General airport remote service man-machine decision-making system based on large model

Abstract

The invention relates to the technical field of general aviation airport management, in particular to a general airport remote service man-machine decision-making system based on a large model, which comprises a general airport large model and a general airport operation post man-machine decision-making module based on the general airport large model, the general airport large model specifically comprises a general airport security check large model, a general airport control command large model, a general airport meteorological service large model and a general airport operation monitoring large model; the general airport operation post man-machine decision module specifically comprises a general airport security check module, a general airport control command module, a general airport meteorological service module and a general airport operation monitoring module. According to the system, the workload of operators is reduced, the airport operation safety and efficiency are improved, and compared with the prior art, the problem that remote intelligent control of airport security check, control command, meteorological service and operation monitoring is lacked is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of general aviation airport management, and in particular to a general airport remote service human-machine decision-making system based on a large model. Background Art

[0002] General aviation transport covers a wide range of sectors, including agriculture, industry, forestry, medical care, tourism, and emergency rescue, and is a key growth driver for regional economies. General airports are the foundation of general aviation operations, providing takeoff, landing, and parking services for general aviation aircraft. These airports drive local economic development, promote tourism and other related industries, and contribute to improving the integrated transportation system, enabling seamless integration of multiple modes of transportation.

[0003] The normal operation of general airports requires a certain number of airport operations personnel and equipment. However, many general airports are located in remote areas, making it difficult to recruit and retain registered and licensed operators. Furthermore, the high investment in airport facilities and equipment makes equipment maintenance a challenge for airports. More importantly, compared to transport airports, the control characteristics of general airports cannot be simply summarized. While controllers at trunk and feeder airports have clear divisions of labor and well-established workflows, general airports, due to their low total flight volume and smaller operations personnel, face a wide variety of flight types, a heavy workload of coordinating multiple functions, and limited and unreliable airport equipment. This makes it more prone to errors and omissions, impacting operational safety.

[0004] The development of general aviation and airports requires overcoming multiple barriers, including the lack of personnel, specialized talent, and intelligent decision-making technology. Remote tower technology, through a centralized and intelligent management model, offers a cost-saving and efficiency-enhancing solution for the industry. Remote towers utilize modern information technology to shift traditional tower management from airports to a centralized remote command center. Remote tower control reduces the cost of decentralized construction at individual airports through centralized construction, making it particularly well-suited to the low flight volume of general airports. It also addresses the challenges of recruiting and retaining controllers at general airports. More importantly, remote towers aggregate richer control information than traditional towers, improving the efficiency and safety of airport operations. However, currently, remote towers are primarily used for remote tower control and command. Airport security checks, equipment maintenance, meteorological services, and operational monitoring are all performed manually at general airports, hindering the realization of remote, intelligent, and cost-effective operations.

[0005] The prior art discloses a flight service system based on a knowledge graph and a large model (Publication No. CN119295018A). The system comprises: an ontology construction module for pre-defining relevant information about the flight service system based on user needs; a data acquisition module for acquiring open-source data related to the flight service system from multiple channels, deduplicating and normalizing the open-source data to generate standardized open-source data; a knowledge graph construction module for extracting triples from the standardized open-source data based on the relevant information to generate a knowledge graph triple set; a prompt generation module for generating corresponding target prompts based on user needs; and a large model fine-tuning module for training a pre-built large model based on the target prompts and triple set to generate a trained large model. The trained large model is used to generate results that match the target task based on the input user prompts and triples. This invention only addresses flight service issues and lacks remote intelligent control for airport security, air traffic control, meteorological services, and operational monitoring. Summary of the Invention

[0006] The purpose of the present invention is to overcome the problems of excessive reliance on manual labor and limited intelligent control in the existing technology, and to provide a general airport remote service human-machine decision-making system and device based on a large model.

[0007] In a first aspect, the present invention provides a general airport remote service human-machine decision-making system based on a large model, comprising a general airport large model and a general airport operation post human-machine decision-making module based on the general airport large model; The general airport large model specifically includes a general airport security inspection large model, a general airport control and command large model, a general airport meteorological service large model and a general airport operation monitoring large model; The general airport operation post human-machine decision-making module specifically includes a general airport security inspection module, a general airport control and command module, a general airport meteorological service module and a general airport operation monitoring module. Preferably, constructing the general airport large model is based on a general airport dataset; The steps of constructing the general airport data set specifically include collecting data from set data sources to obtain multi-source heterogeneous general airport operation data, and performing data preprocessing and data labeling on the multi-source heterogeneous general airport operation data.

[0008] Further preferably, the data source specifically includes video surveillance, X-ray security equipment, ADS-B equipment, airspace environment, facilities and equipment, operating rules and historical data.

[0009] Preferably, the general airport model also needs to be modified and supplemented through the knowledge graph; The steps of constructing the knowledge graph specifically include extracting data from the data source, and performing knowledge extraction, knowledge fusion, and knowledge processing based on the results of the data extraction.

[0010] Further preferably, the knowledge graph constructed is specifically a general airport security knowledge graph, which is used to explain and supplement the results output by the airport security large model.

[0011] Preferably, the general airport security inspection model specifically adopts a multimodal fusion model to obtain the category and location of prohibited items by processing the time series information of facial images, X-ray images and surveillance videos; The general airport control and command model specifically uses a graph neural network to establish the airspace structure and the dynamic relationship between flights to obtain the airport control strategy; The general airport weather service large model obtains airport weather forecast data by processing satellite images, radar images, time series changes of meteorological parameters and meteorological messages; The general airport operation monitoring model specifically monitors the real-time status parameters of airport equipment.

[0012] Further preferably, after the construction of the universal airport large model is completed, the airport large model needs to be trained and human-aligned.

[0013] Preferably, the general airport security module performs a secondary inspection based on the category and location of prohibited items obtained by the general airport security large model; The general airport control and command module combines the control strategy and manual decision-making obtained by the general airport control and command large model; The general airport weather service module determines whether the airport weather forecast data exceeds a set threshold based on the general airport weather model; The general airport operation monitoring module monitors and maintains the alarm equipment obtained by the general airport operation monitoring large model.

[0014] Further preferably, the general airport meteorological service module sets the set threshold according to the safety standards and operation requirements of the general airport.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention provides a general airport remote service human-machine decision-making system and device based on a large model. Based on the large model technology, it provides intelligent decision-making suggestions for general airport security inspection, control command, meteorological services, and operation monitoring. In combination with the human-machine decision-making method, it reduces the workload of operating personnel and improves the safety and efficiency of airport operations. Compared with the existing technology, the present invention solves the problem of lack of remote intelligent control of airport security inspection, control command, meteorological services, and operation monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Schematic diagram of the general airport remote service human-machine decision-making system based on a large model in Example 1.

[0017] Figure 2 Schematic diagram of the knowledge graph construction process in Example 1.

[0018] Figure 3 This is a schematic diagram of the general airport security knowledge graph in Example 1.

[0019] Figure 4 This is a schematic diagram of constructing multiple maps of general airports in Example 1. DETAILED DESCRIPTION

[0020] The present invention will be further described in detail below with reference to specific embodiments. However, this should not be construed as limiting the scope of the present invention to the following embodiments, as all technologies implemented based on the present invention fall within the scope of the present invention.

[0021] Unless otherwise specified, in the description of the specific embodiments of the present invention, the terms indicating the orientation or positional relationship, such as "upper", "lower", "left", "right", "center", "inside", and "outside", are based on the expressions of the orientation or positional relationship shown in the accompanying drawings, or are the orientation or positional relationship in which the invented product / device / apparatus is placed when it is conventionally used. These terms of orientation or positional relationship are merely for the purpose of facilitating the description of the scheme of the present invention or simplifying the description of the specific embodiments to facilitate the rapid understanding of the scheme by technicians, and do not indicate or imply that a specific device / component / element must have a specific orientation, or be constructed and operated in a specific positional relationship, and therefore should not be understood as limiting the present invention.

[0022] In addition, if the terms "horizontal", "vertical", "overhanging", "parallel" and the like appear, it does not mean that the corresponding devices / components / elements are required to be absolutely horizontal or vertical or overhanging or parallel, but may be slightly tilted or have deviations. For example, "horizontal" only means that its direction is more horizontal relative to "vertical", and does not mean that the structure must be completely horizontal, but may be slightly tilted. Alternatively, it can be simply understood that the corresponding devices / components / elements are set in directions such as "horizontal", "vertical", "overhanging", and "parallel", and can have an error / deviation of ±10% relative to the corresponding direction setting, more preferably an error / deviation within ±8%, more preferably an error / deviation within ±6%, more preferably an error / deviation within ±5%, and more preferably an error / deviation within ±4%. As long as the corresponding device / component / element is within the error / deviation range, it can still achieve its role in the solution of the present invention.

[0023] In addition, the expressions “first”, “second”, “third”, etc. in the terms are merely used to distinguish the description of the same or similar components, and should not be understood as emphasizing or implying the relative importance of specific components.

[0024] In addition, in the description of the embodiments of the present invention, "several," "plurality," and "a number" represent at least two. It can also be any number such as two, three, four, five, six, seven, eight, nine, or even more than nine.

[0025] Furthermore, in the description of the technical solution of the present invention, unless otherwise expressly specified, defined, or limited, the terms "disposed," "installed," "connected," "connected," "provided with," "laid," and "arranged" should be understood broadly. For example, they may refer to fixed connections, detachable connections, or integral connections. They may be welded, riveted, bolted, threaded, or other commonly used connection methods in the art. Such connections may be mechanical, electrical, or communicative; they may be direct, indirect via an intermediate medium, or internally connected between two components.

[0026] Example 1 The present invention provides a general airport remote service human-machine decision-making system based on a large model, as shown in the schematic diagram. Figure 1 As shown, it specifically includes: general airport operation data set, general airport operation knowledge graph, general airport large model, and general airport operation human-machine.

[0027] The system for constructing a general airport operation dataset is as follows: General airport data: Build a general airport operation data set, collect various types of multi-source heterogeneous general airport operation data through multiple systems, and pre-process and annotate the data; The general airport data module specifically includes: Data collection module: collects passenger data, X-ray data, meteorological data, surveillance data, facility and equipment data, operating rules, historical data, etc. through video surveillance, X-ray security equipment, sensors, networks, manual entry, etc.; Deploy high-definition cameras at security checkpoints to capture passenger facial images, subject to regulatory compliance, from various angles, expressions, and lighting conditions, for analysis of passenger demeanor and other characteristics. Install surveillance equipment throughout the security checkpoint area to record the entire passenger checkpoint process, including movement patterns and interactions with security personnel. Utilize general airport X-ray security equipment to collect perspective images of luggage items from various angles. Install sensors at key locations such as runways and taxiways to detect aircraft takeoffs, landings, and taxiing, and collect data such as runway occupancy time and aircraft position; Multiple meteorological observation stations set up in and around general airports collect real-time data on common meteorological elements such as temperature, humidity, air pressure, wind speed, and wind direction. These stations are equipped with high-precision data acquisition equipment and record data at standard intervals specified by the International Civil Aviation Organization (ICAO). Use weather radar to obtain information on precipitation intensity, cloud structure, wind shear, and more. Weather radar data can provide weather dynamics over a wider area and is crucial for monitoring severe weather conditions such as thunderstorms and heavy rain. Receive cloud image data from meteorological satellites to analyze cloud distribution, movement, and development trends. Satellite cloud images can visually display large-scale weather systems, providing a macro perspective for forecasting weather evolution. Collect airport aircraft location information, speed, heading, identification number, flight status, intended trajectory, etc. through ADS-B equipment at general airports; Collect airport facility and equipment operation data, communication equipment signal quality data, equipment operating parameters, link status, equipment temperature and voltage, etc. through the network; navigation equipment positioning accuracy, signal status, equipment fault status, etc.; monitoring equipment tracking data, system performance data, etc. Digitize paper-based airport operating rules and collect and store non-paper-based airport operating rules; Collecting years of historical meteorological records for general airports, including daily, monthly, and annual meteorological statistics, helps the model learn the long-term patterns and seasonal characteristics of weather changes.

[0028] Data processing module: processes all types of collected data, improves data quality, and provides data support for subsequent knowledge graph construction and model training; Data is cleaned. Since data collection may come from multiple channels or contain multiple records, there may be duplicate data. This is identified and deleted through algorithms based on unique identification fields. For data with missing values, methods such as deleting missing value records, filling with the mean, and interpolation based on similar data are used according to data characteristics and business needs. The data is checked for logical errors and format errors. For example, if the flight time is obviously unreasonable or the data format does not meet the specifications, these errors can be corrected by checking with other data sources and applying business rules. Grayscale and normalize the collected facial images, unify the image size, remove irrelevant background information, and extract key facial features, such as the location of the eyes, nose, and mouth. Frame extraction is performed on the surveillance video, key frames are selected, and the behavior and actions of the characters in the key frames are annotated, such as whether there are any abnormal behaviors and whether the person cooperates with security checks. Perform noise reduction on X-ray images, enhance image contrast, make object outlines clearer, and mark the category, location, and shape characteristics of various objects in the image; Convert data of different dimensions and value ranges into a unified standard format to make the data comparable in subsequent analysis; use one-hot encoding, label encoding, and other systems to convert categorical data, such as flight type, aircraft model, and passenger type, into numerical data; Data annotation module: formulate various data annotation rules, develop annotation tools, and organize professionals to perform annotation and review; Organize business experts and backbone personnel to formulate rules for marking images during security inspections based on laws and regulations and practical experience. For example, determine whether an item in an X-ray image is a prohibited item based on its shape and material; determine whether a passenger shows suspicious signs based on facial expressions and monitoring behavior; cooperate with meteorological experts and air traffic control personnel to formulate detailed weather airworthiness marking rules based on international and domestic aviation safety standards. For example, it is stipulated that visibility below a certain value or wind speed exceeding a specific range is considered unsuitable for flight. Professional labelers are assigned to label the pre-processed data according to the rules, marking the results corresponding to each data sample, such as labeling X-ray machine data with dangerous goods (knives, explosives, liquids, etc.), classification labels for prohibited items, and bounding box (Bounding Box) labels. After labeling is completed, multiple rounds of review are carried out to ensure the accuracy and consistency of the labeling.

[0029] General Airport Operation Knowledge Graph: Since large models may have hallucinations, and airport operations are highly related to safety, knowledge graphs are used to modify large models. A general airport operation knowledge graph is constructed through data extraction, knowledge extraction, knowledge fusion, and knowledge processing systems. The schematic diagram of the knowledge graph construction process is shown in the figure below. Figure 2 shown.

[0030] The general airport operation knowledge graph specifically includes: Data extraction module: This module extracts knowledge from structured data such as flight dynamics data and airport facility and equipment status. It also uses natural language processing technology to process the large amount of collected unstructured text data, including word segmentation, part-of-speech tagging, and named entity recognition, to prepare for subsequent knowledge extraction. Knowledge extraction module: extracts knowledge based on the general airport operation data set, extracting entities, attributes and relationships; Using named entity recognition technology, entities related to general airport operations, such as aircraft models, runway numbers, navigation equipment names, and meteorological elements, are extracted from the processed data. Through systems such as syntactic analysis and semantic role labeling, we can mine the relationships between entities for relationship extraction, such as "passenger - security check - direct release", "weather conditions - impact - flight operations", etc. For each entity, extract its relevant attributes, such as Figure 3 As shown, such as the attributes of security inspectors, X-ray security equipment, and attributes of items.

[0031] Knowledge fusion module: This module integrates knowledge extracted from different data sources to resolve issues such as entity homonyms and synonyms, ensuring the consistency and integrity of knowledge. It also conducts quality assessment on the integrated knowledge to check its accuracy, consistency, and completeness, and corrects and improves any existing problems. Entity alignment: matching and merging information representing the same entity across different data sources. For example, descriptions of the same aircraft may differ across different systems, requiring alignment using specific algorithms and rules to ensure knowledge consistency. Relationship fusion: Check and merge relationship information from different sources to eliminate relationship conflicts and redundancies. For example, for the "flight-stop-airport" and "flight-arrival-airport" relationships, if the stopover and arrival point to the same airport, they need to be properly merged. Attribute fusion is the process of fusing different attribute values ​​of the same entity and determining the final attribute value based on factors such as data reliability and authority.

[0032] Knowledge processing module: performs knowledge verification, knowledge reasoning and optimization on the established knowledge graph; Knowledge verification: conducts quality checks on the integrated knowledge, using rule verification, logical reasoning and other systems to check the accuracy, completeness and consistency of the knowledge. For example, check whether the logic of "the aircraft pushes out and then taxis" is correct; Knowledge reasoning: Based on existing knowledge and rules, reasoning is performed to acquire new knowledge. For example, based on "Runway A is under maintenance" and "The flight needs to use Runway A to land", it can be inferred that the flight may need to adjust the landing runway or be delayed. Knowledge graph optimization: Based on the results of knowledge verification and reasoning, the structure and content of the knowledge graph are optimized, such as adjusting entity relationships and supplementing missing knowledge, to improve the quality and practicality of the knowledge graph. Choose an appropriate knowledge graph storage method, such as a graph database or relational database, to efficiently support knowledge query and reasoning. Design a reasonable data storage structure and optimize storage performance to ensure that the knowledge graph can quickly respond to various query requests.

[0033] Large model of general airport: Since different operating positions at general airports use different data and operating rules, large models of general airport security, control command, meteorological services and operation monitoring are established respectively.

[0034] The general airport model specifically includes: General airport security inspection model: Security inspectors conduct security checks on passengers, baggage, and cargo to prevent prohibited items from entering. Image recognition is an important part of the security personnel's work, so a model for image recognition is constructed. Based on data characteristics and security inspection task requirements, a multimodal fusion model is adopted. For example, a convolutional neural network (CNN) is used to process facial images and X-ray images, and a recurrent neural network (RNN) or long short-term memory network (LSTM) is used to process the time series information of surveillance video. The different modal features are integrated through a fusion layer. When designing the model structure, the diversity of prohibited items and the complexity of image features in general airport security inspection scenarios are fully considered. The number and parameters of convolutional layers, pooling layers, and fully connected layers are appropriately set to effectively extract image features. The model was trained on airport security X-ray image data from a general airport dataset, which covers a variety of common and rare prohibited items. The images in the dataset have been annotated to clearly indicate the type and location of prohibited items. Divide the dataset into training, validation, and test sets. Use the training set to train the constructed model, setting appropriate hyperparameters such as learning rate, number of iterations, and batch size. Use optimization algorithms such as stochastic gradient descent (SGD), Adagrad, and Adam to continuously adjust model parameters and minimize the loss function, allowing the model to gradually learn accurate image recognition patterns on the training set. During training, regularly use the validation set to evaluate model performance to prevent overfitting. If the model performance no longer improves on the validation set, save the optimal model. The trained model is comprehensively evaluated using the test set, using metrics such as accuracy, recall, F1 score, and mean average precision (mAP) to measure its ability to identify various prohibited items. A confusion matrix is ​​used to visually display the model's predictions for different categories, analyzing situations where the model is prone to misjudgment and omission, such as confusion between similar items and difficulty detecting small objects. Build a general airport security knowledge graph containing knowledge about prohibited item categories, characteristics, relevant regulations, and security inspection procedures. Associate and align model predictions with the knowledge in the knowledge graph. For example, after the model identifies an item, it uses the knowledge graph to obtain detailed information about the item in security inspection regulations, assisting security personnel in making more accurate judgments. The semantic information in the knowledge graph is used to interpret and supplement the model output, improving the interpretability and reliability of the model results. For example, link items detected by the model (such as "lithium batteries") to entities in the knowledge graph and generate disposal recommendations based on regulations in the graph (such as "lithium batteries with a capacity greater than 100Wh are prohibited from checked baggage"). If "liquids" are detected, a graph query is performed on "liquid capacity limits." If the limit is exceeded, an unpacking inspection is triggered.

[0035] General Airport Control and Command Model: Airport control and command is responsible for ensuring safe and orderly aircraft takeoffs and landings and ground transportation within the airport, sequencing takeoffs and landings, planning taxi paths, and performing conflict detection and resolution. Feature engineering and model training are performed using multi-source data such as flight information, meteorological data, and airport airspace collected and preprocessed in the general airport dataset. Graph neural networks (GNNs) are used to establish airspace structures and dynamic relationships between flights (GNNs are machine learning models that process graph-structured data; graphs consist of nodes and edges, such as social networks and molecular structures; traditional neural networks like CNNs and RNNs process Euclidean data, such as images and sequences, while GNNs process non-Euclidean graph data). Four types of spatiotemporal graphs are constructed based on the geographical location and traffic flow attributes of airports. An adjacency matrix is ​​constructed for each graph, including the airport distance (Dis) matrix ADis, the scheduled flight (SF) matrix ASF, the historical similarity (HS) matrix AHS, and the functional relationship (FR) matrix AFR. These graphs are integrated into an atlas AAMRG={ADis,ASF,AHS,AFR}. The schematic diagram of the construction of a general airport multi-graph is shown in the figure below. Figure 4 As shown, the Transformer and other time series models are used to predict the airport operation situation; Based on the data scale and computing resources, determine the number and structure of model parameters, select a training system, set training hyperparameters, conduct training, evaluate model performance, adjust hyperparameters or optimize the model structure based on the indicators. For example, if the accuracy is low, reduce the learning rate or increase the number of model layers.

[0036] The general airport meteorological service large model uses meteorological observation data, radar data, satellite cloud images, and other meteorological data collected and preprocessed by the general airport data center for general airports and surrounding areas. According to the safety standards and operational requirements of general airports, thresholds are set for various meteorological factors. For example, visibility below a certain value and wind speed exceeding a certain range are considered to be meteorological conditions that do not meet the requirements. Data generation technology is used to generate sample data such as sandstorms and freezing rain to verify the model's warning accuracy under low visibility and high humidity conditions. The model inputs multi-source meteorological data and outputs whether the meteorological conditions meet operational requirements. The meteorological data from the general airport dataset is divided into training, validation, and test sets. Hyperparameters such as the learning rate and number of iterations are set, and an appropriate large-scale model architecture is selected for training. CNN processes satellite cloud images and radar imagery, Transformer or LSTM processes temporal changes in meteorological parameters, and the BERT fine-tuned model parses METAR / TAF weather reports. During training, the model's performance metrics on the validation set, such as mean squared error and accuracy, are monitored in real time. Hyperparameters are adjusted based on these metrics to prevent overfitting or underfitting. Comparative experiments are conducted with other traditional meteorological forecasting models and existing advanced models to verify the superiority of the constructed large-scale model. Invite meteorological experts, general airport staff, etc. to evaluate the model's prediction results and service functions, collect their opinions and suggestions, and understand the model's performance and demand satisfaction in actual applications; based on expert feedback, adjust and optimize the model to make the model's output results more in line with human cognition and actual application needs, such as adjusting the expression of prediction results and adding explanatory information.

[0037] Large-scale model for general airport operation monitoring: General airport operation monitors are able to monitor the real-time status parameters (voltage, temperature, signal strength) of communication (VHF radio), navigation (ILS / VOR / DME), surveillance (ADS-B / radar) and other equipment, and promptly detect and address abnormalities. Use a data generation model to generate device anomaly data with a small sample size, such as electromagnetic interference from thunderstorms and device response delays caused by low temperatures. Predefine device thresholds to enhance model robustness. Communication equipment: When visibility is less than 800 meters and wind speed is greater than 15m / s, the reliability of the VHF signal will be reduced. Navigation equipment: When precipitation intensity is greater than 50mm / h, the ILS glide path signal tolerance threshold will be increased by 30%; The model is trained and evaluated based on historical facility and equipment operation data in the general airport dataset, abnormal data annotated by equipment maintenance, and generated data.

[0038] After the general airport large model is built and trained, human alignment is required. Human alignment is to ensure that the behavior of the large language model is consistent with human values, true intentions and social ethics, and to avoid generating biased or erroneous content. The 3H standards of human alignment are further explained: usefulness, honesty and harmlessness, and reinforcement learning based on human feedback (RLHF) is used as the key technology.

[0039] General airport operation post human-machine decision-making module: establish general airport security inspection, control command, meteorological service and operation monitoring human-machine modules respectively.

[0040] The general airport operation position human-machine decision-making module specifically includes: General airport security human-machine module: When the output of the general airport security large model needs to be further unpacked, humans will decide whether security personnel are needed to conduct a secondary inspection. Human security personnel will conduct a secondary inspection of passengers and items.

[0041] General Airport Control Command Human-Machine Module: This module builds a large model that combines reinforcement learning with human feedback (RLHF) with manual decision-making. The model generates control strategies based on airport operational status, integrating perception, reasoning, and decision-making. Airport controllers then decide whether to implement recommendations or make adjustments.

[0042] General Airport Weather Service Human-Machine Module: Based on the safety standards and operational requirements of general airports, thresholds are set for various meteorological factors. For example, visibility below a certain value or wind speed exceeding a certain range is considered to indicate that the required weather conditions are not met. The general airport weather service large model obtains the latest weather data in real time, compares the forecast results with the set thresholds, and determines whether the weather conditions meet the requirements. If the weather conditions do not meet the requirements, a prompt is issued, and the weather service personnel will re-check whether the weather conditions are airworthy.

[0043] General airport operation monitoring human-machine module: The general airport operation monitoring model monitors the status of various communication, navigation, surveillance and other facilities and equipment at the airport, determines the equipment operation status, and issues alarm prompts and provides disposal suggestions when the equipment operates abnormally. Equipment monitoring personnel monitor and repair the alarm equipment.

[0044] The general airport data collected by the large model include: video surveillance images, X-ray security images, ADS-B data, flight plans, meteorological observation data of general airports and surrounding areas, radar data, satellite cloud images, communication equipment status information, navigation equipment status information, monitoring equipment status information and other real-time and historical meteorological data.

[0045] A large-scale, general-purpose airport security inspection model was constructed and compared with traditional security inspection. Traditional security inspection relies on manual experience to predict passenger flow peaks and maintain a fixed number of open security lanes. The large-scale security inspection model dynamically adjusts the number of open security lanes. This reduced the average passenger wait time during peak hours from 30 minutes to 15 minutes, and increased lane utilization to 80%. The traditional model relies on fixed scheduling, resulting in both labor waste and excessive workload. With the large-scale model, machines conduct inspections, and when further baggage opening is required, manual review is reduced from 100% to 1%, reducing labor input by 80%.

[0046] A large-scale universal airport control and command model was constructed and compared with the traditional model. The traditional model relies on manual coordination of data from multiple systems, resulting in significant decision lags. The large-scale model dynamically controls and resolves conflicts in real time, reducing the probability of conflict by 40% and increasing response speed during complex maneuvers by 60%.

[0047] A large-scale model for general airport meteorological services was constructed and compared with traditional models. Traditional manual analysis relies on manual interpretation of multiple data sources, such as radar and satellite imagery. This limited data processing capacity makes it difficult to capture the spatiotemporal evolution of complex meteorological elements. Manual analysis requires hours of data integration, which can lead to delays in handling emergencies. By integrating multi-source data, the large-scale model has increased the lead time for severe convective weather warnings from 40 minutes to over 80 minutes, achieving an accuracy rate of 92%. A large-scale model for general airport operational monitoring was constructed and compared with traditional models. Traditional monitoring models rely on single-point sensors and manual inspections. For example, airport VOR / DME equipment is only manually inspected twice daily, resulting in a 15% miss detection rate. Radar signal anomalies require manual retrieval of data from three independent systems, taking over 20 minutes. The large-scale model monitoring model integrates various parameters, including radar echoes, beacon signals, and equipment power consumption, reducing the frequency of equipment status updates from minutes to 100 milliseconds, increasing anomaly detection coverage to 99%.

Claims

1. A general airport remote service human-machine decision-making system based on a large model, characterized by: It includes a general airport large model and a general airport operation post human-machine decision module based on the general airport large model; The general airport large model specifically includes a general airport security inspection large model, a general airport control and command large model, a general airport meteorological service large model and a general airport operation monitoring large model; The general airport operation post human-machine decision-making module specifically includes a general airport security inspection module, a general airport control and command module, a general airport meteorological service module and a general airport operation monitoring module.

2. The large model-based general airport remote service human-machine decision-making system according to claim 1 is characterized in that: The general airport large model is constructed based on the general airport dataset; The steps of constructing the general airport data set specifically include collecting data from set data sources to obtain multi-source heterogeneous general airport operation data, and performing data preprocessing and data labeling on the multi-source heterogeneous general airport operation data.

3. The large model-based general airport remote service human-machine decision-making system according to claim 2 is characterized in that: The data sources specifically include video surveillance, X-ray security equipment, ADS-B equipment, airspace environment, facilities and equipment, and historical data.

4. The large model-based universal airport remote service human-machine decision-making system according to claim 1 is characterized in that: The general airport model is also explained and supplemented by the knowledge graph; The steps of constructing the knowledge graph specifically include extracting data from the data source, and performing knowledge extraction, knowledge fusion, and knowledge processing based on the results of the data extraction.

5. The large model-based universal airport remote service human-machine decision-making system according to claim 4 is characterized in that: The knowledge graph constructed is specifically a general airport security knowledge graph, which is used to explain and supplement the results output by the airport security large model.

6. The large model-based universal airport remote service human-machine decision-making system according to claim 1 is characterized in that: The general airport security inspection model specifically adopts a multimodal fusion model to obtain the category and location of prohibited items by processing the time series information of facial images, X-ray images and surveillance videos; The general airport control and command model specifically uses a graph neural network to establish the airspace structure and the dynamic relationship between flights to obtain the airport control strategy; The general airport weather service large model obtains airport weather forecast data by processing satellite images, radar images, time series changes of meteorological parameters and meteorological messages; The general airport operation monitoring model is specifically used to monitor the real-time status parameters of airport equipment.

7. The large model-based universal airport remote service human-machine decision-making system according to claim 6 is characterized in that: After the construction of the general airport model is completed, the airport model needs to be trained and aligned by humans.

8. The large model-based universal airport remote service human-machine decision-making system according to claim 1 is characterized in that: The general airport security inspection module performs a secondary inspection based on the category and location of prohibited items obtained by the general airport security inspection large model; The general airport control and command module combines the control strategy obtained by the general airport control and command large model with manual decision-making to obtain the final control strategy; The general airport weather service module determines whether the airport weather forecast data exceeds a set threshold based on the general airport weather model; The general airport operation monitoring module monitors and maintains the alarm equipment obtained by the general airport operation monitoring large model.

9. The large model-based universal airport remote service human-machine decision-making system according to claim 8 is characterized in that: The general airport meteorological service module sets the set threshold according to the safety standards and operation requirements of the general airport.

Images