Airworthiness certification procedure automation system

Abstract

The present invention provides an airworthiness certification procedure automation system for an aircraft or aircraft parts, comprising: a storage unit storing regulation information for the airworthiness certification procedure and hierarchical structure data for documents required at each stage of the airworthiness certification procedure; a document input unit receiving a document file from a user; a dual verification unit including a rule-based document verification unit that detects formal errors in the document file input through the document input unit, and a validity determination unit that determines, through language model-based inference, whether the content of the document file semantically conforms to the hierarchical structure data and the regulations of the airworthiness certification procedure; an error type classification unit that classifies error types in the document file based on the verification results of the dual verification unit; and a procedure control unit that controls the procedure by applying a predefined state application rule based on the judgment results of the dual verification unit and the error type classification unit to apply the progress state of the airworthiness certification procedure corresponding to the document file as one of a state capable of proceeding, a state requiring supplementation, or a state suspended.

Description

Technology Field

[0001] The present invention relates to an airworthiness certification procedure automation system. More specifically, it relates to an airworthiness certification procedure automation system capable of comprehensively verifying documents related to the airworthiness certification procedure from the perspectives of formal, semantic, and procedural requirements, and directly reflecting the results in the actual flow of the airworthiness certification procedure. Background Technology

[0002] The airworthiness certification process for aircraft or aircraft components is an essential procedure for ensuring aviation safety and involves a complex and rigorous document review process based on aviation laws, regulations specific to certification bodies, certification types, and procedural steps. In the airworthiness certification process, various documents, such as design data, test result reports, inspection records, and compliance verification documents, must be submitted at each stage, and each document must meet prescribed formats and content requirements.

[0003] However, in conventional airworthiness certification procedures, it was common practice for personnel to manually review submitted documents to determine whether formal requirements were met and compliance with regulations was observed. This resulted in significant time and manpower consumption for document review, and led to issues where judgment results varied depending on the reviewer's proficiency or differences in interpretation. Furthermore, there were limitations in consistently assessing not only formal errors in the documents but also semantic compliance with the intent of the regulations.

[0004] Although technologies for automatically classifying or analyzing documents using artificial intelligence have recently been proposed, these technologies primarily focus on keyword extraction or simple classification. Consequently, there have been difficulties in directly applying them to environments where regulatory frameworks are complex and procedural connectivity is critical, such as airworthiness certification procedures. In particular, due to the nature of the aviation certification field, which involves frequent regulatory revisions, there have been issues such as the difficulty of maintaining consistent verification accuracy using only a model trained once. Prior art literature

[0005] Korean Registered Patent No. 10-1948257 The problem to be solved

[0006] The present invention was developed to solve the problems of the prior art. The objective of the present invention is to provide an airworthiness certification procedure automation system capable of comprehensively verifying documents related to the airworthiness certification procedure from the perspectives of formal, semantic, and procedural requirements, and directly reflecting the results in the actual flow of the airworthiness certification procedure.

[0007] Another objective of the present invention is to provide an airworthiness certification procedure automation system that clarifies the correspondence between documents and procedures by hierarchically managing regulatory information and step-by-step requirement documents of the airworthiness certification procedure, and simultaneously ensures the accuracy and reliability of document verification through a dual verification structure combining rule-based verification and language model-based reasoning. means of solving the problem

[0008] The present invention provides an airworthiness certification procedure automation system for an aircraft or aircraft parts, comprising: a storage unit storing regulation information for the airworthiness certification procedure and hierarchical structure data for documents required at each stage of the airworthiness certification procedure; a document input unit receiving a document file from a user; a dual verification unit including a rule-based document verification unit that detects formal errors in the document file input through the document input unit, and a validity determination unit that determines, through language model-based inference, whether the content of the document file semantically conforms to the hierarchical structure data and the regulations of the airworthiness certification procedure; an error type classification unit that classifies error types in the document file based on the verification results of the dual verification unit; and a procedure control unit that controls the procedure by applying a predefined state application rule based on the judgment results of the dual verification unit and the error type classification unit to apply the progress state of the airworthiness certification procedure corresponding to the document file as one of a state capable of proceeding, a state requiring supplementation, or a state suspended.

[0009] At this time, the storage unit can index and store hierarchical structure data that defines the hierarchical relationships between aviation regulations, certification bodies, certification types, and subordinate regulations.

[0010] In addition, the language model of the above validation unit can be fine-tuned in advance through a training dataset built based on the above hierarchical structure data.

[0011] In addition, the above training dataset can be constructed through a data augmentation technique that automatically generates multiple prompt-response pairs with different representation modes using a large language model (LLM) based on the above hierarchical structure data.

[0012] In addition, if the validity determination unit determines that the content of the document file does not correspond semantically, it can calculate a similarity score with a specific regulation clause within the hierarchical structure data that served as the basis for such determination.

[0013] In addition, the document input unit can extract text within a document image through OCR for the input document file, structure it while maintaining the mapping structure between the item name (Key) and input value (Value) of the table included in the document image, and transmit it to the dual verification unit.

[0014] In addition, if the text length of the document file exceeds the maximum number of input tokens of the language model, the validity determination unit can maintain the entire context by applying a sliding window algorithm to perform sequential inference for each text unit divided so that certain sections overlap.

[0015] In addition, the above procedure control unit can classify and store the document file according to the steps of the corresponding airworthiness certification procedure.

[0016] In addition, the error type classification unit can classify errors in the document file into format errors, semantic errors, and procedural errors.

[0017] In addition, the above-mentioned procedural state may be a state in which it is determined that there are no format errors, semantic errors, or procedural errors in the document file, thereby allowing progress to the next step of the airworthiness certification procedure.

[0018] In addition, the above supplementary request status may be a state in which a supplementary request or related regulatory information is provided because it is determined that format errors or semantic errors exist in the document file.

[0019] In addition, the above-mentioned suspension status may be a state in which progress to the next step of the airworthiness certification procedure is blocked because it is determined that there is a procedural error in the document file that does not correspond to the corresponding step of the airworthiness certification procedure. Effects of the invention

[0020] According to the present invention, by double-verifying the documents required for the airworthiness certification procedure from the perspectives of formal and semantic requirements, it is possible to comprehensively determine the external completeness of the documents and their substantial compliance with the intent of the regulations.

[0021] In addition, by classifying document verification results by error type and automatically applying the status of the airworthiness certification process to "ready to proceed," "requires supplementary information," or "suspended" according to the error type, the results of the document verification can be directly reflected in the actual flow of the airworthiness certification process. Consequently, this effectively prevents omissions or sequence errors in the certification process and reduces unnecessary re-examinations or repetitive work.

[0022] Furthermore, through a structure that separates and combines rule-based document verification and language model-based semantic judgment, it is possible to flexibly respond to documents of various expression styles while minimizing judgment errors that may occur in simple automation systems. In particular, since verification standards can be continuously updated even when airworthiness certification regulations change, it is effective in proactively responding to the changing regulatory environment.

[0023] In addition, by utilizing a lightweight language model in a local environment isolated from external networks, the security of aviation technical documents can be ensured, and efficient verification is possible even in an environment with limited computational resources. Brief explanation of the drawing

[0024] FIG. 1 is a block diagram illustrating the functional classification of the configuration of an airworthiness certification procedure automation system according to one embodiment of the present invention. Specific details for implementing the invention

[0025] Hereinafter, specific embodiments for implementing the present invention will be described in detail with reference to the drawings.

[0026] First, it should be noted that when assigning reference numerals to the components of each drawing, the same components are assigned the same reference numeral whenever possible, even if they are shown in different drawings. Furthermore, in describing the present invention, if it is determined that a detailed description of related known components or functions could obscure the essence of the invention, such detailed description is omitted.

[0027] Furthermore, when it is stated that one component is 'connected,' 'supported,' 'connected,' 'supplied,' 'transmitted,' or 'contacted' with another component, it should be understood that while the connection, support, connection, supply, transmission, or contact may be direct to that other component, there may also be other components present in between.

[0028] The terms used in this specification are used merely to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0029] Furthermore, it should be noted in advance that expressions such as "upper side," "lower side," and "side" in this specification are described based on the drawings, and may be expressed differently if the orientation of the object changes. For the same reason, some components in the attached drawings may be exaggerated, omitted, or schematically depicted, and the size of each component does not entirely reflect its actual size.

[0030] Additionally, terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but such components are not limited by such terms. These terms are used solely for the purpose of distinguishing one component from another.

[0031] The meaning of "comprising" as used in the specification is to specify certain characteristics, regions, integers, steps, actions, elements, and / or components, and does not exclude the existence or addition of other specific characteristics, regions, integers, steps, actions, elements, components, and / or groups.

[0032] FIG. 1 is a block diagram illustrating the functional classification of the configuration of an airworthiness certification procedure automation system according to one embodiment of the present invention.

[0033] An airworthiness certification procedure automation system according to one embodiment of the present invention may be configured to include a storage unit (100), a document input unit (200), a dual verification unit (300), an error type classification unit (400), and a procedure control unit (500).

[0034] The system of the present invention structurally manages regulatory information and document requirements required throughout the airworthiness certification process, and can comprehensively determine the conformity of documents by performing formal verification and semantic verification in parallel on document files submitted by users.

[0035] The storage unit (100) provides standard information for the entire system by systematically storing the relationship between the required documents and regulations for each stage of the airworthiness certification procedure, and the document input unit (200) can receive document files of various formats and convert them into a verifiable data form. The verification unit (300) can improve the accuracy of document verification by checking the formal requirements of the document through the rule-based document verification unit (310) and determining whether the document content semantically conforms to the regulations and procedure requirements through the validity judgment unit (320). The error type classification unit (400) classifies the error types of the document based on these verification results to provide basic information for subsequent procedure judgments, and the procedure control unit (500) controls the progress status of the airworthiness certification procedure according to the error types and verification results, thereby ensuring that the entire certification procedure is appropriately performed step by step. Accordingly, the system can function as an integrated certification support system that goes beyond simple document verification and automatically manages and controls the flow of the airworthiness certification procedure itself.

[0036] The storage unit (100) is configured to systematically manage standard information required throughout the airworthiness certification process and may be configured to store regulatory information related to the airworthiness certification process and hierarchical structure data regarding documents required at each stage of the airworthiness certification process. This hierarchical structure data may not be a simple list of documents, but may consist of structured data defined by reflecting the hierarchical relationships between aviation regulations, certification bodies, certification types, and subordinate regulations.

[0037] For example, hierarchical data structure can be classified according to the type of aviation regulation, such as domestic aviation safety laws or regulations of overseas certification bodies; under each regulation, types of certifications such as Type Certification (TC), Additional Type Certification (STC), Technical Standard Order (TSO), or Manufacturer's Approval can be hierarchically linked. Furthermore, under each certification type, detailed regulatory provisions, procedural steps, and types of submitted documents required to perform the corresponding certification process can be stored and mapped step-by-step.

[0038] The storage unit (100) can structurally represent the step requirements of the airworthiness certification procedure based on this hierarchical structure data, and can store the types of documents required for each step, document formats, mandatory entries, and related regulatory provisions together. Through this, when a specific document file is input, it is possible to quickly refer to which step of the airworthiness certification procedure the document is associated with and what regulatory requirements must be satisfied at that step.

[0039] To look at it more specifically, this hierarchical data structure can be represented as a tree structure or a hierarchical data format, with aviation regulations or certification bodies located at the top level, and a structure in which certification types, detailed regulations, procedural steps, and required documents for each step are sequentially connected at the lower levels.

[0040] For example, the top-level node of the hierarchical structure data may consist of regulations or certification body information, such as the Aviation Safety Act or regulations of overseas certification bodies, and certification types, such as Type Certification, Supplementary Type Certification, and Technical Standard Product Certification, may be connected as sub-nodes beneath that node. Under each certification type, the step-by-step flow of the certification process may be defined, and stages such as the preliminary consultation stage, certification application stage, technical review stage, testing and verification stage, and approval stage may be connected sequentially.

[0041] In addition, the document types required for each procedure step node can be stored in association with each step, and information regarding submission formats, mandatory entries, reference clauses, and the chronological relationships between documents can be mapped together for each document type. This structure can be represented as a set of key-value pairs and managed as a structure where the document type serves as the key and the requirement information for that document serves as the value.

[0042] The storage unit (100) can store such hierarchical structure data in a hierarchical data format such as JSON, and each data item may include a certification authority identifier, a certification type code, a procedure step identifier, a document type identifier, and a relevant regulation provision identifier. Through this, when a specific document file is input, it is possible to automatically identify which certification authority, which certification type, and which procedure step the document corresponds to through data structure exploration.

[0043] In this way, the hierarchical structure data stored in the storage unit (100) can structurally represent the regulatory framework, procedure flow, and document requirements of the airworthiness certification procedure, thereby supporting document verification, error type classification, and procedure control to be performed based on a consistent data structure.

[0044] Additionally, the storage unit (100) can support the dual verification unit (300) or the error type classification unit (400) in quickly searching for relevant regulatory provisions or procedural information during the document verification process by indexing and storing hierarchical structure data. This indexing structure can be configured based on certification authorities, certification types, procedural steps, or document types, and can be designed to enable searching and referencing based on multiple criteria.

[0045] In addition, the storage unit (100) can be used as reference information for a training dataset to support language model-based inference performed in the validity judgment unit (320). That is, the hierarchical structure data can be used as reference information to help the language model understand the context and regulatory framework of the airworthiness certification procedure, and can be used as basic data for configuring training data that reflects the relationships between regulatory provisions, the flow between procedure steps, and differences in document requirements.

[0046] In this way, the storage unit (100) can function as a standard information storage means that allows document verification, error type classification, and procedure control to be performed according to consistent standards by integrally managing the regulatory framework, procedure structure, and document requirements for the airworthiness certification procedure.

[0047] The document input unit (200) may be configured to serve as an interface for receiving various document files submitted by a user during the process of performing the airworthiness certification procedure into the system. The document input unit (200) may be configured to accept document files of various input formats, such as files written in electronic document format, image files generated by scanning paper documents, or document files containing a mixture of images and text.

[0048] The document input unit (200) can determine the format of the input document file and, if the document file is an image-based document, perform optical character recognition (OCR) to extract text information contained within the document image. At this time, the system may be configured to extract text by distinguishing paragraphs, items, table areas, etc., by analyzing the layout structure of the document together, rather than simply recognizing characters. Through this, verification can be performed in a subsequent step while maintaining the structural meaning of the document.

[0049] In particular, documents related to airworthiness certification procedures may contain a large amount of data organized in the form of tables, such as test result tables, inspection checklists, and compliance checklists. When processing such table-shaped data, the document input unit (200) may be configured to structure the data while maintaining a mapping relationship between the item names included in the table and the input values ​​corresponding to each item, rather than simply listing the text within the table. For example, the data can be converted into a structure in which the item name listed in the left column of the table is set as the Key, and the value entered in the corresponding item or the check result is set as the Value.

[0050] The document input unit (200) transmits the structured text data and table data to the dual verification unit (300), thereby enabling the rule-based document verification unit (310) to accurately detect whether required items are missing or whether format conditions are violated. Additionally, during the process in which the validity judgment unit (320) determines the content meaning of the document, the information within the document is provided in a form clearly separated into item units or paragraph units, so the accuracy of language model-based inference can be improved.

[0051] In addition, the document input unit (200) may be configured to extract or generate metadata of the document file. This metadata may include the file format, number of pages, time of creation, time of input, document classification candidate information, etc., and the metadata may be used as an auxiliary judgment criterion in the error type classification or procedure control process thereafter.

[0052] In this way, the document input unit (200) can function as a preprocessing unit that goes beyond simply uploading document files to the system, converts document information required for the airworthiness certification procedure into a structured data form, and supports subsequent verification and procedure control to be performed efficiently.

[0053] The dual verification unit (300) may be configured to include a rule-based document verification unit (310) and a validity determination unit (320).

[0054] The rule-based document verification unit (310) is a verification unit for detecting formal errors in document files related to the airworthiness certification procedure, and can be configured to determine whether the external structural requirements of the document comply with the regulations and step requirements of the airworthiness certification procedure. The rule-based document verification unit (310) functions as a primary verification means performed prior to the language model-based inference stage, thereby playing a role in ensuring the reliability and consistency of the document verification process.

[0055] The rule-based document verification unit (310) can verify the existence of required input items defined for each document type by referring to document requirement information included in the hierarchical structure data stored in the storage unit (100). For example, in the case of a test result report that must be submitted at a specific stage of the airworthiness certification procedure, the test date, test items, test conditions, test results, and the signature of the person in charge must be entered as required items, and the rule-based document verification unit (310) can detect as a format error if some of these required items are missing.

[0056] Additionally, the rule-based document verification unit (310) may be configured to verify the format conditions of the input value using document data transmitted in a structured form from the document input unit (200). For example, if a date item contains characters other than numbers, if a numeric value outside the range required by the regulation is entered, or if a required check item is not selected in a checklist-type document, etc., these can be determined as format errors.

[0057] In addition, the rule-based document verification unit (310) can analyze the overall structure of the document to detect inconsistencies in the document layout. For example, it can determine that there is a format error if the page composition, the location of tables, or the order of item arrangement differs significantly from the prescribed format despite being of the same document type, or if different document types are mixed and used. This structural verification can contribute to maintaining the standardized format of documents required in the airworthiness certification procedure.

[0058] The rule-based document verification unit (310) can record such verification results in item units or rule units, and can identify the location of the error, the type of error, and related regulation information together for each format error. This information can subsequently be used by the error type classification unit (400) to classify the error type of the document more accurately, and can also be used as supplementary guidance information provided to the user when the procedure control unit (500) applies a supplementary request status.

[0059] In this way, the rule-based document verification unit (310) can perform the function of systematically verifying the formal conformity of the document, thereby filtering out clear errors in advance at a stage prior to semantic judgment based on a language model, and improving the verification efficiency and stability of the entire airworthiness certification procedure automation system.

[0060] The validity determination unit (320) is an inference unit for determining whether the content of a document file is semantically consistent with the regulations and step-by-step requirements of the airworthiness certification procedure, and may be configured to perform a function distinct from the formal verification performed by the rule-based document verification unit (310). The validity determination unit (320) may perform the role of determining whether the document substantially satisfies the purpose and regulatory requirements of the airworthiness certification procedure by comprehensively analyzing the content of text, paragraphs, table items, etc. included in the document.

[0061] The validity determination unit (320) may be configured to perform language model-based inference, and the language model may be pre-fine-tuned through a training dataset built based on hierarchical structure data specialized for airworthiness certification procedures. This training dataset may include prescriptive sentences, procedure descriptions, document examples, etc., actually used in airworthiness certification procedures, and may be configured to include multiple question-response pairs that describe the same prescriptive requirements in different expressions. That is, the training dataset may be constructed through a data augmentation technique that automatically generates multiple question-response pairs with different expressions using a massive language model (LLM) based on hierarchical structure data.

[0062] For example, if a specific regulatory provision is defined by the phrase “test results must be performed under prescribed conditions,” the training dataset may include sentences that are semantically identical but have different expressions, such as “check if test conditions are set according to the regulations” and “does the test environment satisfy the regulatory requirements?” Through this learning process, the validity determination unit (320) can determine whether the document satisfies semantically identical requirements even if the way the document is expressed does not exactly match the regulatory phrase.

[0063] The language model used in the validity determination unit (320) can be configured to run in a local server environment isolated from an external network and can be designed to meet security requirements to prevent the external leakage of aviation technical documents and airworthiness certification data. This language model can be composed of a small large language model (sLLM) with a parameter size of 7B to 20B and can have a lightweight structure so as to be operated stably even in an environment with limited local computing resources.

[0064] In addition, the language model applied to the validity judgment unit (320) can be trained by applying a rank-adaptive fine-tuning technique that differentially optimizes weights according to the importance of the hierarchical structure data stored in the storage unit (100). For example, by assigning relatively high importance to key regulatory provisions of the airworthiness certification procedure or regulatory information that is essential to reference in a specific certification stage, and assigning low importance to regulatory information that is at the level of additional explanation or reference, the performance of regulatory inference can be efficiently improved even within a limited number of parameters. This rank-adaptive fine-tuning technique can strengthen adaptability to specific regulatory areas without retraining the entire language model, thereby enabling efficient training and operation in a local environment.

[0065] In addition, when new airworthiness certification regulations are added to the storage unit (100) or existing regulations or form data are revised, the validity determination unit (320) can update the verification criteria by retraining the language model using an additional training set generated based on the new or revised data. This retraining process can be performed by reflecting only the changed parts of the regulations while maintaining the existing training results, thereby enabling continuous verification in compliance with the latest regulations even in a practical environment where changes to airworthiness certification regulations are frequent.

[0066] Additionally, when the text length of the document file exceeds the maximum number of input tokens of the language model, the validity determination unit (320) may apply a sliding window algorithm to divide the document content into text units of a certain length, and configure adjacent text sections to overlap each other. Through this, sequential inference can be performed without interrupting the entire context of the document, and consistent semantic judgment can be made even for long airworthiness certification documents.

[0067] The validity judgment unit (320) can identify a specific regulation clause that served as the basis for the judgment within the hierarchical structure data when it is determined that the content of the document does not semantically conform to the regulations of the airworthiness certification procedure. At this time, the semantic similarity between the document content and the regulation clause can be quantified and calculated in the form of a similarity score, thereby numerically expressing how much the document deviates from a specific regulation requirement.

[0068] For example, if the submitted test report contains only a portion of the test scope required at a specific certification stage, the validity determination unit (320) determines that the similarity score between the document content and the regulation clause is less than a threshold value and can derive the regulation clause serving as the basis for this. This similarity score and regulation clause information can subsequently be used as a basis for determining whether there is a semantic error in the error type classification unit (400), and can also be used as explanatory information provided to the user when the procedure control unit (500) applies a supplementary requirement status.

[0069] In this way, the validity judgment unit (320) can function as a core inference unit that comprehensively determines whether the document has procedural and substantive validity, going beyond whether the document satisfies the formal requirements of the regulations, by connecting and interpreting the regulation system of the airworthiness certification procedure and the document content from a semantic perspective.

[0070] According to the configuration of the dual verification unit (300), when a user inputs a document file related to the airworthiness certification procedure through the document input unit (200), the document file is converted into a structured data form and then transmitted to the dual verification unit (300). First, the rule-based document verification unit (310) can verify the formal conformity of the document file based on the document requirements defined in the hierarchical structured data to detect whether essential items are missing, whether the input format is violated, or whether there is inconsistency in the document structure. Subsequently, the validity judgment unit (320) can determine, through language model-based inference, whether the content of the document file semantically conforms to the regulations and step-by-step requirements of the airworthiness certification procedure based on the formal verification results and the document content. In this process, even if the expression method of the document differs from the wording of the regulations, whether it satisfies the intent of the regulations and substantive requirements can be comprehensively reviewed.

[0071] In this way, the dual verification unit (300) performs formal verification and semantic verification in separate stages, and by utilizing the two verification results in a mutually complementary manner, it is possible to simultaneously secure the accuracy and reliability of the document verification. The rule-based document verification unit (310) identifies clear formal errors in advance to limit the scope of verification, and the validity judgment unit (320) precisely determines the semantic validity of the document content, thereby comprehensively evaluating the document conformity required in the airworthiness certification procedure.

[0072] The error type classification unit (400) may be configured to receive the formal verification result and semantic verification result produced by the dual verification unit (300) as input, and to perform the function of classifying errors present in a document file from the perspective of error type. The error type classification unit (400) may function as an intermediate judgment unit that not only indicates whether an error exists, but also systematically distinguishes the nature of the cause of the error, thereby enabling the subsequent procedure control unit (500) to accurately apply the progress status.

[0073] The error type classification unit (400) may be configured to classify errors in a document file into format errors, semantic errors, and procedural errors. Format errors may be classified based on results detected by the rule-based document verification unit (310), for example, when required items are missing or format conditions of input values ​​are not met, or when the page composition, table format, or item arrangement of the document differs significantly from the required form. Semantic errors may be classified based on the results of language model-based inference performed by the validity judgment unit (320), for example, when the content written in the document does not substantially conform to the intent of the regulation requirements, or when technical grounds, test conditions, approval criteria, etc. required at a specific stage are not sufficiently explained.

[0074] A procedural error may be classified when a document file does not correspond to the corresponding stage of the airworthiness certification procedure, and this may include cases where the timing or stage of document submission does not match the procedure flow, regardless of the suitability of the document content itself. For example, a procedural error may occur when a document that should be submitted in a preceding stage of the procedure is entered in a subsequent stage, when a document required only for a specific type of certification is entered in a procedure stage for a different type of certification, or when a form from a specific certification body is submitted in a procedure stage of a different certification body. To determine such a procedural error, the error type classification unit (400) may refer to the correspondence relationship information with the required documents for each stage defined in the hierarchical structure data stored in the storage unit (100) to determine whether the entered document file is suitable for a specific procedure stage or is allowed in the current stage.

[0075] The error type classification unit (400) can be configured to handle cases where multiple error types exist simultaneously. For example, if there is a missing required item in a document file, it may be classified as a format error, and if the document content is judged not to comply with regulatory requirements, it may also be classified as a semantic error. Additionally, if the document does not correspond to the procedure step itself, it may be classified as a procedure error, and if a procedure error is confirmed, it may be used as a priority judgment factor for procedure control regardless of whether a format error or a semantic error exists. To this end, the error type classification unit (400) can be configured to define a priority among error types to identify procedure errors preferentially when they exist, and to support the selection of a subsequent state application rule based on a combination of format errors and semantic errors within the scope where procedure errors do not exist.

[0076] The error type classification unit (400) can generate a classification result in a structured form and transmit it to the procedure control unit (500). The structured classification result may include quantitative indicators such as the type of error, the location or item where the error occurred, identification information of the relevant regulation clause, and a semantic similarity score. For example, if the validity judgment unit (320) calculates that the similarity score with a specific regulation clause is less than a threshold value, the error type classification unit (400) may classify the result as a semantic error, and may be configured to transmit together which regulation clause served as the basis and what level the similarity score is. Additionally, if the rule-based document verification unit (310) detects the omission of a specific required item, the error type classification unit (400) may generate a result classified as a format error including the item name.

[0077] In this way, the error type classification unit (400) can provide a standard for the process of applying the progress status of the airworthiness certification procedure as one of the following: a state where progress is possible, a state requiring supplementation, or a state where progress is suspended, by converting the result of the dual verification unit (300) into error type-based judgment information that can be directly utilized for procedure control. Accordingly, the error type classification unit (400) can function as a configuration that links the document verification result with procedure progress control.

[0078] The procedure control unit (500) is configured to control the actual flow of the airworthiness certification procedure so that the document verification results are reflected in the actual flow of the procedure. It may be configured to determine the progress status of the airworthiness certification procedure and control the procedure flow based on the judgment results of the dual verification unit (300) and the error type classification unit (400). The procedure control unit (500) is not limited to the function of simply displaying the document verification results to the user, but can also perform the function of automatically controlling whether the airworthiness certification procedure itself proceeds.

[0079] The procedure control unit (500) can determine the progress status of the airworthiness certification procedure corresponding to the document file by referring to a predefined status application rule. The status application rule may be defined to select one of the following: a state where progress is possible, a state requiring supplementation, or a state where progress is suspended, using the error type classified by the error type classification unit (400) and the procedure step information corresponding to the document as input values. Such status application rule may be set according to the characteristics of the airworthiness certification procedure, the operating standards of the certification body, or requirements for each type of certification.

[0080] The progress-enabled status may be applied when it is determined that there are no format errors, semantic errors, or procedural errors in the document file, in which case the procedure control unit (500) may automatically allow progress to the next step of the airworthiness certification procedure. For example, if all document requirements required for a specific certification step are met, the step may be treated as completed and control may be exercised to automatically transition to the document submission or verification procedure required for the next step.

[0081] A supplementary requirement status may be applied when format errors or semantic errors exist in a document file. In this case, the procedure control unit (500) may temporarily restrict progress to the next step of the airworthiness certification procedure and control the provision of document items requiring supplementation or related regulatory information to the user. For example, if a required item is missing, the item may be explicitly notified, and if a semantic discrepancy with regulatory requirements is confirmed, the regulatory clause that served as the basis and similarity score information may be provided together. Through this, the user can clearly recognize which parts need to be corrected and how.

[0082] A state of suspension of progress may be applied when there is a procedural error in which a document file does not correspond to the corresponding step of the airworthiness certification procedure, and in this case, the procedure control unit (500) may control the process to block progress to the next step of the airworthiness certification procedure. For example, if a document that should be submitted in a preceding step is entered in a subsequent step, or if a document that is allowed only for a specific type of certification is submitted in a procedure step of a different type of certification, the procedure control unit (500) may determine that the document does not conform to the current procedure flow and stop the procedure. This state of suspension of progress may be a control method that reflects the characteristics of the airworthiness certification procedure, which must strictly adhere to the order of the procedure.

[0083] Additionally, the procedure control unit (500) may be configured to classify and store document files according to the stages of the airworthiness certification procedure. This allows documents submitted and verification results for each stage of the procedure to be systematically managed, and makes it easier to track the progress history and status of the certification procedure. Furthermore, if a modified document is re-entered while in a state requiring supplementation, the status application rule is reapplied to the document by passing it through the double verification unit (300) and the error type classification unit (400) again, thereby allowing for an automatic transition to a state where the process is possible or suspended.

[0084] In this way, the procedure control unit (500) can support the airworthiness certification procedure to be systematically performed according to prescribed order and standards by dynamically controlling the progress status of the airworthiness certification procedure based on the document verification results.

[0085] The airworthiness certification procedure automation system configured in this manner hierarchically manages the regulatory information and document structures required for the airworthiness certification process. After double-checking document files input by users both formally and semantically, it can automatically control the progress status of the certification process based on error types. Through this, it is possible to comprehensively assess not only the formal completeness of documents but also their substantive compliance with the intent of the regulations. Furthermore, the results of document verification are not merely reference information but are directly reflected in the decision to proceed with the actual certification process. Additionally, through a structure that separates and combines rule-based verification and language model-based inference, both reliability and flexibility of verification can be secured simultaneously. This enables continuous verification that complies with the latest standards, even in the practical environment of airworthiness certification where regulatory revisions or procedural changes occur frequently. Accordingly, the present invention can be utilized as an integrated airworthiness certification support system that effectively resolves the complexity of the airworthiness certification process and the burden of repetitive document review, while improving the accuracy, consistency, and efficiency of the certification process.

[0086] The airworthiness certification procedure automation system of the present invention may be implemented in a form built within a server and may operate by communicating with one or more user terminals. The server may be configured to perform key functions such as document verification, error type classification, and procedure control, and the user terminal may be configured to provide user interface functions such as document input, verification result confirmation, and response to supplementary requests.

[0087] The server can perform data communication with user terminals via wired or wireless communication networks, and user terminals can be implemented in various forms, such as personal computers, laptops, tablets, smartphones, or dedicated business terminals. Document files, document metadata, or user request information entered from the user terminal can be transmitted to the server via the communication network, and based on this, the server can execute each component of the airworthiness certification procedure automation system to perform document verification and procedure judgment.

[0088] In addition, the server can transmit document verification results, error type classification results, information on the progress status of the airworthiness certification process, and guidance on requests for supplementary information to the user terminal, which can display this information in the form of screen output, notification messages, or reports. Through this, users can intuitively check the current progress status of the airworthiness certification process and whether document supplementation is required.

[0089] Meanwhile, communication between the server and the user terminal may be carried out through authentication and access control procedures, and the scope of access to functions such as document uploading, verification result inquiry, or procedure status change may be restricted depending on user authority. In addition, the system of the present invention can be operated in a closed network or internal private network environment, so it can be configured to enable communication with the user terminal while maintaining the security of aviation technical data and airworthiness certification-related information.

[0090] As such, the airworthiness certification procedure automation system of the present invention is implemented based on a server-client structure, and can systematically support the entire airworthiness certification procedure by efficiently collecting and verifying airworthiness certification-related documents from multiple user terminals and providing the results in real time.

[0091] In the airworthiness certification procedure automation system of the present invention, each component, such as a storage unit, a document input unit, a double verification unit, an error type classification unit, and a procedure control unit placed within the server, is not limited to a hardware device but can be implemented by a program executed on the server. Such a program can be configured to be executed by the server's processor to perform functions corresponding to each component, and can perform data processing, storage, and control in cooperation with memory and storage devices.

[0092] Each component can be implemented as an independent software module or service, and data can be transferred between modules through defined interfaces. For example, the document input unit can be implemented as a program module that receives document files and converts them into structured data, and the verification unit can be implemented as a combination of program modules that perform rule-based document verification and language model-based inference functions, respectively. The error type classification unit and the procedure control unit can also be implemented as program modules that receive verification result data as input and perform judgment and control logic.

[0093] In addition, the storage unit may be implemented as a program module linked to a database management system or a file system, and may be configured to store and manage regulatory information for airworthiness certification procedures, hierarchical structured data, document verification results, and procedure progress status information. This storage unit can perform the role of providing or updating necessary data in response to requests from other program modules within the server.

[0094] The system of the present invention may be configured so that these program modules are executed within a single server, or, if necessary, distributed and executed across multiple servers or a virtualization environment. In this case as well, each component may be implemented through a program and interconnected via a network or an internal communication interface.

[0095] Additionally, the configuration of the airworthiness certification procedure automation system described herein may be realized by digital electronic circuits, integrated circuits, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementation by one or more computer programs executable on a programmable system. A programmable system comprises a storage system, at least one input device, and at least one programmable processor (which may be a special-purpose processor or a general-purpose processor) coupled to receive data and instructions from at least one output device and to transmit data and instructions to them. Computer programs (which are also known as programs, software, software applications, or code) include instructions for the programmable processor and are stored on a "computer-readable recording medium."

[0096] Computer-readable recording media include all types of recording devices in which data that can be read by a computer system is stored. Such computer-readable recording media may further include non-volatile or non-transitory media such as ROM, CD-ROM, magnetic tape, floppy disk, memory card, hard disk, magneto-optical disk, and storage device, or transitory media such as data transmission media. Additionally, computer-readable recording media may be distributed across networked computer systems, and computer-readable code may be stored and executed in a distributed manner.

[0097] Various embodiments of the methods described herein may be implemented by a programmable computer. Here, the computer includes a programmable processor, a data storage system (including volatile memory, non-volatile memory, or other types of storage systems, or a combination thereof), and at least one communication interface. For example, the programmable computer may be one of a server, a network device, a set-top box, an embedded device, a computer expansion module, a personal computer, a laptop, a PDA (Personal Data Assistant), a cloud computing system, or a mobile device.

[0098] The foregoing description is merely an illustrative explanation of the technical concept of the present invention, and those skilled in the art to which the present invention pertains will be able to make various modifications and variations within the scope of the essential characteristics of the present invention. Accordingly, the embodiments disclosed in the present invention are intended to explain, not limit, the technical concept of the present invention, and the scope of the technical concept of the present invention is not limited by these embodiments. The scope of protection of the present invention shall be interpreted by the claims below, and all technical concepts within an equivalent scope shall be interpreted as being included within the scope of rights of the present invention. Explanation of the symbols

[0099] 100: Storage section 200: Document Input Section 300: Double verification unit 310: Rule-based document validation unit 320: Validity determination unit 400: Error Type Classification Section 500: Procedure Control Unit

Claims

Claim 1 A system for automating the airworthiness certification procedure for an aircraft or aircraft parts, comprising: a storage unit that stores regulatory information for the airworthiness certification procedure and hierarchical structure data for documents required at each stage of the airworthiness certification procedure; a document input unit that receives a document file from a user; a dual verification unit including a rule-based document verification unit that detects formal errors in the document file input through the document input unit, and a validity determination unit that determines, through language model-based inference, whether the content of the document file semantically conforms to the hierarchical structure data and the regulations of the airworthiness certification procedure; and an error type classification unit that classifies error types in the document file based on the verification results of the dual verification unit. The airworthiness certification procedure automation system includes a procedure control unit that controls the procedure by applying a predefined state application rule based on the judgment results of the dual verification unit and the error type classification unit to apply the progress state of the airworthiness certification procedure corresponding to the document file to one of a state where the procedure is possible, a state requiring supplementation, and a state where the procedure is suspended; and the validity judgment unit, when the text length of the document file exceeds the maximum number of input tokens of the language model, applies a sliding window algorithm to perform sequential inference for each text unit divided so that a certain section overlaps, thereby maintaining the overall context. Claim 2 In claim 1, the storage unit is an airworthiness certification procedure automation system that indexes and stores hierarchical structure data defining the hierarchical relationship between aviation regulations, certification bodies, certification types, and subordinate regulations. Claim 3 In claim 2, the language model of the validity determination unit is pre-fine-tuned through a training dataset built based on the hierarchical structure data, an airworthiness certification procedure automation system. Claim 4 In claim 3, the above-mentioned training dataset is constructed through a data augmentation technique that automatically generates multiple prompt-response pairs having different representation modes using a large language model (LLM) based on the above-mentioned hierarchical structure data, thereby forming an airworthiness certification procedure automation system. Claim 5 An airworthiness certification procedure automation system according to claim 1, wherein the validity determination unit calculates a similarity score with a specific regulatory provision that served as the basis for the determination within the hierarchical structure data when it determines that the content of the document file does not correspond semantically. Claim 6 An airworthiness certification procedure automation system according to claim 1, wherein the document input unit extracts text within a document image through OCR for the input document file, structures it while maintaining the mapping structure between the item name (Key) and input value (Value) of a table included in the document image, and transmits it to the dual verification unit. Claim 7 delete Claim 8 In claim 1, the procedure control unit classifies and stores the document file according to the corresponding steps of the airworthiness certification procedure, an airworthiness certification procedure automation system. Claim 9 delete Claim 10 A system for automating the airworthiness certification procedure for an aircraft or aircraft parts, comprising: a storage unit that stores regulatory information for the airworthiness certification procedure and hierarchical structure data for documents required at each stage of the airworthiness certification procedure; a document input unit that receives a document file from a user; a dual verification unit including a rule-based document verification unit that detects formal errors in the document file input through the document input unit, and a validity determination unit that determines, through language model-based inference, whether the content of the document file semantically conforms to the hierarchical structure data and the regulations of the airworthiness certification procedure; and an error type classification unit that classifies error types in the document file based on the verification results of the dual verification unit. The airworthiness certification procedure automation system includes a procedure control unit that controls the procedure by applying a predefined state application rule based on the judgment results of the dual verification unit and the error type classification unit to apply the progress state of the airworthiness certification procedure corresponding to the document file as one of a state where progress is possible, a state requiring supplementation, and a state where progress is suspended; the error type classification unit classifies the errors in the document file into format errors, semantic errors, and procedural errors; and the state where progress is possible is a state in which it is determined that no format errors, semantic errors, or procedural errors exist in the document file, thereby allowing progress to the next step of the airworthiness certification procedure. Claim 11 A system for automating the airworthiness certification procedure for an aircraft or aircraft parts, comprising: a storage unit that stores regulatory information for the airworthiness certification procedure and hierarchical structure data for documents required at each stage of the airworthiness certification procedure; a document input unit that receives a document file from a user; a dual verification unit including a rule-based document verification unit that detects formal errors in the document file input through the document input unit, and a validity determination unit that determines, through language model-based inference, whether the content of the document file semantically conforms to the hierarchical structure data and the regulations of the airworthiness certification procedure; and an error type classification unit that classifies error types in the document file based on the verification results of the dual verification unit. The airworthiness certification procedure automation system includes a procedure control unit that controls the procedure by applying a predefined state application rule based on the judgment results of the dual verification unit and the error type classification unit to apply the progress state of the airworthiness certification procedure corresponding to the document file to one of a state where the procedure is possible, a state requiring supplementation, and a state where the procedure is suspended, wherein the error type classification unit classifies errors in the document file into format errors, semantic errors, and procedure errors, and the state requiring supplementation is a state in which a request for supplementation or related regulatory information is provided when it is determined that a format error or a semantic error exists in the document file. Claim 12 A system for automating the airworthiness certification procedure for an aircraft or aircraft parts, comprising: a storage unit that stores regulatory information for the airworthiness certification procedure and hierarchical structure data for documents required at each stage of the airworthiness certification procedure; a document input unit that receives a document file from a user; a dual verification unit including a rule-based document verification unit that detects formal errors in the document file input through the document input unit, and a validity determination unit that determines, through language model-based inference, whether the content of the document file semantically conforms to the hierarchical structure data and the regulations of the airworthiness certification procedure; and an error type classification unit that classifies error types in the document file based on the verification results of the dual verification unit. The airworthiness certification procedure automation system includes a procedure control unit that controls the procedure by applying a predefined state application rule based on the judgment results of the dual verification unit and the error type classification unit to apply the progress state of the airworthiness certification procedure corresponding to the document file as one of a state where progress is possible, a state requiring supplementation, and a state where progress is suspended; the error type classification unit classifies errors in the document file into format errors, semantic errors, and procedural errors; and the state where progress is suspended is a state in which progress to the next stage of the airworthiness certification procedure is blocked when it is determined that there is a procedural error in the document file that does not correspond to the corresponding stage of the airworthiness certification procedure.

Images