Engineering construction technical problem processing management system and method based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based engineering construction technical problem processing management system and method, and the system comprises a data collection module which is used for collecting technical contact lists and historical processing information; the artificial intelligence large model processing module is used for analyzing newly submitted technical problems, identifying similar problems and extracting historical solutions or assisting in generating solutions; the contact list generation and approval module is used for automatically generating a construction technology contact list with a number according to the solution, associating the construction technology contact list with a drawing file, and starting an approval process according to a preset approval rule; the data storage module is used for storing technical contact lists, historical processing information and solutions; the query retrieval module is used for querying historical processing information and solutions; and the user interaction module is used for providing an operation interface. Through intelligent and automatic technical means, efficient and intelligent processing of construction technical problems is achieved, and the overall efficiency and quality level of project management are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of engineering construction management, and in particular to an engineering construction technical problem processing management system and method based on artificial intelligence. BACKGROUND

[0002] In large-scale engineering construction projects, the collaborative work among the engineering user party, the construction party and the engineering design unit is a key factor to ensure the smooth progress of the project. However, the technical problems generated during the construction process have significant diversity and complexity, covering design errors, new requirements proposed by the engineering user party, actual difficulties encountered by the construction party during the construction process, and various special requirements proposed by other related parties.

[0003] Currently, the traditional construction technical problem processing mode mainly relies on the engineering user party or the construction party to feed back the problems occurring on site to the engineering design unit in the form of a technical contact sheet. The design unit then replies to the feedback party with corresponding solutions and processing opinions through the construction technical contact sheet. This processing method has exposed many drawbacks in actual application.

[0004] For projects with series products, the same technical problems often occur repeatedly. The staff responsible for technical problem processing has to invest a lot of time and effort in repetitive work, which not only is inefficient, but also is prone to human error. At the same time, due to the lack of unified and perfect management and query mechanism, the processing methods for the same problems by different personnel at different stages often have contradictions, making it difficult for relevant personnel to quickly and accurately obtain effective processing solutions for the same problems. The existence of these problems not only seriously delays the progress of the project, but also causes huge economic losses, time cost waste and human cost consumption, which has become an important bottleneck restricting the smooth development and economic benefit improvement of the project. Therefore, it is urgent to design an engineering construction technical problem processing management system and method based on artificial intelligence to solve the above problems in the prior art. SUMMARY

[0005] In view of the above defects or improvement needs of the prior art, the present application provides an engineering construction technical problem processing management system and method based on artificial intelligence, which aims to realize rapid and accurate processing of construction technical problems through intelligent and automated technical means, effectively avoid repetitive work, eliminate contradictions and inconsistencies in the processing process, and thus greatly improve the engineering management efficiency and significantly reduce the construction cost.

[0006] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0007] The present application provides an engineering construction technical problem processing management system based on artificial intelligence, comprising:

[0008] a data collection module configured to collect technical contact sheets submitted by engineering user parties or construction parties, construction technical contact sheets replied by design parties, and relevant historical processing information;

[0009] an artificial intelligence large model processing module configured to receive data collected by the data collection module, analyze newly submitted technical problems, identify similar problems, and extract historical solutions or assist in generating new solutions;

[0010] a contact sheet generation and approval module configured to automatically generate construction technical contact sheets with numbers and associate drawing files according to solutions output by the artificial intelligence large model processing module, and start an approval process according to preset approval rules;

[0011] a data storage module configured to store data collected by the data collection module, solutions generated by the artificial intelligence large model processing module, and technical contact sheets and associated drawing files generated by the contact sheet generation and approval module;

[0012] a query and retrieval module configured to query relevant historical processing information and solutions in the data storage module according to various conditions;

[0013] a user interaction module configured to provide an operation interface for users to submit technical contact sheets, query processing progress, view solutions, participate in the approval process, and provide feedback.

[0014] As an embodiment of the present application, the artificial intelligence large model processing module compares the newly submitted technical problem with the historical problem data stored in the system in all aspects to determine whether the newly submitted technical problem is a similar problem:

[0015] If it is a similar problem, relevant solutions are extracted from a historical solution library and optimized based on the current scenario;

[0016] If it is not a similar problem, the engineering field knowledge is called to assist the design party in generating a solution.

[0017] As an embodiment of the present application, the data storage module uses a database management system to classify and store data and establish an indexing mechanism.

[0018] As an embodiment of the present application, the query and retrieval module queries according to contact sheet numbers, problem categories, product model codes, and processing personnel, and supports fuzzy query function.

[0019] As an embodiment of the present application, the contact sheet generation and approval module automatically generates a unique number for each construction technical contact sheet.

[0020] As an embodiment of the present application, the user interaction module also has a message reminding function, which is used to inform the user of the relevant operation and processing result.

[0021] The second aspect of the present application also provides a method of the artificial intelligence-based engineering construction technical problem processing management system as described in the first aspect, comprising the following steps:

[0022] Step S1, collecting technical contact sheets submitted by engineering user parties or construction parties, design party replies to construction technical contact sheets and related historical processing information in real time and dynamically through a data collection module;

[0023] Step S2, receiving data collected by the data collection module through an artificial intelligence large model processing module, analyzing newly submitted technical problems, identifying similar problems and extracting historical solutions or assisting in generating new solutions;

[0024] Step S3, generating a technical contact sheet based on the solution output by the artificial intelligence large model processing module through a contact sheet generation and approval module, automatically generating a construction technical contact sheet with a number and associating a drawing file, and starting an approval process according to a preset approval rule;

[0025] Step S4, storing the data collected by the data collection module, the solution generated by the artificial intelligence large model processing module, and the technical contact sheet and associated drawing file generated by the contact sheet generation and approval module through a data storage module;

[0026] Step S5, the user queries the relevant historical processing information and solutions in the data storage module through a query retrieval module;

[0027] Step S6, the user submits a technical contact sheet, queries the processing progress, views the solution, participates in the approval process, and provides feedback through the user interaction module.

[0028] As an embodiment of the present application, when the artificial intelligence large model processing module analyzes the newly submitted technical problem in step S2, it first determines whether it is a similar problem:

[0029] If it is a similar problem, relevant solutions are extracted from a historical solution library and optimized based on the current scene;

[0030] If it is not a similar problem, the engineering field knowledge is called to assist the design party in generating a solution.

[0031] As an embodiment of the present application, when the approval process is started according to the preset approval rule in step S3, the approval personnel need to complete the approval operation within the specified time limit, and the technical contact sheet that passes the approval will be sent to the engineering construction unit in time, and the technical contact sheet that does not pass the approval will be returned to the design party for modification.

[0032] As an embodiment of the application, in the step S5, the user can query according to multiple conditions such as contact sheet number, problem classification, product model code, and processing personnel, and the query search module supports fuzzy query function.

[0033] The application has the following beneficial effects:

[0034] 1. The application collects technical contact sheets and related historical processing information through the data acquisition module and transmits them to the artificial intelligence big model processing module, which has strong data analysis and processing capabilities, can quickly identify similar problems, and automatically provide corresponding solutions, effectively reducing the time and manpower waste caused by repetitive labor and processing conflicts, greatly shortening the processing period of technical problems, significantly improving the efficiency of engineering management, saving valuable time for the smooth progress of engineering construction projects, effectively reducing the economic cost, time cost and manpower cost in the engineering construction process, saving a large amount of resources for enterprises, improving the economic benefit and competitiveness of the project;

[0035] 2. The application stores all data collected during system operation safely and efficiently through the data storage module, including technical contact sheet data, historical processing information, and solution library and other key data resources, effectively ensuring processing consistency, and processing personnel can refer to previous processing methods and experience at any time when facing new problems, ensuring that the processing method for similar problems remains consistent, effectively avoiding conflicts and errors caused by inconsistent processing methods, and ensuring the accuracy and standardization of technical problem processing;

[0036] 3. The application provides convenient and flexible query search services for users through the query search module, and users can accurately query according to multiple conditions such as contact sheet number, problem classification, product model code, and processing personnel, while supporting fuzzy query function, effectively improving the flexibility and accuracy of the query, facilitating comprehensive and systematic management and analysis of engineering construction technical problems, and providing strong data support for engineering decision-making;

[0037] 4. The application automatically triggers a new technical contact sheet generation process according to the solution generated by the artificial intelligence big model processing module through the contact sheet generation and approval module, realizes the automation of the contact sheet generation, approval and sending process, reduces human intervention, improves work efficiency, reduces the risk of human operation errors, and ensures the efficient and accurate operation of the problem processing process. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 The application provides an engineering construction technical problem processing management system block diagram based on artificial intelligence;

[0039] Figure 2The method flowchart for processing and managing engineering construction technical problems based on artificial intelligence provided in the embodiments of the present application. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present application.

[0041] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present application are only used to explain the relative positional relationship, movement condition, etc. between components in a certain posture (as shown in the drawings), and if the certain posture changes, the directional indications also change accordingly.

[0042] In the present application, unless otherwise explicitly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense, for example, "fixation" can be fixed connection, or detachable connection, or integral; can be mechanical connection, or electrical connection; can be direct connection, or indirect connection through an intermediate medium; can be internal communication of two elements or interaction relationship between two elements, unless otherwise explicitly limited. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0043] In addition, if the present application has descriptions involving "first", "second", etc., the descriptions of "first", "second", etc. are only for the purpose of description, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features. Therefore, the features limited by "first" and "second" can explicitly or implicitly include at least one of the features. In addition, the meaning of "and / or" appearing throughout the text includes three parallel solutions. Taking "A and / or B" as an example, it includes A solution, or B solution, or A and B solutions. In addition, the technical solutions of each embodiment can be combined with each other, but it must be based on the fact that a person skilled in the art can realize it. When the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist, and is not within the protection scope claimed by the present application.

[0044] Reference Figure 1 The first aspect of the present application provides an engineering construction technical problem processing and management system based on artificial intelligence, comprising a data acquisition module, an artificial intelligence large model processing module, a data storage module, a query retrieval module, a contact single generation and approval module and a user interaction module, wherein:

[0045] a data collection module configured to collect technical contact sheets submitted by engineering user parties or construction parties, construction technical contact sheets replied by design parties, and relevant historical processing information.

[0046] an artificial intelligence large model processing module configured to receive data collected by the data collection module, analyze newly submitted technical problems, identify similar problems, and extract historical solutions or assist in generating new solutions;

[0047] a contact sheet generation and approval module configured to automatically generate construction technical contact sheets with numbers and associate drawing files according to solutions output by the artificial intelligence large model processing module, and start an approval process according to preset approval rules;

[0048] a data storage module configured to store data collected by the data collection module, solutions generated by the artificial intelligence large model processing module, and technical contact sheets and associated drawing files generated by the contact sheet generation and approval module;

[0049] a query and retrieval module configured to query relevant historical processing information and solutions in the data storage module according to various conditions;

[0050] a user interaction module configured to provide an operation interface for users to submit technical contact sheets, query processing progress, view solutions, participate in the approval process, and provide feedback.

[0051] As an embodiment of the present application, the data collection module undertakes the key task of system data source, and is responsible for collecting technical contact sheets submitted by engineering user parties or construction parties, technical contact sheets replied by design parties, and relevant historical processing information. Each technical contact sheet records detailed description of technical problems, including time, place, specific phenomena, influence on the project, etc., and uploads relevant drawing files, photos, videos, and other materials, fills in important information such as product model code, product specific equipment name, etc. The technical contact sheets replied by design parties cover solutions, processing opinions, etc. In addition, historical processing information includes contact sheet number, associated drawing number and file number, processing time, processing personnel, contact sheet recipient, problem classification, and other multi-dimensional data, providing a solid data foundation for subsequent intelligent analysis and processing of the system.

[0052] As an embodiment of the present application, the artificial intelligence large model processing module serves as the core hub of the entire system, is constructed and trained based on an open-source deep learning framework such as TensorFlow or PyTorch, to implement advanced artificial intelligence algorithms and models for receiving various data collected by the data collection module, performing deep intelligent analysis on newly submitted technical problems, identifying similar problems and extracting historical solutions or assisting in generating new solutions, effectively avoiding repetitive work, greatly shortening the processing period of technical problems, significantly improving engineering management efficiency, and gaining valuable time for the smooth progress of engineering construction projects.

[0053] Specifically, the artificial intelligence large model processing module compares with the historical problem data stored in the system in all directions, quickly and accurately judges whether the newly submitted technical problem is a similar problem:

[0054] If it is a similar problem, relevant solutions are accurately extracted from the historical solution library and combined with the specific scene and actual demand of the current problem for targeted optimization and adjustment;

[0055] If it is not a similar problem, the professional knowledge and rich experience in the engineering field are fully called upon to assist the design party in efficiently generating scientific and reasonable solutions.

[0056] As an embodiment of the present application, the specific steps of the artificial intelligence large model processing module to determine whether it is a similar problem are as follows:

[0057] First, text preprocessing: standardizing the text description of the newly submitted technical problem, including word segmentation (using tools such as Jieba to split continuous text into independent words), stop word removal (filtering out words such as "of" and "in" that have no actual meaning), part-of-speech tagging (annotating nouns, verbs, etc.), and entity recognition (extracting key entities such as engineering terms, equipment names, and product models in the problem), to form a structured text feature sequence.

[0058] Second, feature extraction: based on a pre-trained language model (such as BERT, RoBERTa, etc.), the preprocessed text is encoded to generate a vector representation containing semantic information, capturing the context meaning of the problem description. Extract structured features, including the engineering type (building / bridge / water conservancy, etc.) to which the problem belongs, the professional field (structure / mechanical and electrical / rock and soil, etc.) involved, the severity of the problem (mild / normal / urgent), and the associated equipment model, to form a multi-dimensional feature set.

[0059] Next, establish a historical data index: the historical problem data stored in the system is preprocessed and feature extracted using the same method, a structured feature library is constructed, and efficient indexing is established through a vector database (such as Milvus, FAISS) to support fast similarity retrieval.

[0060] Then, similarity calculation in three aspects is performed, semantic similarity: the matching degree of the new question text vector and the historical question text vector is calculated by cosine similarity, Euclidean distance and other algorithms, and the similarity at the semantic level is quantified. Feature matching degree: the structured features are matched with weights, such as setting the matching weight of core features such as engineering type and equipment model to 0.3, and setting the weight of auxiliary features such as professional field and problem severity to 0.2, to form a comprehensive feature matching score. Comprehensive similarity: the semantic similarity and the feature matching degree are weighted and fused in the ratio of 6:4 to obtain a comprehensive similarity score between 0 and 1.

[0061] Finally, threshold determination: the system presets a similarity threshold (the default value is 0.75, which can be dynamically adjusted according to the engineering scene), and when the comprehensive similarity score of the new question and a certain historical question is ≥ the threshold, it is determined to be a similar problem; if the scores of all historical questions are < the threshold, it is determined to be a new problem.

[0062] In addition, the artificial intelligence large model processing module also continuously iteratively optimizes, regularly collects user feedback on similar problem identification results (such as misjudgment and missed cases), updates language model parameters and feature weights through incremental training, and continuously optimizes identification accuracy.

[0063] The artificial intelligence large model processing module adjusts the retrieved historical solutions through the Transformer decoder, such as automatically supplementing "low-temperature maintenance measures" in the historical solutions according to the "winter construction" scene in the new question. For problems without matching history, call the engineering knowledge graph (stored in Neo4j) for reasoning, combine the specification clauses to generate a preliminary solution framework, and then fill in the specific implementation steps through the GPT-2 fine-tuning model to assist in generating a new solution.

[0064] As an embodiment of the present application, the contact sheet generation and approval module automatically triggers the new construction technology contact sheet generation process. The approval personnel need to complete the approval operation within the specified time limit, and the approved contact sheet will be sent to the engineering construction unit in time to ensure that the problem processing result can be executed in time, and to ensure the efficient operation of the problem processing process. Among them, the number of each construction technology contact sheet generated by the contact sheet generation and approval module is a unique number, which ensures the uniqueness and traceability of each contact sheet. The contact sheet generation and approval module of the present application realizes the automation of the contact sheet generation, approval and sending process, reduces human intervention, improves work efficiency, reduces the risk of human operation errors, and ensures the efficient and accurate operation of the problem processing process.

[0065] Specifically, the specific content of the pre-set approval rules in the contact sheet generation and approval module is as follows:

[0066] First, the approval level is set. First-level approval: For general technical problems (such as minor level design adjustment, routine construction questions), the professional engineer of the engineering design unit (with corresponding professional intermediate and above title) acts as the approver. Second-level approval: For important technical problems (such as major level structural optimization, modification involving safety specifications), the professional engineer of the design unit needs to pass the preliminary examination, and then submit it to the project chief engineer for approval. Third-level approval: For major technical problems (such as critical level design scheme change, adjustment that may affect the project period or cost), after the second-level approval, it needs to be reported to the technical person in charge of the engineering construction unit for final approval.

[0067] Second, the approval authority is distributed. The system controls the approval authority through user roles and permission matrix. The specific distribution is shown in the following table:

[0068] Approval level Approver role Scope of authority Approval time limit First-level approval Professional engineer General issues within the professional field 24 hours Second-level approval Project chief engineer Cross-professional coordination issues, important technical solutions 48 hours Third-level approval Technical director of the construction unit Major changes, cost impact issues 72 hours

[0069] Next, set the approval process trigger condition. Automatic triggering: The system automatically matches the corresponding approval level and assigns the approver according to the problem classification label (identified by the AI module) and the preset rules (such as "structure safety" label automatically triggers third-level approval). Manual intervention: When there is a dispute in AI classification, the design party can apply for manual adjustment of the approval level, which will take effect after being reviewed by the system administrator.

[0070] Finally, execute the approval operation specification. Approval passed: The approver needs to fill in the "agree to execute" opinion, and can add specific implementation suggestions. The system automatically records the approval time and electronic signature of the approver. Approval not passed: The reason for rejection needs to be clearly filled out (from the preset options such as "incomplete scheme", "not in line with the specification", "need to supplement data", and can add custom explanation), the system automatically marks the contact sheet status as "rejected for modification" and returns to the design party.

[0071] As an embodiment of the present application, if the approver does not complete the approval operation within the specified time limit, a timeout unprocessed coping mechanism is started, which includes a timeout warning mechanism, a timeout processing rule, a timeout record and trace, and a special rule for emergency situations. The timeout warning mechanism includes: first reminder: when the remaining time is less than 1 / 3 of the total time limit (such as 8 hours before the 24-hour approval time limit), the system sends a timeout warning notice to the approver through in-station messages, SMS and email, and clearly prompts the remaining time and the link of the matter to be approved. Second reminder: if it is still not handled 1 / 5 of the time limit away from the timeout (such as 4.8 hours before the 24-hour approval time limit), the system automatically sends a reminder notice to the approver and his direct superior, and simultaneously displays the urgency of the contact sheet to be approved.

[0072] The overtime processing rules include: first-level approval overtime, more than 24 hours without processing, the system automatically transfers the contact sheet to other engineers with approval authority in the same profession, and sends an overtime record notice to the original approver. Second-level approval overtime: more than 48 hours without processing, the system automatically upgrades to the third-level approver, and sends a warning notice to the original second-level approver, and the third-level approver decides to continue approval or reassign. Third-level approval overtime: more than 72 hours without processing, the system triggers the highest level of warning, synchronously notifies the project manager of the construction unit and the manager of the design unit, starts the manual coordination process, and ensures that the approval responsibility person is clear within 24 hours.

[0073] The overtime record and traceability includes: the system adds "overtime identifier", "transfer reason" and "overtime processing result" fields in the approval record table (approval_record), which records the processing process of the overtime event, and serves as the basis for optimizing the approval process.

[0074] The emergency special rules include: for technical contact sheets marked as "urgent" (such as issues affecting construction safety and causing stoppage), the approval time limit is shortened by 50% (8 hours for the first level, 24 hours for the second level, and 48 hours for the third level), and the overtime warning is advanced to 1 / 2 of the remaining time, ensuring rapid response.

[0075] As an embodiment of the present application, the data storage module uses a high-performance database management system to store data safely and efficiently. During storage, the data is carefully classified and managed, and an efficient indexing mechanism is established to ensure that data can be quickly stored, retrieved, and updated. This greatly facilitates users to quickly query and call the required data, ensures that the handling methods for similar problems remain consistent, effectively avoids conflicts and errors caused by inconsistent handling methods, and provides reliable data storage protection for stable operation and efficient service of the system.

[0076] Preferably, the data storage module uses MySQL 8.0 as the main database management system, and combines Redis cache database to build a two-level storage architecture. The main database is responsible for persistently storing full data, supporting transaction processing and complex queries; Redis is used to cache high-frequency access data (such as recent technical contact sheets and mainstream solutions), improving query response speed. The database uses a master-slave replication architecture, where the master database is responsible for data write operations and the slave database handles query requests, ensuring data security and system availability.

[0077] The data storage module classifies technical contact forms by data source and state into three subcategories: pending contact forms, handled contact forms, and rejected contact forms. The pending contact forms store new, unprocessed contact forms submitted by engineering users or construction parties, containing fields such as contact form unique ID, submission time, submitting unit, problem description text, associated drawing file path, product model, equipment name, problem classification label, and processing status (pending receipt / processing). The handled contact forms store contact forms with completed responses from the design side, adding fields such as solution text, response time, processing personnel ID, and approval status based on the pending contact form fields. The rejected contact forms store contact forms that have failed approval, adding fields such as rejection reason, rejection time, and rejection personnel ID.

[0078] The data storage module divides historical processing information into annual historical data and engineering type data based on time dimension and engineering type. The annual historical data is stored by natural year and contains full life cycle data of contact forms (complete records from submission to processing to approval to archiving). The engineering type data is stored in tables by categories such as building engineering, bridge engineering, and water conservancy engineering, with each sub-table containing exclusive fields (such as span and structure type fields for bridge engineering).

[0079] The data storage module divides solutions into structured solutions and case library based on problem areas and applicable scenarios. The structured solutions store reusable standardized solutions, containing solution ID, problem type label, applicable product model, solution text, attachment file path, usage frequency, and update time. The case library stores processing cases for complex problems, adding fields such as case analysis report, expert comments, and on-site processing photo path.

[0080] The data storage module uses a relational data table structure, with core data tables including technical contact form table (contact_form), user information table (user_info), solution table (solution_library), approval record table (approval_record), and historical operation log table (operation_log). The technical contact form table is associated with the solution table and the approval record table in a one-to-many manner through the contact form ID. The user information table is associated with other tables through the user ID to achieve data traceability. Binary files such as drawings and photos are stored using a file server, and the database only records the unique identifier and access path of the file (format: / engineering number / contact form ID / file name.extension). A distributed file system (such as MinIO) is also used to manage file storage, supporting file sharding upload, breakpoint resume, and redundant backup to ensure the safe storage of large engineering drawings.

[0081] The data storage module builds an index mechanism including a primary key index, a secondary index design, and a vector index supplement. The primary key index establishes a primary key index for all data tables. The technical contact sheet takes the "contact ID" as the primary key, and the solution sheet takes the "solution ID" as the primary key to ensure data uniqueness. The secondary index design includes the technical contact sheet: a composite index (submission time + processing status) and a single field index (product model, problem classification label) are established to accelerate time range queries and state screening. The solution sheet: a full-text index (problem description + solution text) is established to support keyword fuzzy queries; a hash index (applicable product model) is established to optimize exact match queries. The historical data table: a partition index is established, and a combined index is built according to the annual partition key and the engineering type label to improve cross-year historical data query efficiency. The vector index supplement is the BERT encoding vector of the solution text. A Milvus vector database is used to establish an approximate nearest neighbor search index to support solution recommendation queries based on semantic similarity. The vector index is associated with the relational database through the solution ID to realize joint query of structured data and unstructured vector data.

[0082] The data storage module also establishes a data management mechanism. Through an automatic archiving strategy, the system automatically migrates processed contact sheets that are more than 1 year old from the main table to the historical archive table every month. Data partition rotation is realized through a stored procedure to keep the main table lightweight. Through a data backup mechanism, full backup is performed every morning, and incremental backup is performed every 6 hours. Backup files are encrypted and stored on an off-site server, and backup history is retained for 30 days to support data recovery at a time point. An index maintenance plan is developed to perform index optimization tasks every Sunday morning, including index fragmentation cleanup and statistical information update to ensure stable query performance. For large tables with more than 100 million records, online index reconstruction is enabled to avoid business interruption during maintenance.

[0083] As an embodiment of the present application, the query retrieval module obtains data from the data storage module to meet the user's query requirements and provide convenient and flexible query retrieval services for users. Specifically, users can perform accurate queries according to contact number, problem classification, product model code, and processing personnel, and also support fuzzy query functions to effectively improve the flexibility and accuracy of queries. Through this module, users can quickly obtain relevant historical processing information and solutions, which is convenient for reference and provides strong support for the analysis and solution of technical problems.

[0084] As an embodiment of the present application, the user interaction module is dedicated to providing an easy-to-use operation interface for engineering users, construction builders, designers and managers. Users can conveniently submit technical contact sheets, query processing progress, view solutions, participate in approval processes, and provide feedback through the module. In addition, the user interaction module also has a message reminder function to inform users of relevant operations and processing results, and to keep users informed of the latest developments and results of problem processing. If users are not satisfied with the processing results, they can provide feedback through the module, and the system will further process the feedback information to achieve efficient interaction between users and the system and improve user experience.

[0085] Referring to Figure 2 The second aspect of the present application also provides a method for processing and managing engineering construction technical problems based on artificial intelligence, as described in the first aspect, comprising the following steps:

[0086] Step S1, collecting technical contact sheets submitted by engineering users or construction builders, construction technical contact sheets replied by designers and related historical processing information in real time and dynamically through the data collection module. After collection is completed, these data are transmitted to the artificial intelligence large model processing module and the data storage module in time to provide data support for subsequent processing;

[0087] Step S2, receiving data collected by the data collection module through the artificial intelligence large model processing module, analyzing newly submitted technical problems, identifying similar problems and extracting historical solutions or assisting in generating new solutions;

[0088] Step S3, generating a technical contact sheet with a number and associating a drawing file according to the solution output by the artificial intelligence large model processing module through the contact sheet generation and approval module, and starting an approval process according to a pre-set approval rule;

[0089] Step S4, classifying and storing the data collected by the data collection module, the solutions generated by the artificial intelligence large model processing module, and the technical contact sheets and associated drawing files generated by the contact sheet generation and approval module through the data storage module, and establishing detailed indexes for subsequent user quick query and retrieval;

[0090] Step S5, users query relevant historical processing information and solutions in the data storage module through the query and retrieval module;

[0091] Step S6, users submit technical contact sheets, query processing progress, view solutions, participate in approval processes and provide feedback through the user interaction module.

[0092] As an embodiment of the present application, in the step S2, the advanced algorithm and model of the artificial intelligence large model processing module are used to analyze the newly submitted technical problem in depth, identify similar problems, and extract historical solutions or assist in generating new solutions. When the artificial intelligence large model processing module analyzes the newly submitted technical problem, it first determines whether it is a similar problem:

[0093] If it is a similar problem, relevant solutions are extracted from the historical solution library and optimized and improved based on the current scenario;

[0094] If it is not a similar problem, the engineering field knowledge and experience are comprehensively called to assist the design party in generating a feasible solution.

[0095] As an embodiment of the present application, in the step S5, the user can input the corresponding query conditions through the query retrieval module according to the user's own needs, and the system performs rapid retrieval in the data storage module, queries the relevant historical processing information and solutions, and returns them to the user in a timely manner, facilitating the user to refer to them. Specifically, the user can query according to the contact sheet number, problem classification, product model code, processing personnel, processing time, etc. The query retrieval module supports fuzzy query function, effectively improving the flexibility and accuracy of the query.

[0096] As an embodiment of the present application, in the step S3, when the approval process is started according to the preset approval rules, the approval personnel need to complete the approval operation within the specified time limit. The technical contact sheet that passes the approval will be sent to the engineering construction unit in a timely manner, and the technical contact sheet that does not pass the approval will be returned to the design party for modification.

[0097] As an embodiment of the present application, in the step S6, the user can perform various operations through the user interaction module, such as submitting a technical contact sheet, querying the processing progress, viewing the solution, participating in the approval process, etc. If the user is not satisfied with the processing result, he can fill in the feedback opinion through the feedback function of the user interaction module and submit it to the system. The system further processes according to the feedback information, such as reanalyzing the problem, adjusting the solution, etc., until the user is satisfied. At the same time, the user can also receive the message reminder sent by the system in real time through the module, and timely understand the latest dynamics and results of the problem processing.

[0098] As a preferred embodiment of the present application, the system operating environment specifically includes:

[0099] System Hardware Environment: This system is deployed on a high-performance server, and the server hardware configuration requirements are as follows: CPU uses multi-core processor (such as Intel Xeon series), main frequency not less than 2.5GHz, to ensure that the system has strong computing power; memory not less than 32GB, to ensure the smoothness of the system when processing large amounts of data and complex operations; hard disk capacity not less than 1TB of high-speed storage device, to meet the needs of system data storage and fast read-write. At the same time, it needs to be equipped with stable network environment to ensure the rapidity and stability of data interaction between each user terminal and the system.

[0100] System Software Environment: This system is developed using mainstream programming languages such as Java or Python, fully utilizing their powerful functions and extensive application ecosystem. The database uses large relational database management systems such as MySQL or Oracle to ensure the safety, reliability and efficiency of data storage. The artificial intelligence big model processing module is built and trained based on open-source deep learning frameworks such as TensorFlow or PyTorch to implement advanced artificial intelligence algorithms and models. The user interaction interface is developed using front-end technologies such as HTML, CSS and JavaScript to create a friendly and easy-to-use user interface, improving user experience.

[0101] Specifically, the special field AI model built based on the TensorFlow or PyTorch framework adopts a double-layer architecture of "base model + field fine-tuning", and the main steps include base model layer and field adaptation layer construction. The base model layer selects BERT-Base (Chinese pre-training version) as the core framework, and its 12-layer Transformer structure can effectively capture the text context semantics. The pre-training weight is loaded in the PyTorch framework through the transformers library, and the original word embedding layer and encoder structure are retained, laying the foundation for field adaptation. The field adaptation layer construction includes adding an engineering entity recognition unit, adding a problem classification layer and optimizing the attention mechanism. The engineering entity recognition unit includes: connecting the BiLSTM-CRF structure to the output layer of the base model, which is realized by TensorFlow tf.keras.layers or PyTorch nn.LSTM, and is specially used to identify engineering-specific entities such as "device model", "construction process", "specification standard" and the like in the text; the problem classification layer includes: using a fully connected layer + SoftMax activation function, designing 32-dimensional engineering problem classification labels (such as "structural safety", "material selection", "process conflict", etc.), and realizing feature mapping through PyTorch nn.Linear layer; the attention mechanism optimization includes: adding a field attention layer after the Transformer encoder, giving higher weights to key technical terms such as "steel strength", "concrete grade", etc., and improving the sensitivity of the model to engineering professional vocabulary.

[0102] Next, the special field AI model built based on the TensorFlow or PyTorch framework is subjected to dataset construction and training. First, technical contact sheets (containing problem description, solution) of large-scale engineering projects in the past 5 years, construction specification documents (such as GB 50204 Concrete Structure Specification), expert review opinions and other text data are collected, with a total amount of more than 100,000 pieces. The above data is used as raw data, and then the raw data is cleaned by removing random codes and redundant symbols through regular expressions, and the text is subjected to entity annotation (annotating entity type and boundary) and classification label annotation by using an artificial annotation tool (such as LabelStudio) to construct a structured training set. Then, the data is enhanced by using synonym replacement (such as replacing “steel” with “steel”), sentence reordering, context expansion and other methods to expand the dataset and solve the problem of data scarcity in the engineering field. The training process is optimized by using the fine-tuning strategy, multi-task training and training monitoring. The fine-tuning strategy uses the tf.keras.optimizers.Adam optimizer in TensorFlow with an initial learning rate of 2e-5 and uses a linear learning rate decay strategy. In PyTorch, torch.optim.AdamW is used to solve the instability problem of small batch training by gradient accumulation (Gradient Accumulation). The multi-task training simultaneously trains the “entity recognition” and “problem classification” two tasks, shares the basic model parameters, and realizes joint optimization through a weighted loss function (entity recognition loss weight 0.6, classification loss weight 0.4). The training monitoring uses TensorBoard to record the loss curve and accuracy index, sets an early stopping mechanism (EarlyStopping), stops training when the accuracy of the validation set does not improve for 5 consecutive epochs, and saves the optimal model weight.

[0103] Preferably, the trained model is quantized and compressed by TensorFlow Lite or PyTorch TorchScript in the application, reducing the model size by 60%, meeting the deployment requirements of edge devices, and encapsulating it into a RESTful API interface through a lightweight Web service framework, including: a problem analysis interface ( / analyze_problem): receiving user-submitted JSON format technical problem description, triggering AI analysis process; a solution matching interface ( / match_solution): outputting JSON structured data of historical solutions or newly generated solutions; the response time of the above interfaces is controlled within 500 milliseconds, guaranteeing the real-time processing requirements of engineering sites.

[0104] Preferably, the system is also configured with a verification mechanism including specification compliance checks and case library feedback iteration. The specification compliance checks are automatically compared with the national or industry standard database after the scheme is generated, marking the content in the scheme that conflicts with the specification and prompting modification suggestions. The case library feedback iteration is to collect the application effect data of the scheme in actual engineering (such as "adoption rate" and "problem solving rate"), and perform incremental training of the model every month to continuously optimize the accuracy of scheme generation.

[0105] Preferably, the system integrates a data collection module, an artificial intelligence large model processing module, and a contact single generation and approval module. After the data collection module receives technical contact single data, the artificial intelligence large model processing module automatically triggers the AI analysis process, and the analysis results are written into the MySQL database in real time. The problem classification results are used as the basis for approval level determination (such as "structural safety class" automatically triggering three-level approval), and intelligent auxiliary decision support is provided for approval personnel.

[0106] As a preferred embodiment of the present application, the implementation steps of the system are as follows:

[0107] System deployment and initialization: deploy the developed system to the server and operate according to the established deployment process. After deployment is completed, perform database initialization work and import historical construction technical problem processing data to provide initial data support for the system. At the same time, pre-train the artificial intelligence large model to have initial problem analysis and solution generation capabilities, ensuring that the system can operate normally and provide services.

[0108] User registration and permission setting: engineering users, construction builders, designers, and management personnel register as users. The system administrator sets the corresponding operation permissions according to different user roles to ensure data security and operation specification. Users with different roles can only access and operate functions and data that match their permissions, effectively preventing data leakage and illegal operations.

[0109] In the present application, when the engineering user party or the construction party encounters technical problems during the construction process, first log in to the system to enter the user interaction module, click to fill in the contact form button, and describe the problem in detail in the contact form filling page, including the time, place, specific phenomena, and impact on the project, etc., and upload relevant drawing files, photos, videos, etc. Information such as product model code, product specific equipment name, etc. After confirming that the filling content is correct, submit it, send the technical contact form to the system, collect it through the data collection module and transmit it to the artificial intelligence big model processing module for analysis and generate a solution, and store it in the data storage module. The contact form generation and approval module automatically generates a new construction technical contact form according to the final determined solution, attaches relevant drawing files, generates a unique number, and sends the contact form to relevant approval personnel according to the preset approval process. The approval personnel will review the contact form within the specified time limit, check the problem description and solution, and if the approval personnel agrees, execute the pass instruction; if the approval personnel disagrees, fill in the approval opinion and execute the rejection instruction, and the contact form will be returned to the design party for modification. The contact form approved by the approval module will be sent to the engineering construction unit within the specified time. The engineering construction unit user logs in to the system and views the received contact form on the designated page, obtains the solution, and performs construction operation according to the scheme. The user can query the processing progress, historical processing information and solution of the technical problem at any time through the query retrieval module. In the query retrieval page, input the query conditions (such as contact form number, problem classification, processing time, etc.) for query, and the system will quickly retrieve and display the relevant results. If the user is not satisfied with the processing result, he can fill in the feedback opinion through the feedback function of the user interaction module and submit it to the system. The system will further process according to the feedback information, such as reanalyzing the problem, adjusting the solution, etc., until the user is satisfied.

[0110] The above description is only some of the preferred embodiments of the present disclosure and an explanation of the principles of the technology used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combinations of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with each other to form a technical solution with similar functions disclosed in the embodiments of the present disclosure (but not limited to).

Claims

1. An artificial intelligence-based engineering construction technical problem-solving management system, characterized in that, include: The data acquisition module is used to collect technical liaison forms submitted by project users or construction parties, technical liaison forms replied by designers, and related historical processing information. The artificial intelligence large model processing module is used to receive data collected by the data acquisition module, analyze newly submitted technical problems, identify similar problems and extract historical solutions or assist in generating new solutions; The contact form generation and approval module is used to automatically generate numbered technical contact forms and associate them with drawing files based on the solutions output by the artificial intelligence big data model processing module, and to initiate the approval process according to preset approval rules. The data storage module is used to store data collected by the data acquisition module, solutions generated by the artificial intelligence large model processing module, and technical contact sheets and related drawing files generated by the contact sheet generation and approval module. The query and retrieval module is used to query relevant historical processing information and solutions in the data storage module based on various conditions. The user interaction module provides an interface for users to submit technical contact forms, check processing progress, view solutions, participate in approval processes, and provide feedback.

2. The engineering construction technical problem handling and management system based on artificial intelligence according to claim 1, characterized in that, The AI ​​large-scale model processing module determines whether a newly submitted technical question is of the same type by comprehensively comparing it with historical question data stored in the system. If it is a similar problem, relevant solutions are extracted from the historical solution library and optimized based on the current scenario; If the problem is not of the same type, then engineering domain knowledge is used to assist the designer in generating a solution.

3. The engineering construction technical problem handling and management system based on artificial intelligence according to claim 1, characterized in that, The data storage module uses a database management system to classify and store data and establish an indexing mechanism.

4. The engineering construction technical problem handling and management system based on artificial intelligence according to claim 1, characterized in that, The query and retrieval module allows users to search based on contact order number, problem category, product model code, and the person handling the issue, and also supports fuzzy search functionality.

5. The engineering construction technical problem handling and management system based on artificial intelligence according to claim 1, characterized in that, Each construction technical liaison form automatically generated by the liaison form generation and approval module is assigned a unique number.

6. The engineering construction technical problem handling and management system based on artificial intelligence according to claim 1, characterized in that, The user interaction module also has a message notification function, which is used to notify users of relevant operations and processing results.

7. A method for an engineering construction technical problem handling and management system based on artificial intelligence as described in any one of claims 1-6, characterized in that, Includes the following steps: Step S1: Collect technical liaison forms submitted by the project user or construction party, construction technical liaison forms replied by the design party, and relevant historical processing information through the data acquisition module; Step S2: Receive data collected by the data acquisition module through the artificial intelligence big data model processing module, analyze newly submitted technical issues, identify similar issues and extract historical solutions or assist in generating new solutions; Step S3: Based on the solution output by the AI ​​big data model processing module, the contact form generation and approval module automatically generates a numbered technical contact form and associates it with drawing files, and initiates the approval process according to the preset approval rules. Step S4: Store the data collected by the data acquisition module, the solution generated by the artificial intelligence big model processing module, and the technical contact forms and related drawing files generated by the contact form generation and approval module through the data storage module; Step S5: The user queries the relevant historical processing information and solutions in the data storage module through the query and retrieval module; Step S6: Users can submit technical contact forms, check processing progress, view solutions, participate in the approval process, and provide feedback through the user interaction module.

8. The method for the engineering construction technical problem handling and management system based on artificial intelligence according to claim 7, characterized in that, In step S2, when the large-scale artificial intelligence model processing module analyzes newly submitted technical issues, it first determines whether the issues are of the same type. If it is a similar problem, relevant solutions are extracted from the historical solution library and optimized based on the current scenario; If the problem is not of the same type, then engineering domain knowledge is used to assist the designer in generating a solution.

9. The method for the engineering construction technical problem handling and management system based on artificial intelligence according to claim 7, characterized in that, In step S3, when the approval process is initiated according to the preset approval rules, the approver must complete the approval operation within the specified time limit. The approved technical contact form will be sent to the engineering construction unit in a timely manner, and the unapproved technical contact form will be returned to the design party for modification.

10. The method for the engineering construction technical problem handling and management system based on artificial intelligence according to claim 7, characterized in that, In step S5, the user can query based on multiple conditions such as contact order number, problem category, product model code, and handling personnel. At the same time, the query and retrieval module supports fuzzy search function.

Images