on the risk probability.

[0130] After preprocessing, the data is input into a CNN neural network to extract spatial features for each feature. This extracted feature is then fed into a BILSTM layer to train the forward and backward networks to fit the data and obtain temporal features. The output of the BILSTM network is used as input to the fully connected layer, ultimately yielding the probability of an explosion caused by leakage of the new energy medium or circuit overload in charging / hydrogen refueling / battery swapping facilities.

[0131] As an example, the parameters of the CNN-BILSTM neural network can be set as shown in Table 3.

[0132] Table 3

[0133] Based on the probability of fires and explosions at new energy supply infrastructure, a health level is defined for the new energy supply infrastructure. As shown in Table 4.

[0134] Table 4

[0135] Step 2.3, Health assessment of communication and navigation infrastructure;

[0136] Table 5 shows a health assessment index system for UAV take-off and landing site communication and navigation facilities, including satellite communication, 5G-A, and ADS-B communication and navigation methods. A quantitative assessment model for the health of UAV take-off and landing site communication and navigation facilities is constructed using a BP deep learning neural network.

[0137] Table 5

[0138] The input to the BP neural network is the feature values ​​of satellite communication, 5G-A communication, and ADS-B communication. The output is the health status of the communication and navigation infrastructure. The input to the BP neural network is the indicator features of the UAV's communication and navigation infrastructure, and the network output is the health status of the UAV's communication and navigation infrastructure, with the output layer nodes displaying the expected output. The corresponding values ​​of 0, 30, 50, 70, and 100 represent changes in signal interference intensity from strong to weak, and corresponding changes in health level from low to high. An example of BP neural network parameter settings is shown in Table 6.

[0139] Table 6

[0140] The model is trained using monitoring data from the communication and navigation infrastructure of UAV take-off and landing sites. The model parameters are iteratively updated repeatedly to ultimately obtain a BP neural network-based health assessment model for communication and navigation infrastructure that meets application requirements, and the health score is obtained. .

[0141] Step 3: Overall health assessment of the takeoff and landing field

[0142] Taking into account the health of takeoff and landing site infrastructure, including drone takeoff and landing platforms, hangars / charging stations, and communication and navigation facilities, the overall health of the takeoff and landing site infrastructure is calculated using a weighted average. The dynamic weighted calculation formula for the overall health is as follows:

[0143]

[0144] in, These are the health statuses of the drone take-off and landing platform, hangar / charging station, and communication and navigation facilities. These are the dynamic weighting coefficients corresponding to the three types of infrastructure health. A risk-priority-driven dynamic weighting method is set, meaning that facilities and systems with lower health and wider impact receive higher weights, ensuring that the overall health accurately reflects key risks. The risk-priority-driven classification is based on two dimensions: risk severity and facility impact range. Therefore:

[0145]

[0146] in, For the first The weighting coefficients corresponding to the health of each infrastructure item These correspond to drone take-off and landing platforms, hangars / charging stations, and communication and navigation facilities, respectively. for Weighting coefficients in the dimension of risk severity. for Weighting coefficients for the scope of impact of facilities; , , . The calculation formula is:

[0147]

[0148] The scope and weight of the facility's impact on drone take-off and landing safety The assignment relationships are shown in Table 7 below.

[0149] Table 7

[0150] Step 4: Plot the life cycle curve.

[0151] Assume the monitoring point receives the first The health sequence over time is The health score sequence at historical moments is normalized. Based on the normalized health scores, a multinomial regression model is constructed:

[0152]

[0153] in, It is calculated based on the polynomial function. Health prediction values ​​at each time point The input is the health status independent variable. These are the regression coefficients that need to be determined.

[0154] Define a loss function to quantify the difference between predicted and actual values:

[0155]

[0156] Using the least squares method, with the goal of minimizing the loss function, we find the parameter combination that minimizes the squared error loss through matrix operations. :

[0157]

[0158] in , It is a matrix The transpose of the matrix, It is a matrix The inverse matrix, These are historical health observations. .

[0159] The parameters were solved using the least squares method, and a multiple linear regression model was trained using historical health data. The trained model was then used to predict the performance of the takeoff and landing site infrastructure at specific time points. Health value .

[0160] As an example, such as Figure 2 As shown. Using historical health data of a certain UAV take-off and landing site infrastructure, data points were sampled every half hour for a total of 15 hours. The resulting cubic regression equation, fitted with 30 data points, is as follows:

[0161]

[0162] Using the fitted lifecycle curve of the UAV take-off and landing site infrastructure, the health status of future time intervals is predicted. The predicted health status value at the 15.5-hour mark (31st data point) is 44.88; the predicted health status value at the 16-hour mark (32nd data point) is 39.85.

[0163] Step 5: Intelligent Early Warning and Self-Repair

[0164] Step 5.1: Intelligent Tiered Early Warning

[0165] After obtaining the health prediction results of the takeoff and landing field infrastructure, abnormal situations are managed in a tiered manner. By pre-setting risk indicators and classification standards, potential safety issues are divided into different levels, which facilitates the rational arrangement of operation and maintenance tasks according to their priority. The correspondence between the health level, health status, and early warning indicators of the takeoff and landing field infrastructure is shown in Table 8.

[0166] Table 8

[0167] Step 5.2: Autonomously implement operation and maintenance measures

[0168] Based on the safety risk assessment results of different drone take-off and landing sites, the tiered emergency response mechanism of the take-off and landing site is automatically triggered to eliminate and control safety hazards. Hazard investigation priorities are set according to the level of safety risk; different hazard investigation measures are taken according to the type of safety risk. Examples of emergency measures for safety hazards in take-off and landing site infrastructure are shown in Table 9.

[0169] Table 9

[0170] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0171] Corresponding to the above embodiment, a health-state-driven predictive maintenance method for UAV take-off and landing sites, Figure 3 The diagram shows a structural block diagram of a health-state-driven predictive maintenance system for UAV take-off and landing sites provided in an embodiment of this application. For ease of explanation, only the parts related to the embodiments of this application are shown.

[0172] Reference Figure 3 The system includes:

[0173] The status awareness module is used to collect real-time operational status data for each infrastructure and drive the digital twin model to be updated synchronously based on the real-time operational status data; the infrastructure includes UAV take-off and landing platforms, new energy supply facilities, and communication and navigation facilities.

[0174] The health assessment and prediction module is used to identify security risks of each infrastructure based on real-time operational status data and multiple preset risk identification models, obtain multiple risk assessment results, and calculate multiple sub-health levels based on the multiple risk assessment results; based on the multiple sub-health levels, a dynamic weighted fusion algorithm is used to calculate the comprehensive health index of the UAV take-off and landing field;

[0175] The life cycle curve analysis module is used to fit the life cycle curve of the health status of UAV take-off and landing sites using a multinomial regression model based on historical comprehensive health data series, and obtain the predicted comprehensive health index based on the life cycle curve.

[0176] The early warning and self-repair module compares the predicted comprehensive health index with the preset multi-level alarm thresholds. When the prediction result reaches the corresponding threshold, the corresponding level of early warning mechanism is automatically triggered, and the operation and maintenance execution resources in the take-off and landing field are scheduled to perform maintenance or emergency response operations according to the predefined operation and maintenance strategy. At the same time, the execution results are fed back to the digital twin model to realize predictive maintenance of the take-off and landing field.

[0177] In practical use, the health-state-driven predictive maintenance system for UAV take-off and landing sites provided in this application embodiment can be configured in any terminal device to execute the aforementioned health-state-driven predictive maintenance method for UAV take-off and landing sites.

[0178] This application provides a health-state-driven predictive maintenance system for UAV take-off and landing sites. First, it collects real-time operational status data for each infrastructure component and updates a digital twin model synchronously based on this data. Then, using multiple preset risk identification models, it identifies safety risks for each infrastructure component based on the real-time operational status data, obtaining multiple risk assessment results. Based on these results, it calculates multiple sub-health levels. Next, using a dynamic weighted fusion algorithm, it calculates the comprehensive health index of the UAV take-off and landing site. Based on historical comprehensive health data sequences, it uses a multinomial regression model to fit the lifecycle curve of the UAV take-off and landing site's health status and obtains the predicted comprehensive health index. Finally, it compares the predicted comprehensive health index with preset multi-level alarm thresholds. When the predicted result reaches the corresponding threshold, it automatically triggers the corresponding level of early warning mechanism and schedules maintenance or emergency response operations within the take-off and landing site according to predefined maintenance strategies. Simultaneously, the execution results are fed back to the digital twin model, achieving predictive maintenance of the take-off and landing site. This application has achieved a fundamental shift in the maintenance mode of UAV take-off and landing sites from passive response to proactive prediction by constructing an intelligent closed-loop operation and maintenance system that integrates state perception, dynamic assessment, trend prediction and autonomous repair, which significantly improves operational safety and maintenance efficiency.

[0179] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0180] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0181] To implement the above embodiments, this application also proposes a terminal device.

[0182] Figure 4 This is a schematic diagram of the structure of a terminal device according to an embodiment of this application.

[0183] like Figure 4 As shown, the terminal device 200 includes:

[0184] The system includes a memory 210 and at least one processor 220, and a bus 230 connecting different components (including the memory 210 and the processor 220). The memory 210 stores a computer program, which, when executed by the processor 220, implements the health-state-driven predictive maintenance method for UAV take-off and landing fields described in this application embodiment.

[0185] Bus 230 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0186] Terminal device 200 typically includes various electronically readable media. These media can be any available media that can be accessed by terminal device 200, including volatile and non-volatile media, removable and non-removable media.

[0187] Memory 210 may also include computer system readable media in the form of volatile memory, such as random access memory (RAM) 240 and / or cache memory 250. Terminal device 200 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 260 may be used to read and write non-removable, non-volatile magnetic media (… Figure 4 Not shown; usually referred to as a "hard drive"). Although Figure 4 As not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 230 via one or more data media interfaces. Memory 210 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of this application.

[0188] A program / utility 280 having a set (at least one) of program modules 270 may be stored in, for example, memory 210. Such program modules 270 include—but are not limited to—an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 270 typically perform the functions and / or methods described in the embodiments of this application.

[0189] Terminal device 200 can also communicate with one or more external devices 290 (e.g., keyboard, pointing device, display 291, etc.), and with one or more devices that enable a user to interact with terminal device 200, and / or with any device that enables terminal device 200 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed via input / output (I / O) interface 292. Furthermore, terminal device 200 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 293. As shown, network adapter 293 communicates with other modules of terminal device 200 via bus 230. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with terminal device 200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0190] The processor 220 performs various functional applications and data processing by running programs stored in the memory 210.

[0191] It should be noted that the implementation process and technical principles of the terminal device in this embodiment are explained in the foregoing description of the health status-driven predictive maintenance method for UAV take-off and landing sites in this application embodiment, and will not be repeated here.

[0192] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0193] This application provides a computer program product that, when run on a terminal device, enables the terminal device to implement the steps described in the various method embodiments above.

[0194] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0195] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0196] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0197] In the embodiments provided in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0198] 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0199] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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. Such 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 this application, and should all be included within the protection scope of this application.

Claims

1. A predictive maintenance method for UAV takeoff and landing sites based on health status, characterized in that, include: S1. Collect real-time operational status data corresponding to each infrastructure, and drive the digital twin model to be updated synchronously based on the real-time operational status data; The infrastructure includes unmanned aerial vehicle (UAV) take-off and landing platforms, new energy supply facilities, and communication and navigation facilities; S2. Based on the real-time operating status data, multiple preset risk identification models are used to identify security risks of each infrastructure, obtain multiple risk assessment results, and calculate multiple sub-health scores based on the multiple risk assessment results. S3. Based on multiple sub-health scores, a dynamic weighted fusion algorithm is used to calculate the comprehensive health index of the UAV take-off and landing site; S4. Based on the historical comprehensive health data sequence, a multinomial regression model is used to fit the life cycle curve of the health status of the UAV take-off and landing field, and the predicted comprehensive health index is obtained according to the life cycle curve. S5. Compare the predicted comprehensive health index with the preset multi-level alarm thresholds. When the prediction result reaches the corresponding threshold, the corresponding level of early warning mechanism is automatically triggered. According to the predefined operation and maintenance strategy, the operation and maintenance execution resources in the take-off and landing field are scheduled to perform maintenance or emergency response operations. At the same time, the execution results are fed back to the digital twin model to realize predictive maintenance of the take-off and landing field.

2. The method as described in claim 1, characterized in that, S1 involves collecting real-time operational status data for each infrastructure component and driving synchronous updates of the digital twin model based on this data. Specifically, this includes: By collecting surface status data of the UAV take-off and landing platform, operational parameter data of the new energy supply facility, and communication quality data of the communication and navigation facility through multi-source sensors deployed in the UAV take-off and landing site, real-time operational status data of each infrastructure is formed. The real-time operating status data is subjected to outlier detection, missing data filling, and data format standardization to obtain standardized status data for subsequent analysis; The digital twin model is updated by synchronizing data based on the standardized state data. A periodic synchronization strategy is used for normal state data, and a priority synchronization strategy is used for abnormal state data to improve the timeliness of response to abnormal states in the digital twin model. The synchronized updated digital twin model will be used as a virtual mapping of the physical entities of the UAV take-off and landing site for subsequent health status assessment, prediction and operation and maintenance decisions.

3. The method as described in claim 1, characterized in that, The preset risk identification model is an image semantic segmentation model based on a fully convolutional neural network; step S2, based on the real-time operating status data, uses multiple preset risk identification models to identify security risks in each infrastructure, obtains multiple risk assessment results, and calculates multiple sub-health scores based on the multiple risk assessment results, specifically including: Real-time image data of the UAV take-off and landing platform is acquired by optical cameras and thermal imaging cameras, and the real-time image data is input into the image semantic segmentation model based on a fully convolutional neural network. The image semantic segmentation model is used to perform pixel-level semantic segmentation on the real-time image data to identify the percentage of water, snow and ice areas on the drone take-off and landing platform, i.e., the coverage rate, and to identify the length, width and location of cracks on the drone take-off and landing platform. Based on coverage, the primary risk assessment result is determined to be the health of the drone take-off and landing platform under slippery conditions. ; Based on the three preset ranges corresponding to the crack width, the two classification regions corresponding to the crack location, and the three preset ranges corresponding to the crack length, the second risk assessment result is determined as the health status of the UAV take-off and landing platform under damage conditions, according to the crack length, crack width, and crack location of the UAV take-off and landing platform. ; Based on health Health And with preset weights, a weighted fusion algorithm is used to determine the sub-health of the drone take-off and landing platform. .

4. The method as described in claim 3, characterized in that, The preset risk identification model is a CNN-BiLSTM neural network model; step S2, based on the real-time operating status data, uses multiple preset risk identification models to identify security risks in each infrastructure, obtains multiple risk assessment results, and calculates multiple sub-health scores based on the multiple risk assessment results, specifically including: Collect real-time operating status data of new energy supply facilities. The real-time operating status data includes at least temperature parameters, voltage parameters, current parameters, and combustible gas concentration parameters, and construct the real-time operating status data into time series data. The time series data is input into the CNN-BiLSTM neural network model, which outputs a risk assessment result, i.e., the probability of accident occurrence, specifically: The time series data is input into a CNN network to extract features from the time series data in order to obtain the spatial correlation features between various operating parameters. The spatial correlation features are input into a BiLSTM network, and forward and backward long short-term memory networks are used to perform temporal modeling on the spatial correlation features in order to extract the time-related features of the operation status of the new energy supply facilities. The time-related features are input into a fully connected layer, and the probability of the accident is output. Mapping the probability of the aforementioned accident to the sub-health of the new energy supply facility. .

5. The method as described in claim 4, characterized in that, The preset risk identification model is a BP deep learning neural network; S2, based on the real-time operating status data, uses multiple preset risk identification models to identify security risks in each infrastructure, obtains multiple risk assessment results, and calculates multiple sub-health scores based on the multiple risk assessment results, specifically including: Collect real-time operational status data of communication and navigation facilities. The real-time operational status data includes at least the signal strength, signal-to-noise ratio, number of tracked satellites, transmission delay, data packet loss rate, and bit error rate parameters of satellite communication, 5G-A communication, and ADS-B communication. The real-time operational status data of the communication and navigation facilities is normalized and then used to construct a feature vector, which is then input into a trained BP deep learning neural network. The trained BP deep learning neural network is trained based on the monitoring status data of the communication and navigation facilities. The trained BP deep learning neural network is used to perform a nonlinear mapping on the feature vector, and the risk assessment results of communication and navigation facilities are output. Based on the risk assessment results of the communication and navigation facilities, the operational status of the communication and navigation facilities is mapped to the sub-health level of the communication and navigation facilities. .

6. The method as described in claim 5, characterized in that, S3, based on multiple sub-health scores, uses a dynamic weighted fusion algorithm to calculate the comprehensive health index of the UAV take-off and landing site, specifically including: Based on the sub-health of the drone take-off and landing platform Sub-health of new energy supply facilities and the sub-health of communication and navigation facilities Weighted calculations were performed to obtain the comprehensive health index H of the drone take-off and landing site: ； in, , , They are respectively , as well as The corresponding dynamic weighting coefficients; , , For the first The dynamic weighting coefficients corresponding to the health of each infrastructure sub-item for The weighting coefficients in the dimension of risk severity. for The weighting coefficient for the facility's impact range, , , ;in, The calculation formula is: ； in, The assignment relationship is as follows: when the influence range of the drone take-off and landing platform is to affect all drones, The value is 0.5; when the impact range of the new energy supply facility is limited to affecting new energy drones, The value is 0.3; when the impact range of communication and navigation facilities is limited to affecting communication but can be switched to a backup link, It is 0.

2.

7. The method as described in claim 6, characterized in that, S4, based on historical comprehensive health data sequences, uses a multinomial regression model to fit the lifecycle curve of the UAV takeoff and landing field health status, and obtains a predicted comprehensive health index based on the lifecycle curve, specifically including: Obtain historical comprehensive health data sequences of drone take-off and landing sites And normalize the historical comprehensive health data sequence; Based on the normalized historical comprehensive health data series, a multinomial regression model is constructed: ； in, It is the first Health prediction values ​​at each time point These are the regression coefficients; Define a loss function to quantify the difference between the predicted and actual health scores: ； in, This represents the actual health level. Using the least squares method, with the goal of minimizing the loss function, we find the parameter combination that minimizes the squared error loss through matrix operations. : ； in, , It is a matrix The transpose of the matrix, It is a matrix The inverse matrix, This is the comprehensive health index of the aforementioned drone take-off and landing site. ; Based on the parameter combination substituted A multinomial regression model is used to obtain a life cycle curve, and a predicted comprehensive health index is obtained based on the life cycle curve.

8. A health-state-driven predictive maintenance system for UAV takeoff and landing sites, characterized in that, include: The status awareness module is used to collect real-time operational status data corresponding to each infrastructure and drive the digital twin model to be updated synchronously based on the real-time operational status data. The infrastructure includes unmanned aerial vehicle (UAV) take-off and landing platforms, new energy supply facilities, and communication and navigation facilities; The health assessment and prediction module is used to identify security risks of each infrastructure based on the real-time operating status data and using multiple preset risk identification models to obtain multiple risk assessment results, and calculate multiple sub-health levels based on the multiple risk assessment results. Based on multiple sub-health scores, a dynamic weighted fusion algorithm is used to calculate the comprehensive health index of the UAV take-off and landing site; The life cycle curve analysis module is used to fit the life cycle curve of the health status of the UAV take-off and landing field using a multinomial regression model based on the historical comprehensive health data sequence, and to obtain the predicted comprehensive health index based on the life cycle curve. The early warning and self-repair module is used to compare the predicted comprehensive health index with the preset multi-level alarm thresholds. When the prediction result reaches the corresponding threshold, the corresponding level of early warning mechanism is automatically triggered, and the operation and maintenance execution resources in the take-off and landing field are scheduled to perform maintenance or emergency response operations according to the predefined operation and maintenance strategy. At the same time, the execution results are fed back to the digital twin model to realize predictive maintenance of the take-off and landing field.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images