Method and system for generating an initial aircraft support plan

Abstract

The present invention relates to a method and system for generating an initial aircraft support plan. First, an aircraft identification code is extracted from a historical document, and stored in an aircraft information table of a database with corresponding basic aircraft information. A named entity recognition method based on transfer learning and BiLSTM is used to perform entity recognition and annotation of the plan text. A relation extraction method based on text features is used to extract entries and texts according to the identified plan text, and store them in the plan information table of the database. Then, the existing aircraft in the database are sorted by similarity based on the aircraft identification code. Finally, according to the principle of deferred substitution, the corresponding plan text is extracted from the plan information table, filled in a template, and the generated plan is output. The present invention realizes the fully automated generation of the initial aircraft support plan, greatly reducing the formulation time and cost of the aircraft support plan.

Description

Technical Field

[0001] The present invention relates to the technical field of aircraft device support, and in particular to a method and system for generating an initial aircraft support plan. Background Art

[0002] The support concept is a complete overall description of the support system, which consists of a set of comprehensive support element plans and is coordinated with the design plan and the use plan. In the field of modern equipment support, aircraft is the main source of equipment comprehensive support engineering and the place where it is most used. For aircraft equipment, the aircraft support plan is a summary of the overall support work of the aircraft, an overall plan for implementing the aircraft support requirements and achieving the support goals, and a key work in the comprehensive support project. The formulation of the aircraft support plan is a dynamic process. The initial support plan is proposed in the aircraft demonstration stage. It is the basis for studying the impact of support issues on aircraft design and an important basis for determining the reliability and maintainability indicators of the aircraft. It is composed of preliminary ideas of various comprehensive support elements; the optimized support plan in the program stage and the engineering development stage is the basic basis for formulating the support plan and developing support resources. Only from the optimized support plan can the best support resource requirements be obtained; in the deployment and use stage, the maintenance level and the main work of each level specified in the support plan are the basis for establishing the support system and maintenance system of the newly developed aircraft.

[0003] In the current maintenance plan formulation process, the human link accounts for the vast majority of the workload. In the existing research on aircraft maintenance plan generation methods, the research direction is to use expert systems, only for the aircraft maintenance link, to generate plans based on the mechanical structure, historical fault information and maintenance methods of relatively specific parts and structures to be repaired. The advantage of this method is that the generated plan is relatively more specific and more feasible, but on the one hand, for aircraft under development, this method is not applicable due to the lack of specific fault history data; on the other hand, this method is aimed at maintenance personnel. From the perspective of designers, it is impossible to generate plans based on aircraft and their support systems, and it is impossible to provide effective assistance in the design link.

[0004] The method of manually generating aircraft support plans with computer assistance is to simply transfer the work to the computer to facilitate communication between different personnel and standardize the format of output documents. Although this method is easy to understand logically and does not require the design of additional algorithms, it simply computerizes the manual process to make the output more standardized, but the problems of high time cost and difficulty in solving the manual process still exist. Although computer assistance has reduced the workload of collaborative work and adding, deleting, modifying and checking relevant data to a certain extent, it is still necessary to advance the preparation of the plan according to the traditional manual steps, with a low degree of automation and a very limited degree of reduction in the overall workload. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for generating an initial aircraft support plan, so as to reduce the time and cost of formulating the aircraft support plan.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for generating an initial aircraft support plan, the method comprising:

[0008] Extracting an aircraft identification code from historical documents related to the aircraft support plan and storing it together with the corresponding aircraft basic information as a new data in an aircraft information table of a database; the aircraft information table includes an aircraft number, aircraft basic information and an aircraft identification code;

[0009] According to the historical documents, a named entity recognition method based on transfer learning and BiLSTM is used to obtain the guarantee plan text of the labeled plan item entities;

[0010] According to the support plan text of the marked plan item entity, a relation extraction method based on text features is used to obtain the marked plan item entity corresponding to each necessary tag, and the plan item text of the marked plan item entity corresponding to each necessary tag is stored in a plan information table of the database; the necessary tag is an item that must be included in the aircraft support plan; the plan information table includes the aircraft number and all necessary tags;

[0011] Calculate the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sort the aircraft information table in descending order according to the similarity, and re-sort the solution information table according to the aircraft number sequence of the sorted aircraft information table;

[0012] According to the principle of postponement and substitution, the text of the plan entry that appears for the first time for each necessary label in the sorted plan information table is filled into the corresponding necessary label position of the support plan template of the aircraft to be generated, so as to obtain the initial aircraft support plan of the aircraft to be generated.

[0013] Optionally, extracting the aircraft identification code from the historical documents related to the aircraft support plan specifically includes:

[0014] The information in the historical documents is stored in txt text in the form of strings through the OCR extension package;

[0015] According to the txt text, the aircraft identification code is extracted according to the aircraft identification code encoding rules.

[0016] Optionally, according to the safeguard solution text of the labeled solution item entity, a relation extraction method based on text features is used to obtain the labeled solution item entity corresponding to each necessary tag, and the solution item text of the labeled solution item entity corresponding to each necessary tag is stored in the solution information table of the database, specifically including:

[0017] Using a matching field method to identify necessary tags in the safeguard plan text of the marked plan entry entity, and marking an identifier to obtain a safeguard plan identification text;

[0018] Segment the security solution identification text according to text features to obtain multiple sentences;

[0019] Integrate a sentence without a necessary label between two adjacent sentences containing necessary labels and a sentence containing necessary labels that comes first in the word order among the two adjacent sentences containing necessary labels into a semantic block, thereby obtaining multiple semantic blocks;

[0020] The necessary tags in each semantic block are used as the necessary tags corresponding to the annotated solution entry entities in the respective semantic blocks, all the annotated solution entry entities corresponding to each necessary tag are obtained, and the solution entry text of the annotated solution entry entity corresponding to each necessary tag is stored in the solution information table of the database.

[0021] Optionally, the calculating the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sorting the aircraft information table in descending order according to the similarity, and re-sorting the solution information table according to the aircraft number sequence of the sorted aircraft information table, further comprises:

[0022] Get the basic information field data of the aircraft to be generated;

[0023] According to the basic information field data of the aircraft to be generated and in accordance with the aircraft identification code coding rules, the aircraft identification code to be generated is extracted.

[0024] Optionally, the calculating the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, sorting the aircraft information table in descending order according to the similarity, and re-sorting the solution information table according to the aircraft number sequence of the sorted aircraft information table specifically includes:

[0025] The similarity between the aircraft identification code in the aircraft information table and the aircraft identification code to be generated is calculated according to the ratio of the intersection and union of the sets in the character string constituting the aircraft identification code;

[0026] Sort the aircraft information table once according to the order of similarity from large to small;

[0027] If there are multiple aircraft with the same similarity, the aircraft information table parts with the same similarity are sorted again in descending order of the service years in the aircraft information table;

[0028] The plan information table is re-sorted according to the aircraft number sequence of the aircraft information table after the secondary sorting.

[0029] Optionally, according to the principle of postponement and substitution, the text of the solution item in which each necessary tag appears for the first time in the sorted solution information table is filled into the corresponding necessary tag position of the support solution template of the aircraft to be generated, so as to obtain the initial aircraft support solution of the aircraft to be generated, specifically including:

[0030] Using the solution entry texts corresponding to all necessary tags of the first data in the sorted solution information table as pre-selected texts for the aircraft to be generated;

[0031] Determine whether the pre-selected text contains a null value placeholder, and obtain a determination result;

[0032] If the judgment result indicates yes, then fill the pre-selected text with the solution entry text that appears first under the necessary tag where the empty value placeholder is located according to the arrangement order;

[0033] If the judgment result indicates no, the pre-selected text is filled into the corresponding position of the support plan template of the aircraft to be generated to obtain the initial aircraft support plan of the aircraft to be generated.

[0034] A system for generating an initial aircraft support plan, the system comprising:

[0035] An aircraft information table update module is used to extract an aircraft identification code from historical documents related to the aircraft support plan, and store it in an aircraft information table of the database as a new data item together with the corresponding aircraft basic information; the aircraft information table includes an aircraft number, aircraft basic information and aircraft identification code;

[0036] A guarantee plan text acquisition module is used to obtain the guarantee plan text of the marked plan item entity according to the historical document by adopting a named entity recognition method based on transfer learning and BiLSTM;

[0037] A scheme information table updating module is used to obtain the marked scheme item entity corresponding to each necessary tag according to the support scheme text of the marked scheme item entity and the relation extraction method based on text features, and store the scheme item text of the marked scheme item entity corresponding to each necessary tag into the scheme information table of the database; the necessary tag is an item that must be included in the aircraft support scheme; the scheme information table includes the aircraft number and all necessary tags;

[0038] A sorting module is used to calculate the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sort the aircraft information table in descending order according to the similarity, and at the same time re-sort the solution information table according to the aircraft number sequence of the sorted aircraft information table;

[0039] The aircraft initial support plan generation module is used to fill in the plan entry text of each necessary tag that appears for the first time in the sorted plan information table into the corresponding necessary tag position of the support plan template of the aircraft to be generated based on the principle of deferred substitution, so as to obtain the aircraft initial support plan of the aircraft to be generated.

[0040] Optionally, the scheme information table updating module specifically includes:

[0041] A submodule for obtaining a protection scheme identification text is used to identify necessary tags in the protection scheme text of the marked scheme entry entity by using a matching field method, and to add an identifier to obtain the protection scheme identification text;

[0042] A segmentation submodule, used for segmenting the security solution identification text according to text features to obtain multiple sentences;

[0043] A semantic block integration submodule is used to integrate a sentence without necessary tags between two adjacent sentences containing necessary tags and a sentence containing necessary tags that comes first in the word order of the two adjacent sentences containing necessary tags into one semantic block, thereby obtaining multiple semantic blocks;

[0044] The scheme entry text storage submodule is used to use the necessary tags in each semantic block as the necessary tags corresponding to the labeled scheme entry entities in the respective semantic blocks, obtain all the labeled scheme entry entities corresponding to each necessary tag, and store the scheme entry text of the labeled scheme entry entities corresponding to each necessary tag into the scheme information table of the database.

[0045] Optionally, the sorting module specifically includes:

[0046] A similarity calculation submodule, used to calculate the similarity between the aircraft identification code in the aircraft information table and the aircraft identification code to be generated according to the ratio of the intersection and the union of the set in the character string constituting the aircraft identification code;

[0047] A one-time sorting submodule is used to sort the aircraft information table in descending order of similarity;

[0048] A secondary sorting submodule is used to sort the aircraft information table parts with the same similarity in descending order of the service years in the aircraft information table if there are multiple aircraft information tables with the same similarity;

[0049] The scheme information table sorting submodule is used to re-sort the scheme information table according to the aircraft number sequence of the aircraft information table after the secondary sorting.

[0050] Optionally, the aircraft initial support plan generation module specifically includes:

[0051] A pre-selected text obtaining submodule, used to use the solution item texts corresponding to all necessary tags of the first data in the sorted solution information table as pre-selected texts for the aircraft to be generated;

[0052] A judgment result obtaining submodule is used to judge whether the pre-selected text contains a null value placeholder and obtain a judgment result;

[0053] A filling submodule is used for filling the pre-selected text with the solution entry text that appears for the first time under the necessary tag where the empty value placeholder is located according to the arrangement order if the judgment result indicates yes;

[0054] The aircraft initial support plan generating submodule is used to fill the pre-selected text into the corresponding position of the support plan template of the aircraft to be generated if the judgment result indicates no, so as to obtain the aircraft initial support plan of the aircraft to be generated.

[0055] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0056] The present invention discloses a method and system for generating an initial aircraft support plan. First, an aircraft identification code is extracted from a historical document, and the code is stored in an aircraft information table of a database as a new data together with the corresponding basic information of the aircraft. A named entity recognition method based on transfer learning and BiLSTM is used to obtain the support plan text of the labeled plan item entity, and a relationship extraction method based on text features is used to obtain the labeled plan item entity corresponding to each necessary label, and the labeled plan item entity is stored in the plan information table of the database. Then, the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated is calculated, and the aircraft information table and the plan information table are sorted in descending order according to the similarity. Finally, according to the principle of deferred substitution, the plan item text of each necessary label appearing for the first time in the sorted plan information table is filled into the corresponding necessary label position of the support plan template of the aircraft to be generated, so as to obtain the aircraft initial support plan of the aircraft to be generated. Based on a neural network and historical data, the present invention realizes the fully automated generation of the initial aircraft support plan, and greatly reduces the formulation time and cost of the aircraft support plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0058] Figure 1 A flowchart of a method for generating an initial aircraft support plan provided by the present invention;

[0059] Figure 2 A schematic diagram of a method for generating an initial aircraft support plan provided by the present invention;

[0060] Figure 3 A schematic diagram of the working process of the historical document structured storage module provided by the present invention;

[0061] Figure 4 Schematic diagram of text segmentation and semantic block integration provided by the present invention

[0062] Figure 5 A schematic diagram of the working process of the protection scheme generation module provided by the present invention. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] The purpose of the present invention is to provide a method and system for generating an initial aircraft support plan, so as to reduce the time and cost of formulating the aircraft support plan.

[0065] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0066] In order to solve the problems of complex manual process, high labor cost and time cost, and great influence of subjective opinions in the process of aircraft support plan formulation, the present invention provides a method for generating an initial aircraft support plan, such as Figure 1 As shown, the method includes:

[0067] Step 101, extracting an aircraft identification code from historical documents related to an aircraft support plan, and storing the aircraft identification code and the corresponding aircraft basic information as a new data in an aircraft information table of a database; the aircraft information table includes an aircraft number, aircraft basic information and an aircraft identification code.

[0068] The aircraft identification code is extracted from the historical documents related to the aircraft support plan, including:

[0069] The information in the historical documents is stored in txt text in the form of strings through the OCR extension package;

[0070] According to the txt text and the aircraft identification code encoding rules, extract the aircraft identification code.

[0071] Step 102: According to the historical documents, a named entity recognition method based on transfer learning and BiLSTM is used to obtain the security plan text of the labeled plan item entities.

[0072] Step 103, according to the support plan text of the labeled plan item entity, a relational extraction method based on text features is used to obtain the labeled plan item entity corresponding to each necessary tag, and the plan item text of the labeled plan item entity corresponding to each necessary tag is stored in the plan information table of the database; the necessary tags are items that must be included in the aircraft support plan; the plan information table includes the aircraft number and all necessary tags.

[0073] Specifically include:

[0074] The necessary tags in the assurance plan text of the marked plan entry entity are identified by using the matching field method, and the identifiers are added to obtain the assurance plan identification text;

[0075] According to the text features, the guarantee scheme identifies the text and segments it to obtain multiple sentences;

[0076] Integrate a sentence without a necessary label between two adjacent sentences containing necessary labels and a sentence containing necessary labels that comes first in the word order among the two adjacent sentences containing necessary labels into a semantic block, thereby obtaining multiple semantic blocks;

[0077] The necessary tags in each semantic block are used as the necessary tags corresponding to the annotated solution entry entities in the respective semantic blocks, all the annotated solution entry entities corresponding to each necessary tag are obtained, and the solution entry text of the annotated solution entry entity corresponding to each necessary tag is stored in the solution information table of the database.

[0078] Step 104, calculate the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sort the aircraft information table in descending order according to the similarity, and re-sort the solution information table according to the aircraft number sequence of the sorted aircraft information table.

[0079] Specifically include:

[0080] The similarity between the aircraft identification code in the aircraft information table and the aircraft identification code to be generated is calculated according to the ratio of the intersection and union of the sets in the character string constituting the aircraft identification code;

[0081] Sort the aircraft information table once according to the order of similarity from large to small;

[0082] If there are multiple aircraft with the same similarity, the aircraft information table parts with the same similarity are sorted again in descending order of the service years in the aircraft information table;

[0083] The plan information table is re-sorted according to the aircraft number sequence of the aircraft information table after the secondary sorting.

[0084] The extraction process of the aircraft identification code to be generated is: obtaining the basic information field data of the aircraft to be generated; extracting the aircraft identification code to be generated according to the basic information field data of the aircraft to be generated and the aircraft identification code coding rules.

[0085] Step 105, based on the principle of postponement and substitution, fill the solution entry text of each necessary tag that appears for the first time in the sorted solution information table into the corresponding necessary tag position of the support solution template of the aircraft to be generated, and obtain the aircraft initial support solution of the aircraft to be generated.

[0086] Specifically include:

[0087] The scheme entry texts corresponding to all necessary tags of the first data in the sorted scheme information table are used as pre-selected texts for the aircraft to be generated;

[0088] Determine whether the pre-selected text contains a null value placeholder and obtain the determination result;

[0089] If the judgment result indicates yes, the pre-selected text is filled with the solution entry text that first appears under the required tag where the empty value placeholder is located according to the arrangement order;

[0090] If the judgment result indicates no, the pre-selected text is filled into the corresponding position of the support plan template of the aircraft to be generated to obtain the initial aircraft support plan of the aircraft to be generated.

[0091] The formulation of aircraft support plans is a work that is carried out during the development process. As the aircraft development process is continuously optimized and iterated, the initial support plan is the basis for the formulation of the aircraft support plan, and its formulation or generation efficiency is very important. In the initial stage of aircraft development, since there is relatively little available information and iterative optimization is required according to actual conditions, the accuracy of the plan generation is not very high. Therefore, the historical plans of existing aircraft can be used as references, that is, a benchmark comparison system can be established. However, the establishment of an automated benchmark comparison system faces two major problems: one is the structuring of historical document information, and the other is the quantitative establishment of the rules for the benchmark comparison system. With the development of artificial intelligence technology, named entity recognition (NER) has made great progress in recent years. Combined with the clear structure of the support plan document, it can be used to structure the historical document information; and similarity and its related theories also provide tools for the quantitative evaluation of the selection of similar aircraft in the construction of the benchmark comparison system.

[0092] The present invention provides an aircraft initial support plan generation tool based on neural network and historical data, which uses BiLSTM and transfer learning-based named entity recognition technology and text feature-based relationship extraction methods to structure and store aircraft support plan texts in historical documents, extract aircraft identification codes, and use aircraft similarity ranking methods based on similarity theory to achieve quantitative establishment and evaluation of a benchmark comparison system, and finally achieve aircraft initial support plan generation. The present invention can help solve the problems of complex manual processes, high labor and time costs, and great influence of subjective opinions in the early stages of the formulation of most aircraft initial support plans under current standards.

[0093] In order to further illustrate the aircraft initial support plan generation method of the present invention, the detailed process is as follows:

[0094] The present invention proposes a tool for generating an initial aircraft support plan based on a neural network and historical data. Historical documents may refer to support plan documents of past or current aircraft, or documents and texts containing support plan items and contents that can be obtained for past or current aircraft. These documents contain a large amount of unstructured aircraft support plan information. By using a named entity recognition method and, on this basis, a relationship extraction method based on text features is adopted to structure and store the support plan information in the historical documents. At the same time, the aircraft identification code of the aircraft in the historical document is extracted based on information such as the aircraft type and usage conditions. When generating a plan, the aircraft in the database are sorted by similarity theory, and the entry text is filled into the template according to the deferred substitution rule to realize the automatic generation of the initial aircraft support plan.

[0095] Figure 2 This is the basic principle diagram of the aircraft initial support plan generation tool based on neural network and historical data provided by the present invention. Figure 2 As shown, the aircraft initial support plan generation tool mainly includes a historical document structured storage module, a database module and a support plan generation module. Among them, the main function of the historical document structured storage module is to extract the support plan entry text from the historical documents imported by the user according to the named entity recognition method based on transfer learning and BiLSTM and the relationship extraction method based on text features and store it in the solution information table of the database module; at the same time, the aircraft identification code is extracted according to the type and use conditions of the aircraft, and combined with the basic information of the aircraft input by the user, it is stored in the aircraft information table of the database module. The main function of the database module is to store aircraft information and plan information, and to associate the two tables with each other in the form of aircraft numbers. The main function of the support plan generation module is to extract the aircraft identification code of the aircraft to be generated according to the type, use conditions and other information of the aircraft to be generated input by the user, and to sort the similarity between the existing aircraft in the database and the aircraft to be generated according to the similarity theory, and after sorting, select the plan text according to the rule of postponement and substitution, and fill the plan text into the preset template, and output the generated aircraft initial support plan.

[0096] See also Figure 2 The working process of the aircraft initial support plan generation tool includes:

[0097] Step 1: Structured storage of historical plans. Historical plan information refers to the content information text of each item in the support plan. The structured storage of plan information is a document containing aircraft support plan information, which can be an alternative plan, initial plan, or a report or summary document. The plan structured storage step is the basis of the plan generation step. Its main task is to structure the plan item text in the support plan information document, extract it, and store it in the database for use in the generation step.

[0098] Step 1-1: Document reading. The user needs to enter the following information: the name of the model involved in the imported document, the year the model was in service, and the document path. Considering the general format of the current document, the imported document should be in PDF format. In the document reading step, the tool extracts the information in the PDF text in the form of a string through the OCR extension package and saves it in a txt text for subsequent operations.

[0099] Step 1-2: Extraction of aircraft identification code (PC): The aircraft identification code is a string of five characters divided and defined according to the type and use conditions of the aircraft. The encoding rules of the identification code, that is, the items represented by each digit, and the types and use conditions represented by different digits are shown in Table 1:

[0100] Table 1 Aircraft identification code encoding rules

[0101]

[0102] In this step, the characters of each digit are determined by checking whether there is a corresponding field in the historical document text. If the description of a certain digit does not exist, it is set to the default value. The default values ​​of each digit are defined as follows: the default value of the first digit (aircraft type) is "fighter" (i.e. "1"); the default value of the second digit (mission type) is "multi-purpose" (i.e. "3"); the default value of the third digit (range) is "medium" (i.e. "2"); the default value of the fourth digit (whether it is carrier-based) is "no" (i.e. "0"); the default value of the fifth digit (whether it is stealth) is "no" (i.e. "0").

[0103] Step 1-3: Aircraft information storage. Aircraft information refers to the model name, model service year and aircraft identification code of the currently imported aircraft data document. The model name and service year are the information entered by the user in step 1-1, and the aircraft identification code is obtained in this step. These three pieces of information, plus the aircraft number as the primary key, constitute the information of the aircraft information table. The aircraft information is stored in the database module built using MySQL through the extension package. The design of the aircraft information table is shown in Table 2.

[0104] Table 2 Aircraft information table

[0105] Data Types int string int string Data Item Aircraft number Model Name Year of service Aircraft identification code

[0106] Step 1-4: Annotate the entity of the protection plan item. i ) refers to a descriptive text phrase for a support plan item, such as maintenance strategy or maintenance method. Figure 3As shown in the figure, according to the txt text saved in step 1-1, read the content and use the trained BiLSTM and transfer learning-based named entity recognition model to identify and extract the security plan item text in the text. First, identify the plan item entities {CD1, CD2, …, CD n} and temporarily store it in the intermediate variable CDtempt, while removing duplicate text fields. Then, the scheme entry entities in the entire text are annotated according to the fields saved as intermediate variables, and the annotated text is saved in .txt format, with the annotation format of "[@text entity#Other*]".

[0107] Step 1-5: Mark the required label identifier. i ) refers to the items that must be included in the protection plan, and its contents are shown in Table 3:

[0108] Table 3 List of required tag entries

[0109]

[0110]

[0111] Since the input historical document data is mostly existing solution information, the text is highly standardized and has a relatively uniform format, so the matching field method is directly used to directly mark the necessary tags with identifiers based on the text obtained in steps 1-4. The format of the identifier is: "[A Necessary Tag BNecessary*]". The text marked with the identifier is stored in the form of .txt.

[0112] Step 1-6: Corresponding entity extraction. In step 1-4, all solution entry entities are extracted and annotated in general, without specifically dividing the necessary tags corresponding to each entity. Therefore, it is necessary to associate the solution entry entities with the necessary tags based on the text annotated with the solution entry entities and necessary tags in step 1-5. Figure 3 As shown in the figure, the text is first divided into different sentences according to the meaning of the sentence, and then the sentences are integrated and spliced ​​into semantic blocks according to the text features, corresponding to the necessary tags. The specific steps are:

[0113] 1) Based on the text features, the text is segmented according to periods and semicolons;

[0114] 2) According to the manual feature engineering analysis of the program text input, it is found that in most cases, the necessary label will appear before the text describing the necessary label. Therefore, the sentences are combined into semantic blocks according to the presence and position of the necessary label. That is, the sentence between the two sentences containing the necessary label is integrated with the sentence containing the necessary label in front into a semantic block, such as Figure 4 shown.

[0115] 3) According to the necessary tag sorting in steps 1-5, traverse the document and store the scheme entry entities in the semantic blocks containing the corresponding necessary tags into the corresponding temporary variable list CDOtempt. If a necessary tag that does not exist in the document is encountered, NULL is used as an identifier placeholder.

[0116] Step 1-7: Save the solution text. Figure 3 As shown, based on the temporary variable CDOtempt in step 1-6, the solution item text is filled into the solution information table of the database module in order. In addition to the solution item text, the solution information table also has the aircraft number as the primary key, where the aircraft number is unified with the aircraft number in the aircraft information table above to achieve the association between the two tables. The solution information table includes the aircraft number and the necessary label texts arranged in order, as shown in Table 4.

[0117] Table 4 Scheme information table

[0118] Data Types int string string string string Data Item Aircraft number Maintenance guarantee system Maintenance strategy ······ Types of security resources

[0119] Step 2: Solution generation. The solution generation step refers to selecting the best solution text based on the aircraft use condition information input by the user, combined with the aircraft information and corresponding support solution information saved in step 1, sorting by similarity and deferral rules, and filling it into the preset template, and saving it in a word format file.

[0120] Step 2-1: Extract the aircraft identification code to be generated. Figure 5 As shown, according to the aircraft type, use conditions and other field data entered by the user in the generation plan interface, the aircraft identification code to be generated is extracted according to the same identification code encoding rules in step 1-2.

[0121] Step 2-2: Sort by aircraft similarity. Similar aircraft refer to aircraft that are similar in type and use conditions. The purpose of selecting similar aircraft is to select the protection plan item text of similar aircraft. The selection of similar aircraft is based on the aircraft identification code, and the following rules are followed during selection:

[0122] (1) If there is an aircraft with the same identification code in the aircraft information table of the database, then the aircraft is selected as a similar aircraft. If there are multiple aircraft, then the aircraft with the latest service year is selected as the similar aircraft;

[0123] (2) For aircraft with inconsistent identification codes in the aircraft information table of the database, the aircraft are sorted according to the similarity of the identification codes, and are sorted after the aircraft with the same identification codes, in the following manner:

[0124] For the aircraft to be generated, there is an aircraft identification code PCw For the aircraft in the database with aircraft numbers 1, 2, ..., n, there are aircraft identification codes {PC1, PC2, ..., PC n The aircraft identification code is a string of 5 digits. According to the ratio of the intersection of the set and the union of the set in the string, the similarity between the aircraft identification code of the existing aircraft in the database and the aircraft identification code of the aircraft to be generated is calculated {PCS1, PCS2, ..., PCS n Then, the aircraft numbers are sorted in descending order of similarity, and the aircraft numbers in this sequence that are the same as the aircraft identification code of the solution to be generated are deleted, and the sequence is sorted after the aircraft numbers sorted in (1).

[0125] Step 2-3: Substitute text selection. Text selection refers to the process of extracting and combining the support plan text from the database. Substitute refers to the process of selecting, considering that some input documents may lack descriptions and texts of certain necessary tags, so it is necessary to select the following plan text to fill the gap according to the sorting. The specific process is: first, select the support plan entry text of the most similar aircraft i according to the aircraft similarity sorting {CD i1 ,CD i2 ,…CD in}, set this set of entry text to be pre-selected to be filled in, and then determine whether the entry contains an empty value placeholder, that is, whether there is a CD in = NULL, that is, whether the data set lacks the necessary label. In the case of a null placeholder, fill in the corresponding missing part of the aircraft support plan text with the next highest similarity. Obviously, if the next aircraft has a description of the required necessary label, the vacancy is filled and the text selection is completed; if the next aircraft still does not have a description of the necessary label, the filled character is still a null placeholder, and the pre-selected text is repeatedly judged to select the next aircraft plan text in turn. Repeat this judgment process until there is no null placeholder in the pre-selected plan text or the existing plans in the database are traversed.

[0126] Step 2-4: Fill in the text. Fill the selected protection plan text into the corresponding position of the protection plan template word file. The method used is: preset the plan text placeholder in the template, each placeholder corresponds to a plan text with a necessary label, and the necessary labels represented by the placeholder are shown in Table 5. When performing the filling operation, traverse the placeholders in the template and replace them with the plan text to be filled in according to the set list element order. After filling, save it as a word file, name it according to the system time, and output the generated plan.

[0127] Table 5 List of required tags and corresponding placeholders

[0128]

[0129]

[0130] The method for generating an initial aircraft support plan based on a neural network and historical data proposed in the present invention has the following advantages over the prior art:

[0131] 1. The data source of the present invention is the guarantee plan or the historical document containing the guarantee plan information, which ensures the validity and credibility of the work items in the generated plan on the one hand, and allows the user to further improve the credibility of the output plan by importing documents with actual application conditions according to the actual situation on the other hand;

[0132] 2. The present invention utilizes the characteristics of the relatively standardized formats of the security scheme and related documents, and adopts a relationship extraction method based on text features. On the one hand, it avoids the re-introduction of deep learning methods in relationship extraction, which brings new uncertainties to the tool and improves the interpretability of the structuring process. On the other hand, it also improves the accuracy and completeness of the structuring;

[0133] 3. The present invention adopts an automated structured storage method for historical document data, which, to a certain extent, relieves the manpower and time costs of manually looking up and refining data. In addition, this method can also be used to specifically construct structured storage methods for other similar fields, which helps artificial intelligence methods to be further applied to these fields;

[0134] 4. In the present invention, the sorting and text selection of similar aircraft corresponds to the construction of a benchmark comparison system in the traditional scheme formulation. The use of similarity theory and the sorting based on similarity reduce the problem of the process being affected by subjective opinions. The automated sorting and priority substitution rules further reduce the time and labor cost of the work at this stage.

[0135] 5. The present invention is highly automated and can generate an initial aircraft support plan based on different needs in a relatively short time. The extremely short generation time cycle and extremely low cost help to quickly obtain a large number of solution texts that can be used for trade-off comparison and iterative optimization in the early stages of aircraft research and development, which can speed up the progress of aircraft comprehensive support work to a certain extent.

[0136] The present invention also provides an aircraft initial support plan generation system, the system comprising:

[0137] The aircraft information table update module is used to extract the aircraft identification code from the historical documents related to the aircraft support plan, and store it together with the corresponding aircraft basic information as a new data in the aircraft information table of the database; the aircraft information table includes the aircraft number, aircraft basic information and aircraft identification code;

[0138] The guarantee plan text acquisition module is used to obtain the guarantee plan text of the annotated plan item entity based on historical documents by using the named entity recognition method based on transfer learning and BiLSTM;

[0139] The scheme information table updating module is used to obtain the marked scheme item entity corresponding to each necessary tag according to the support scheme text of the marked scheme item entity and the relation extraction method based on text features, and store the scheme item text of the marked scheme item entity corresponding to each necessary tag into the scheme information table of the database; the necessary tags are the items that must be included in the aircraft support scheme; the scheme information table includes the aircraft number and all necessary tags;

[0140] A sorting module is used to calculate the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sort the aircraft information table in descending order according to the similarity, and at the same time re-sort the solution information table according to the aircraft number sequence of the sorted aircraft information table;

[0141] The aircraft initial support plan generation module is used to fill in the plan entry text of each necessary tag that appears for the first time in the sorted plan information table into the corresponding necessary tag position of the support plan template of the aircraft to be generated based on the principle of deferred substitution, so as to obtain the aircraft initial support plan of the aircraft to be generated.

[0142] The scheme information table update module includes:

[0143] The submodule for obtaining the identification text of the guarantee plan is used to identify the necessary tags in the guarantee plan text of the marked plan entry entity by using the matching field method, and to add identifiers to obtain the identification text of the guarantee plan;

[0144] The segmentation submodule is used to segment the text according to the text features and the security scheme identification text to obtain multiple sentences;

[0145] A semantic block integration submodule is used to integrate a sentence without necessary tags between two adjacent sentences containing necessary tags and a sentence containing necessary tags that comes first in the word order of the two adjacent sentences containing necessary tags into one semantic block, thereby obtaining multiple semantic blocks;

[0146] The scheme entry text storage submodule is used to use the necessary tags in each semantic block as the necessary tags corresponding to the labeled scheme entry entities in the respective semantic blocks, obtain all the labeled scheme entry entities corresponding to each necessary tag, and store the scheme entry text of the labeled scheme entry entities corresponding to each necessary tag into the scheme information table of the database.

[0147] Sorting module, including:

[0148] A similarity calculation submodule, used to calculate the similarity between the aircraft identification code in the aircraft information table and the aircraft identification code to be generated according to the ratio of the intersection and the union of the set in the character string constituting the aircraft identification code;

[0149] A one-time sorting submodule is used to sort the aircraft information table in descending order of similarity;

[0150] A secondary sorting submodule is used to sort the aircraft information table parts with the same similarity in descending order of the service years in the aircraft information table if there are multiple aircraft information tables with the same similarity;

[0151] The scheme information table sorting submodule is used to re-sort the scheme information table according to the aircraft number sequence of the aircraft information table after the secondary sorting.

[0152] The aircraft initial support plan generation module includes:

[0153] A pre-selected text obtaining submodule is used to use the solution entry texts corresponding to all necessary tags of the first data in the sorted solution information table as pre-selected texts for the aircraft to be generated;

[0154] The judgment result obtaining submodule is used to judge whether the pre-selected text contains a null value placeholder and obtain the judgment result;

[0155] A filling submodule is used for filling the pre-selected text with the solution entry text that first appears under the necessary tag where the empty value placeholder is located according to the arrangement order if the judgment result indicates yes;

[0156] The aircraft initial support plan generation submodule is used to fill the pre-selected text into the corresponding position of the support plan template of the aircraft to be generated if the judgment result indicates no, so as to obtain the aircraft initial support plan of the aircraft to be generated.

[0157] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0158] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for generating an initial aircraft support plan, characterized in that: The method comprises: Extracting an aircraft identification code from historical documents related to the aircraft support plan and storing it together with the corresponding aircraft basic information as a new data in an aircraft information table of a database; the aircraft information table includes an aircraft number, aircraft basic information and an aircraft identification code; According to the historical documents, a named entity recognition method based on transfer learning and BiLSTM is used to obtain the guarantee plan text of the labeled plan item entities; According to the support plan text of the marked plan item entity, a relation extraction method based on text features is used to obtain the marked plan item entity corresponding to each necessary tag, and the plan item text of the marked plan item entity corresponding to each necessary tag is stored in a plan information table of the database; the necessary tag is an item that must be included in the aircraft support plan; the plan information table includes the aircraft number and all necessary tags; Calculate the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sort the aircraft information table in descending order according to the similarity, and re-sort the solution information table according to the aircraft number sequence of the sorted aircraft information table; According to the principle of postponement and substitution, the text of the plan item that appears for the first time for each necessary tag in the sorted plan information table is filled into the corresponding necessary tag position of the support plan template of the aircraft to be generated, so as to obtain the initial aircraft support plan of the aircraft to be generated; According to the security plan text of the labeled plan item entity, a relationship extraction method based on text features is used to obtain the labeled plan item entity corresponding to each necessary tag, and the plan item text of the labeled plan item entity corresponding to each necessary tag is stored in the plan information table of the database, specifically including: Using a matching field method to identify necessary tags in the safeguard plan text of the marked plan entry entity, and marking an identifier to obtain a safeguard plan identification text; Segment the security solution identification text according to text features to obtain multiple sentences; Integrate a sentence without a necessary label between two adjacent sentences containing necessary labels and a sentence containing necessary labels that comes first in the word order among the two adjacent sentences containing necessary labels into a semantic block, thereby obtaining multiple semantic blocks; The necessary tags in each semantic block are used as the necessary tags corresponding to the annotated solution entry entities in the respective semantic blocks, all the annotated solution entry entities corresponding to each necessary tag are obtained, and the solution entry text of the annotated solution entry entity corresponding to each necessary tag is stored in the solution information table of the database.

2. The method for generating an initial aircraft support plan according to claim 1, characterized in that: The extracting of the aircraft identification code from the historical documents related to the aircraft support plan specifically includes: The information in the historical documents is stored in txt text in the form of strings through the OCR extension package; According to the txt text, the aircraft identification code is extracted according to the aircraft identification code encoding rules.

3. The method for generating an initial aircraft support plan according to claim 1, characterized in that: The calculating the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sorting the aircraft information table in descending order according to the similarity, and re-sorting the solution information table according to the aircraft number sequence of the sorted aircraft information table, also includes: Get the basic information field data of the aircraft to be generated; According to the basic information field data of the aircraft to be generated and in accordance with the aircraft identification code coding rules, the aircraft identification code to be generated is extracted.

4. The method for generating an initial aircraft support plan according to claim 1, characterized in that: The calculating the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sorting the aircraft information table in descending order according to the similarity, and re-sorting the solution information table according to the aircraft number sequence of the sorted aircraft information table, specifically includes: The similarity between the aircraft identification code in the aircraft information table and the aircraft identification code to be generated is calculated according to the ratio of the intersection and union of the sets in the character string constituting the aircraft identification code; Sort the aircraft information table once according to the order of similarity from large to small; If there are multiple aircraft with the same similarity, the aircraft information table parts with the same similarity are sorted again in descending order of the service years in the aircraft information table; The plan information table is re-sorted according to the aircraft number sequence of the aircraft information table after the secondary sorting.

5. The method for generating an initial aircraft support plan according to claim 1, characterized in that: According to the principle of postponement and substitution, the text of the plan item that appears for the first time for each necessary tag in the sorted plan information table is filled into the corresponding necessary tag position of the support plan template of the aircraft to be generated, so as to obtain the initial aircraft support plan of the aircraft to be generated, specifically including: The scheme entry texts corresponding to all necessary tags of the first data in the sorted scheme information table are used as pre-selected texts for the aircraft to be generated; Determine whether the pre-selected text contains a null value placeholder, and obtain a determination result; If the judgment result indicates yes, then fill the pre-selected text with the solution entry text that appears first under the necessary tag where the empty value placeholder is located according to the arrangement order; If the judgment result indicates no, the pre-selected text is filled into the corresponding position of the support plan template of the aircraft to be generated to obtain the initial aircraft support plan of the aircraft to be generated.

6. An aircraft initial support plan generation system, characterized in that: The system comprises: An aircraft information table update module is used to extract an aircraft identification code from historical documents related to the aircraft support plan, and store it in an aircraft information table of the database as a new data item together with the corresponding aircraft basic information; the aircraft information table includes an aircraft number, aircraft basic information and aircraft identification code; A guarantee plan text acquisition module, used to obtain the guarantee plan text of the labeled plan item entity according to the historical document by adopting a named entity recognition method based on transfer learning and BiLSTM; A scheme information table updating module is used to obtain the marked scheme item entity corresponding to each necessary tag according to the support scheme text of the marked scheme item entity and the relation extraction method based on text features, and store the scheme item text of the marked scheme item entity corresponding to each necessary tag into the scheme information table of the database; the necessary tag is an item that must be included in the aircraft support scheme; the scheme information table includes the aircraft number and all necessary tags; A sorting module is used to calculate the similarity between each aircraft identification code in the aircraft information table and the aircraft identification code to be generated, and sort the aircraft information table in descending order according to the similarity, and at the same time re-sort the solution information table according to the aircraft number sequence of the sorted aircraft information table; The aircraft initial support plan generation module is used to fill the plan entry text of each necessary tag that appears for the first time in the sorted plan information table into the corresponding necessary tag position of the support plan template of the aircraft to be generated according to the principle of postponement and substitution, so as to obtain the aircraft initial support plan of the aircraft to be generated; The scheme information table updating module specifically includes: A submodule for obtaining a protection scheme identification text is used to identify necessary tags in the protection scheme text of the marked scheme entry entity by using a matching field method, and to add an identifier to obtain the protection scheme identification text; A segmentation submodule, used for segmenting the security solution identification text according to text features to obtain multiple sentences; A semantic block integration submodule is used to integrate a sentence without necessary tags between two adjacent sentences containing necessary tags and a sentence containing necessary tags that comes first in the word order of the two adjacent sentences containing necessary tags into one semantic block, thereby obtaining multiple semantic blocks; The scheme entry text storage submodule is used to use the necessary tags in each semantic block as the necessary tags corresponding to the labeled scheme entry entities in the respective semantic blocks, obtain all the labeled scheme entry entities corresponding to each necessary tag, and store the scheme entry text of the labeled scheme entry entities corresponding to each necessary tag into the scheme information table of the database.

7. The aircraft initial support plan generation system according to claim 6, characterized in that: The sorting module specifically includes: A similarity calculation submodule, used to calculate the similarity between the aircraft identification code in the aircraft information table and the aircraft identification code to be generated according to the ratio of the intersection and the union of the set in the character string constituting the aircraft identification code; A one-time sorting submodule is used to sort the aircraft information table in descending order of similarity; A secondary sorting submodule is used to sort the aircraft information table parts with the same similarity in descending order of the service years in the aircraft information table if there are multiple aircraft information tables with the same similarity; The scheme information table sorting submodule is used to re-sort the scheme information table according to the aircraft number sequence of the aircraft information table after the secondary sorting.

8. The aircraft initial support plan generation system according to claim 6, characterized in that: The aircraft initial support plan generation module specifically includes: A pre-selected text obtaining submodule, used to use the solution entry texts corresponding to all necessary tags of the first data in the sorted solution information table as pre-selected texts for the aircraft to be generated; A judgment result obtaining submodule is used to judge whether the pre-selected text contains a null value placeholder and obtain a judgment result; A filling submodule is used for filling the pre-selected text with the solution entry text that appears for the first time under the necessary tag where the empty value placeholder is located according to the arrangement order if the judgment result indicates yes; The aircraft initial support plan generating submodule is used to fill the pre-selected text into the corresponding position of the support plan template of the aircraft to be generated if the judgment result indicates no, so as to obtain the aircraft initial support plan of the aircraft to be generated.

Images