Civil aviation data automatic classification and management method based on large model training

Abstract

The invention relates to a civil aviation data automatic classification and management method based on large model training. The method comprises the following steps: collecting navigation notification information from a plurality of official platforms; defining professional classification term dictionary interpretation according to specific use scenes and meanings in the notification information; carrying out manual annotation processing on the obtained navigation announcement data; performing large model training on the civil aviation notification information by using the marked data; deploying the trained model to a data application platform for real-time processing; the model automatically classifies the announcement into a preset professional field category according to the property of the announcement; and a manual verification system is adopted to verify and adjust the classification accuracy. The invention further relates to a corresponding device, a processor and a storage medium thereof. By adopting the civil aviation data automatic classification and management method and device based on large model training, the processor and the storage medium thereof, the efficiency of aviation safety management and operation decision can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of computer software, and in particular to the field of artificial intelligence large model data management, and specifically refers to a method, device, processor and computer-readable storage medium for automatic classification and management of civil aviation data based on large model training. Background Art

[0002] With the rapid development of the aviation industry, the number and complexity of navigation notices are increasing, and traditional manual processing methods can no longer meet the needs of efficiency and accuracy. Therefore, it is particularly important to use machine learning and natural language processing technologies to automate the processing of these notices. These technologies can identify and understand the content of the notices and classify them into preset professional fields, thereby improving the efficiency and accuracy of information processing. The latest progress includes the application of deep learning models, which are trained with large amounts of data and can provide higher classification confidence and more sophisticated management modes. In addition, with the advancement of artificial intelligence technology, these systems are becoming more intelligent and can adapt to changing data patterns and notice formats, providing strong support for aviation safety management and operational decisions.

[0003] In this regard, the existing technology mainly relies on manual methods to classify and manage civil aviation notices, including flight plans, safety warnings, airport operation updates, etc., which are usually read by professionals and filed into corresponding categories based on experience. This method is inefficient and susceptible to human errors. With the development of big data and artificial intelligence technologies, some automated tools have begun to be introduced, such as simple classification systems based on keyword matching, which can reduce the manual burden to a certain extent. However, these tools often lack the ability to deeply understand the semantics of the notice content, resulting in insufficient classification accuracy.

[0004] On this basis, since the existing civil aviation notice information processing technology mainly relies on manual operation, the manual reading and classification of notices is inefficient, especially when faced with a large number of notices, it is difficult for staff to respond quickly, resulting in delays in information processing. Due to the intervention of human factors, the accuracy of classification is affected by subjective judgment, and different staff may have different understandings of the classification of the same notice, which leads to inconsistency in classification results. In addition, the manual processing method is difficult to adapt to changes in the notice format. Whenever a new notice format appears, staff need to be retrained, which not only increases costs but also affects the continuity of processing. The manual method cannot effectively process large-scale data. As the amount of data increases, the accuracy and efficiency of manual operations will decrease. The existing technology lacks a real-time monitoring and feedback mechanism, and cannot verify and adjust the classification results in a timely manner, which limits the further improvement of system performance. Therefore, the existing technology faces challenges in efficiency, accuracy and scalability when processing civil aviation notice information. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a method, device, processor and computer-readable storage medium for automatic classification and management of civil aviation data based on large model training.

[0006] In order to achieve the above objectives, the method, device, processor and computer-readable storage medium for automatic classification and management of civil aviation data based on large model training of the present invention are as follows:

[0007] The main feature of the automatic classification and management method of civil aviation data based on large model training is that the method comprises the following steps:

[0008] (1) Obtain the latest NOTAMs from the official channels of the Civil Aviation Administration of China, airport announcements, airline notices, and related aviation information release platforms, and collect key information including flight plan changes, weather conditions, air traffic control instructions, and safety warnings;

[0009] (2) Define dictionary interpretations of professional classification terms based on the specific usage scenarios and meanings in the notice information;

[0010] (3) Manually annotate the acquired NOTAM data;

[0011] (4) Use labeled data to train large models for civil aviation announcement information;

[0012] (5) Deploy the trained model to the data application platform for real-time processing;

[0013] (6) the model will automatically classify the notification into pre-set professional field categories based on the nature of the notification;

[0014] (7) Use a manual verification system to verify and adjust the classification accuracy to improve the reliability of the model.

[0015] Preferably, the step (2) is specifically as follows:

[0016] The acquired navigation notice information is set with a notice type tag to define a professional classification term dictionary, which is used to explain in detail the specific meaning and usage scenario of each classification term.

[0017] Preferably, the step (3) is specifically as follows:

[0018] Aviation experts will assign the currently acquired notices to corresponding categories based on a defined dictionary of classification terms. During the labeling process, experts need to carefully read each notice, understand its meaning, and classify it into the most appropriate category.

[0019] Preferably, the step (4) is specifically as follows:

[0020] The labeled data is used to train a large machine learning model, which learns how to automatically classify new announcements into the correct categories by analyzing the content of the announcement text, and uses LoRA to continuously adjust its corresponding parameters during the training process to minimize classification errors.

[0021] Preferably, the step (5) is specifically as follows:

[0022] Once the model training is completed and verified, it will be deployed on the data application platform, which will process navigation notices in real time and use the trained model for automatic classification processing to complete the model deployment.

[0023] Preferably, the step (6) is specifically as follows:

[0024] The deployed model is able to identify notifications related to flight safety, updates related to airport operations, or notifications related to changes in aviation regulations. The model will automatically classify the notification into preset professional field categories based on the nature of the notification for automated classification processing.

[0025] Preferably, the step (7) is specifically as follows:

[0026] Aviation experts review the model’s classification results, including assessing the model’s confidence and adjusting the classification results when necessary, to improve the model’s accuracy and reliability.

[0027] The main features of the automatic classification and management device for civil aviation notice information based on large model training are that the device comprises:

[0028] a processor configured to execute computer-executable instructions;

[0029] A memory stores one or more computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the above-mentioned method for automatic classification and management of civil aviation announcement information based on large model training are implemented.

[0030] The processor for automatic classification and management of civil aviation announcement information based on large model training has the main feature that the processor is configured to execute computer executable instructions, and when the computer executable instructions are executed by the processor, the steps of the above-mentioned method for automatic classification and management of civil aviation announcement information based on large model training are implemented.

[0031] The main feature of the computer-readable storage medium is that a computer program is stored thereon, and the computer program can be executed by a processor to implement the steps of the above-mentioned method for automatic classification and management of civil aviation announcement information based on large model training.

[0032] The method, device, processor and computer-readable storage medium for automatic classification and management of civil aviation data based on large model training of the present invention are adopted, and the high efficiency and high accuracy of civil aviation notice information processing are achieved by constructing a professional terminology dictionary, using machine learning technology to train models, automating classification processes and continuous manual verification and feedback loops. In practical applications, this technical solution can reduce manual processing time by about 60%, improve classification accuracy to more than 95%, and shorten response time to real-time automatic processing, greatly improving the timeliness of information processing; through automation, human errors are reduced, potential safety risks caused by classification errors are reduced, and aviation safety levels are improved. Economically, this method reduces labor costs and is expected to save about 50% of related expenses each year. Technically, the application of this method improves the intelligent level of data processing and provides strong technical support for big data analysis and decision support in the aviation industry. Therefore, this technical solution not only achieves innovation in technology, but also brings significant benefits at the economic level. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The present invention is a flowchart of the method for automatic classification and management of civil aviation data based on large model training.

[0034] Figure 2 A schematic diagram of defining professional classification terms for the automatic classification and management method of civil aviation data based on large model training of the present invention.

[0035] Figure 3 A schematic diagram of data labeling for automatic classification and management of civil aviation data based on large model training according to the present invention.

[0036] Figure 4 This is a schematic diagram of large model training for the automatic classification and management method of civil aviation data based on large model training of the present invention.

[0037] Figure 5 A schematic diagram of deploying a large model for the automatic classification and management method of civil aviation data based on large model training of the present invention. DETAILED DESCRIPTION

[0038] In order to more clearly describe the technical content of the present invention, further description is given below in conjunction with specific embodiments.

[0039] Before describing in detail embodiments according to the present invention, it should be noted that, hereinafter, the terms "comprises", "includes" or any other variations are intended to cover non-exclusive inclusion, whereby a process, method, article or apparatus comprising a series of elements includes not only these elements, but also other elements not explicitly listed or inherent to such process, method, article or apparatus.

[0040] See also Figure 1 As shown, the method for automatic classification and management of civil aviation data based on large model training, wherein the method comprises the following steps:

[0041] (1) Obtain the latest NOTAMs from the official channels of the Civil Aviation Administration of China, airport announcements, airline notices, and related aviation information release platforms, and collect key information including flight plan changes, weather conditions, air traffic control instructions, and safety warnings;

[0042] (2) Define dictionary interpretations of professional classification terms based on the specific usage scenarios and meanings in the notice information;

[0043] (3) Manually annotate the acquired NOTAM data;

[0044] (4) Use labeled data to train large models for civil aviation announcement information;

[0045] (5) Deploy the trained model to the data application platform for real-time processing;

[0046] (6) the model will automatically classify the notification into pre-set professional field categories based on the nature of the notification;

[0047] (7) Use a manual verification system to verify and adjust the classification accuracy to improve the reliability of the model.

[0048] In a preferred embodiment of the present invention, the step (2) is specifically as follows:

[0049] The acquired navigation notice information is set with a notice type tag to define a professional classification term dictionary, which is used to explain in detail the specific meaning and usage scenario of each classification term.

[0050] In a preferred embodiment of the present invention, the step (3) is specifically:

[0051] Aviation experts will assign the currently acquired notices to corresponding categories based on a defined dictionary of classification terms. During the labeling process, experts need to carefully read each notice, understand its meaning, and classify it into the most appropriate category.

[0052] In a preferred embodiment of the present invention, the step (4) is specifically as follows:

[0053] The labeled data is used to train a large machine learning model, which learns how to automatically classify new announcements into the correct categories by analyzing the content of the announcement text, and uses LoRA to continuously adjust its corresponding parameters during the training process to minimize classification errors.

[0054] In a preferred embodiment of the present invention, the step (5) is specifically:

[0055] Once the model training is completed and verified, it will be deployed on the data application platform, which will process navigation notices in real time and use the trained model for automatic classification processing to complete the model deployment.

[0056] In a preferred embodiment of the present invention, the step (6) is specifically as follows:

[0057] The deployed model is able to identify notifications related to flight safety, updates related to airport operations, or notifications related to changes in aviation regulations. The model will automatically classify the notification into preset professional field categories based on the nature of the notification for automated classification processing.

[0058] In a preferred embodiment of the present invention, the step (7) is specifically as follows:

[0059] Aviation experts review the model’s classification results, including assessing the model’s confidence and adjusting the classification results when necessary, to improve the model’s accuracy and reliability.

[0060] The following will describe in detail the various steps of this technical solution through specific embodiments:

[0061] Step 1: Collect NOTAMs

[0062] In the patented process of automated classification and refined management of civil aviation notice information, the first step is to collect navigation notices. This involves obtaining the latest navigation notices from the official channels of the Civil Aviation Administration, airport announcements, airline notices, and related aviation information release platforms. These notices may contain key information such as flight plan changes, weather conditions, air traffic control instructions, safety warnings, etc. The collection process needs to ensure the integrity and timeliness of the data in order to provide accurate raw data for subsequent classification and analysis.

[0063] Step 2: Define dictionary definitions for specialized taxonomy terms

[0064] like Figure 2As shown, in the second step, we defined a professional terminology dictionary containing 50 categories. This dictionary is the basis of the classification algorithm, which explains in detail the specific meaning and usage scenarios of each classification term. For example, the flight safety category may include terms such as "emergency landing" and "air approach", while the airport operation category may cover "runway closure" and "flight delay". This dictionary not only helps standardize the classification process, but also ensures consistency and communicability between different users and systems. The diagram is as follows:

[0065] “

[0066] Announcement Type Label Explanation

[0067] Airport closed Airport closed, Airport operating hours, Airport not in service

[0068] No alternate landings are accepted. No alternate landings are accepted. No alternate landings are accepted for non-emergency situations. No aircraft with a wingspan of more than xx meters are accepted.

[0069] Operation standard changes Landing and takeoff minimum standards are adjusted, and the minimum operation standard is adjusted. The landing standard is unavailable.

[0070] Airport restrictions: Ascending and descending within xxkm of the airport, the airport cannot be used for certain flights, VIP activities

[0071] Airport operating hours changed

[0072] Airport fuel fuel service is not provided, no fuel supply

[0073] Airport equipment and services: Air truck, power truck, snow removal equipment, weather radar, low-level wind shear radar are not working

[0074] Fire protection level Changes in airport fire protection level

[0075] Airport construction WIP, CONSTRUCTION, non-stop construction, if there is information about the height / height of the equipment, it should also be identified as an "obstacle"

[0076] STRIKE, STRIKE

[0077] Runway closed RWY xx / xx CLSD, RUNWAY CLOSURE PERIOD CHG, RWY xx / xx cannot be used for landing

[0078] New RWY / parallel TWY New parallel taxiway, new runway

[0079] Runway operation schedule changed. Runway closure maintenance schedule changed. RWY xx / xx NOT CLSD. Runway closure schedule suspended.

[0080] Runway restrictions: Is the runway wet and contaminated? Does the runway provide low visibility landings? Which taxiway to use to enter or exit the runway?

[0081] Runway data changed TORATODAASDALDA length changed

[0082] Bearing strength PCN related ...

[0084] ”

[0085] Step 3: Label the data

[0086] like Figure 3 As shown in Figure 1, the third step is to manually annotate the collected NOTAM data. This step requires professional aviation experts who assign the content of the NOTAM to the corresponding categories based on the defined classification term dictionary. During the annotation process, experts need to carefully read each NOTAM, understand its meaning, and classify it into the most appropriate category. This step is crucial for training an accurate classification model because it provides the basic data for model learning.

[0087] S3-1 Prepare the annotation environment so that experts can easily read the content of the notice and annotate it. Ensure that all data that needs to be annotated has been collected and prepared, including navigation notices obtained from different channels.

[0088] S3-2 assigns category labels based on the understood content. Experts assign each notice to one of the 50 categorized professional terminology dictionary categories in step 2.

[0089] S3-3 Annotation Results After the initial annotation, cross-validation review is conducted, and different experts check the consistency of the same batch of data. In case of inconsistent annotation, discussion and review are carried out to reach a consensus.

[0090] S3-4 The annotation results are organized into a structured data format CSV file, and the completeness and accuracy of the annotation data are checked to ensure that there are no omissions or errors for subsequent model training.

[0091] The following is an example of the file content:

[0092] “

[0093] Notice type label marked original sample

[0094] ATIS changes ATIS126.400MHZ U / SDUE TO MAINTENANCE.

[0095] ATIS changeD-ATIS AND PDC(PRE DEPARTURE CLEARANCE)DATALINKSERVICESIMPLEMENTED ON TEST,DO NOT USE FOR OPERATIONAL PURPOSES REF AIP AD2LIMC 1-15

[0096] ATIS Change: The ATIS / D-ATIS (working frequency: 126.25MHZ) is not available for use due to software upgrade.

[0097] ATIS changes ATIS / D-ATIS U / SDUE TO MAINT.

[0098] ATIS CHANGE D-ATIS COMPLIANT WITH EUROCAE ED-89AAVBL FOR OPR PURPOSES.D-ATIS CONTRACT AVBL O / R FOR AMAX DURATION OF 30MIN.RMK:PRE-DEPARTURE CLEARANCEDATALINK ON TEST,DO NOT USE FOR OPR PURPOSES.REF AIP AD 2LIMC 1-15

[0099] ATIS changesATIS FREQUENCY 127.000MHZ BE USED FOR CHINESE(ARRIVAL / DEPARTURE).ATIS FREQUENCY 128.600MHZ BE USED FOR ENGLISH(ARRIVAL / DEPARTURE).

[0100] EOSID key station CENCUN VOR / DME'CEN'114.6MHZ / CH93X U / S.

[0101] EOSID key station Shaoguan VOR / DME'UQN'113.65MHZ / CH83Y is restricted as follows: 1. The signal between the VOR / DME 225 degree radial line clockwise to the 315 degree radial line is unavailable. 2. The VOR 005 degree radial line 18NM to 22NM is unavailable.

[0102] GPS related navigation station BELOZERSK VORDME BLZ 110.40MHZ: VOR U / S.

[0103] The following ILS-related procedures are not available due to calibration: 1. RNAV CAT-Ⅱ / ⅢILS / DME ZRWY02 instrument approach procedure. 2. CAT-Ⅱ / ⅢILS / DME Y RWY02 instrument approach procedure.

[0104] ILS related RWY08 ILS U / SDUE TO FLTCK.

[0105] ILS related FLW APPROACH PROCEDURES U / S:1.RNAV CAT-II / III ILS / DME ZRWY02.2.CAT-II / III ILS / DME Y RWY02.

[0106] ILS related INSTRUMENT LANDING SYSTEM RWY 17L UNSERVICEABLE DUE TO FLTCKIN PROGRESS RMK:ARR TFC EXPECT RNP 17L APCH REF AIP AD 2LIMC 1-16 ...

[0108] ”

[0109] Step 4: Training Data

[0110] like Figure 4 As shown in the figure, in the fourth step, we use the labeled data to train an open source large model glm-4-9b-chat. This model learns how to automatically classify new announcements into the correct category by analyzing the content of the announcement text. During the training process, the model will continuously adjust its parameters to minimize classification errors.

[0111] S4-1 Data preprocessing requires data cleaning and conversion of the labeled data into json format to ensure data consistency {"messages":[{"role":"user","content":""},{"role":"assistant","content":""}]}. In addition, natural language processing steps such as text segmentation and stem extraction may also be required to simplify the input of the model.

[0112] Here is an example:

[0113] “

[0114] {"messages":[{"role":"user","content":"ATIS FREQUENCY 127.000MHZ BEUSED FOR CHINESE(ARRIVAL / DEPARTURE).ATIS FREQUENCY 128.600MHZ BE USED FORENGLISH(ARRIVAL / DEPARTURE)."},{"role":"assistant","content":"ATIS changes"}]}

[0115] {"messages":[{"role":"user","content":"CENCUN VOR / DME'CEN'114.6MHZ / CH93XU / S."},{"role":"assistant","content":"EOSID key station"}]}

[0116] {"messages":[{"role":"user","content":"Shaoguan VOR / DME'UQN'113.65MHZ / CH83Y is restricted as follows:1.VOR / DME 225 degree radial line clockwise to 315 degree radial line signal is unavailable.2.VOR 005 degree radial line 18NM to 22NM is unavailable."},{"role":"assistant","content":"EOSID key station"}]} ...

[0118] ”

[0119] S4-2 implements real-time monitoring and logging mechanisms. Using the deep learning visualization tool Wandb to monitor the log system can track the model's training indicators, training learning rate, system load, etc. Logging is used to record the system's operating status and any abnormal conditions, which is crucial for subsequent troubleshooting and system optimization.

[0120] During the S4-3 model training process, the model will learn how to predict the category of the notification based on the input features. In order to optimize the model performance, Figure 4 By tuning parameters such as learning rate, neuron parameters, network depth, etc., the model can gradually learn how to accurately and automatically classify new announcements into the correct categories, thereby realizing the automatic classification of civil aviation announcement information.

[0121] Step 5: Deploy the trained large model

[0122] like Figure 5As shown in Figure 1, once the model training is completed and verified, the fifth step is to deploy it on a data application platform. This platform will process real-time navigation notices and use the trained model for automatic classification. The deployed model needs to be able to handle large-scale data streams and return classification results quickly and accurately.

[0123] Model integration First, you need to integrate the trained model node into glm-4-9b-chat and name it glm-4-9b-aviation. You need to ensure that the integration of the model does not affect the system function, and load and call the system with the python command line for integration testing and performance testing. Select the local server deployment environment and deploy and optimize based on the ubuntu system cuda12 version.

[0124] Step 6: Automatic Classification Notice

[0125] In actual applications, the deployed model will automatically classify the notifications into preset professional field categories based on the nature of the notifications. The model can identify notifications related to flight safety, updates related to airport operations, or notifications related to changes in aviation regulations. In addition, the model can also provide confidence for each classification to help users understand the reliability of the classification. The confidence calculation algorithm is based on the probability distribution predicted by the model and is estimated by statistically analyzing the performance of the model on the validation set.

[0126] The model deployed by S6-1 began to receive real-time navigation notice data and input the notice text into the model through the data application platform API interface. The model determined the category of each notice through its learned features and pattern recognition.

[0127] S6-2 Confidence Assessment For each classification result given by the model, the Top-p algorithm (kernel sampling) is used to calculate a confidence score based on the probability distribution predicted by the model. Confidence reflects the certainty of the model in its classification decision. High confidence means that the model strongly correlates the classification result with the database example, while low confidence may require further manual review.

[0128] S6-3 sets the three most relevant results as output and the confidence threshold as 0.7 based on the automatic classification of the model. When the confidence is lower than the threshold, the classification result will be marked as pending verification and submitted to the manual verification system.

[0129] Step 7: Human verification and feedback

[0130] In actual applications, in order to ensure the accuracy of classification, manual verification of the system is involved. This step requires professional aviation experts to review the classification results of the model. They will evaluate the confidence of the model and adjust the classification results if necessary. This provides feedback for the continuous improvement of the model.

[0131] Repeat steps 3-6 above:

[0132] As new NOTAMs are released and model performance is monitored, we will continue to repeat steps 3 to 6 to continuously optimize the model's classification capabilities. This iterative process ensures that the model can adapt to new NOTAM formats and content changes, maintaining high confidence and accuracy in classification results. Through this continuous optimization, we can ensure that the automated classification and refined management system for civil aviation notification information is always in the best condition.

[0133] The automatic classification and management device for civil aviation notice information based on large model training, wherein the device comprises:

[0134] a processor configured to execute computer-executable instructions;

[0135] A memory stores one or more computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the above-mentioned method for automatic classification and management of civil aviation announcement information based on large model training are implemented.

[0136] The processor for automatic classification and management of civil aviation announcement information based on large model training, wherein the processor is configured to execute computer executable instructions, and when the computer executable instructions are executed by the processor, the steps of the above-mentioned method for automatic classification and management of civil aviation announcement information based on large model training are implemented.

[0137] The computer-readable storage medium stores a computer program thereon, and the computer program can be executed by a processor to implement the steps of the above-mentioned method for automatic classification and management of civil aviation announcement information based on large model training.

[0138] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0139] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution device.

[0140] A person of ordinary skill in the art may understand that all or part of the steps of the method for implementing the above-mentioned embodiment may be completed by instructing the relevant hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one of the steps of the method embodiment or a combination thereof.

[0141] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0142] In the description of this specification, the description with reference to the terms "an embodiment", "some embodiments", "example", "specific example", or "embodiment" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0143] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

[0144] The method, device, processor and computer-readable storage medium for automatic classification and management of civil aviation notice information based on large model training of the present invention are adopted, and the high efficiency and high accuracy of civil aviation notice information processing are achieved by constructing a professional terminology dictionary, using machine learning technology to train models, automating classification processes and continuous manual verification and feedback loops. In practical applications, this technical solution can reduce manual processing time by about 60%, improve classification accuracy to more than 95%, and shorten response time to real-time automatic processing, greatly improving the timeliness of information processing; through automation, human errors are reduced, potential safety risks caused by classification errors are reduced, and aviation safety levels are improved. Economically, this method reduces labor costs and is expected to save about 50% of related expenses each year. Technically, the application of this method improves the intelligent level of data processing and provides strong technical support for big data analysis and decision support in the aviation industry. Therefore, this technical solution not only achieves innovation in technology, but also brings significant benefits at the economic level.

[0145] In this specification, the present invention has been described with reference to specific embodiments thereof. However, it is apparent that various modifications and variations may be made without departing from the spirit and scope of the present invention. Therefore, the specification and drawings should be regarded as illustrative rather than restrictive.

Claims

1. A method for automatic classification and management of civil aviation data based on large model training, characterized in that: The method comprises the following steps: (1) Obtain the latest NOTAMs from the official channels of the Civil Aviation Administration of China, airport announcements, airline notices, and related aviation information release platforms, and collect key information including flight plan changes, weather conditions, air traffic control instructions, and safety warnings; (2) Define dictionary interpretations of professional classification terms based on the specific usage scenarios and meanings in the notice information; (3) Manually annotate the acquired NOTAM data; (4) Use labeled data to train large models for civil aviation announcement information; (5) Deploy the trained model to the data application platform for real-time processing; (6) the model will automatically classify the notification into pre-set professional field categories based on the nature of the notification; (7) Use a manual verification system to verify and adjust the classification accuracy to improve the reliability of the model.

2. The method for automatic classification and management of civil aviation data based on large model training according to claim 1 is characterized in that: The step (2) is specifically as follows: The acquired navigation notice information is set with a notice type tag to define a professional classification term dictionary, which is used to explain in detail the specific meaning and usage scenario of each classification term.

3. The method for automatic classification and management of civil aviation data based on large model training according to claim 1 is characterized in that: The step (3) is specifically as follows: Aviation experts will assign the currently acquired notices to corresponding categories based on a defined dictionary of classification terms. During the labeling process, experts need to carefully read each notice, understand its meaning, and classify it into the most appropriate category.

4. The method for automatic classification and management of civil aviation data based on large model training according to claim 1 is characterized in that: The step (4) is specifically as follows: The labeled data is used to train a large machine learning model, which learns how to automatically classify new announcements into the correct categories by analyzing the content of the announcement text, and uses LoRA to continuously adjust its corresponding parameters during the training process to minimize classification errors.

5. The method for automatic classification and management of civil aviation data based on large model training according to claim 4 is characterized in that: The step (5) is specifically as follows: Once the model training is completed and verified, it will be deployed on the data application platform, which will process navigation notices in real time and use the trained model for automatic classification processing to complete the model deployment.

6. The method for automatic classification and management of civil aviation data based on large model training according to claim 5 is characterized in that: The step (6) is specifically as follows: The deployed model is able to identify notifications related to flight safety, updates related to airport operations, or notifications related to changes in aviation regulations. The model will automatically classify the notification into preset professional field categories based on the nature of the notification for automated classification processing.

7. The method for automatic classification and management of civil aviation data based on large model training according to claim 6 is characterized in that: The step (7) is specifically as follows: Aviation experts review the model’s classification results, including assessing the model’s confidence and adjusting the classification results when necessary, to improve the model’s accuracy and reliability.

8. An automatic classification and management device for civil aviation data based on large model training, characterized in that: The device comprises: a processor configured to execute computer-executable instructions; A memory stores one or more computer executable instructions, which, when executed by the processor, implement the steps of the method for automatic classification and management of civil aviation data based on large model training as described in any one of claims 1 to 7.

9. A processor for automatic classification and management of civil aviation data based on large model training, characterized in that: The processor is configured to execute computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the method for automatic classification and management of civil aviation data based on large model training described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer program can be executed by a processor to implement the steps of automatic classification and management of civil aviation data based on large model training as described in any one of claims 1 to 7.

Images