A nuclear power plant handover and delivery whole-process collaborative management system and method

Abstract

The application provides a nuclear power plant handover and production transfer whole process collaborative management system and method, the system includes main data management, handover package plan, business process engine, knowledge graph and improved GNN intelligent reasoning, joint inspection collaboration, rule verification, data storage and unified workbench and other modules. The core uses five tuple knowledge graph, improved GNN and BERT / LSTM / SOM AI model, realizes dependent reasoning, completeness verification, text analysis, construction period prediction and defect clustering. The system abandons the traditional extensive mode, solves the pain points such as difficult artificial combing dependence, weak data isolation and loose closed-loop management through semantic modeling, graph intelligent reasoning and rule rigid execution. Realize multi-unit parallel management and control, full-factor structured embedding, digital signature storage, opinion item forced closed loop and report automatic generation, promote the upgrade of business from online to technology driven mode.

Description

Technical Field

[0001] This application belongs to the fields of information technology in nuclear power engineering, big data governance, knowledge graphs in the five-tuple domain, improved GNN graph neural networks and digital collaborative management technology, specifically involving a collaborative management system and method for the entire process of nuclear power plant handover and commissioning. Background Technology

[0002] The handover and commissioning of a nuclear power plant is a critical stage connecting engineering construction and production operation, involving various types such as system equipment isolation handover (Take Over for Blocking, TOB), maintenance handover (Take Over for Maintenance, TOM), temporary operation handover (Take Over for Temporary Operation, TOTO), building handover (Building HandOver, BHO), and reverse handover from commissioning to construction (such as construction to maintenance handover - TOM_CONS, construction to operation handover - TOTO_CONS). This process encompasses dozens of complex steps, including the Take Over Package Scope Definition (TPSD), Complete Package Scope Definition (CPSD), preparation and review of the End of Erection Status Report (EESR), handover application, multiple rounds of document review, joint on-site inspection, submission and closed-loop processing of Nonconformance Reports (NCRs), and signing and issuance of the handover certificate. Numerous parties are involved (such as engineering companies, construction units, commissioning units, production units, and supervisors), with each step closely linked and adhering to strict rules.

[0003] Traditional management models rely heavily on paper-based document circulation, offline meetings, manual telephone / email tracking, and Excel spreadsheet statistics, which have the following significant drawbacks: Unstructured data and difficulty in knowledge accumulation – massive amounts of handover documents, review comments, comment items, certificates, etc., exist in the form of unstructured documents or scattered tables, making it difficult to conduct effective queries, statistics, in-depth analysis, and the construction of project knowledge bases; Heavy workload and error-proneness and efficiency bottleneck – a large amount of repetitive manual filling, data transfer, report preparation, and status verification work consumes a lot of manpower and is prone to errors, becoming an efficiency bottleneck for project progress.

[0004] There is a lack of solidification and automated execution capabilities for core business rules such as the handover package coding system, strict pre- and post-transfer logical relationships between handover stages (e.g., TOB must be initiated after the corresponding EESR is completed), and the management of comment item classification (Category I / II) and mandatory closed-loop verification. Comment item control is lax, and Category I items lack a mandatory closed-loop interception mechanism, making them prone to entering the signature process in violation of regulations.

[0005] The lack of a nuclear power-specific knowledge graph makes it impossible to abstract units, systems, handover phases, and process constraints into reasonable entities and relationships, and dependencies cannot be converted into digital knowledge.

[0006] Insufficient compatibility with multiple power plants and units: It only supports single power plant and single unit management, and cannot meet the needs of multiple projects in parallel, data isolation and unified aggregation. Outdated certificate management: Transfer certificates rely on manual compilation and offline signatures, resulting in inconsistent formats, errors, low efficiency, easy tampering, poor traceability, and failure to achieve full-element structured automatic filling of transfer information; The multi-tenant logical isolation architecture without cluster factories and clusters results in data being scattered across multiple projects and unable to be uniformly aggregated. Passive progress control: There are no intelligent reminders for planned overdue projects, and the reliance on manual tracking can easily lead to delays in handover. Inefficient reporting: Weekly reports and PPT presentations rely on manual data compilation, which is time-consuming, error-prone, and cannot automatically generate standardized reporting materials; Weak compliance control: Key node verification relies on manual execution, which can easily lead to quality and safety risks such as skipping steps in the process and signing off on comments before they are closed.

[0007] Existing general-purpose OA and project management systems can only realize form circulation, which is merely the online version of ordinary business and cannot meet the control requirements of strong rules, strong constraints, reasoning, and traceability in nuclear power plant handover. Therefore, there is an urgent need for a collaborative management system and method for handover and production acceptance that is adaptable to multiple plants and reactors, with a structured process, digital certificates, intelligent progress tracking, and automated reporting. Summary of the Invention

[0008] The purpose of this application is to overcome the shortcomings of existing technologies, such as reliance on human experience, hard-coded rules, lack of semantic modeling, lack of automatic dependency reasoning, and poor adaptability to general OA systems. It creates a unique five-tuple knowledge graph modeling architecture of unit-system-transfer phase-process entity-dependency relationship, which transforms the implicit business dependencies of nuclear power transfer into a digital semantic network that can be parsed by machines and can be reasoned in multiple hops. With the improvement of GNN strong dependency reinforcement mechanism and multi-hop path pruning algorithm, it builds the underlying intelligent technology foundation, eliminates the need for manual rule configuration, and realizes fully automatic intelligent verification of pre-constraints.

[0009] To achieve the above objectives, this application proposes a collaborative management system for the entire process of nuclear power plant handover and commissioning, comprising: The master data and access control module is used to configure and maintain the baseline data for system operation and a unified access control mechanism. The handover package plan management module is used to create, maintain, and manage the entire lifecycle status of handover package plans. The business process engine module has a built-in standardized process model library covering the entire process of nuclear power plant handover and production. It is used to create process instances, push task nodes to designated personnel and intelligently control the flow path, and record the history of process status changes. It also supports process suspension, rollback, cancellation and version upgrade operations. The core intelligent module comprises a quintuple knowledge graph and an improved GNN intelligent reasoning module. This module constructs the quintuple knowledge graph and assigns differentiated weights to different dependencies. It also improves GNN message passing by strengthening strong pre-dependencies through weighted aggregation, gating truncation, and hierarchical aggregation. After multi-hop path pruning, it outputs completeness, risk level, and a missing list, providing intelligent decision-making basis for process initiation and node release. Decision nodes, deeply integrated into the business process engine module, run as pre-services or automatic verifiers to enforce business compliance logic. The module also includes lightweight BERT, LSTM, and SOM AI models, respectively implementing semantic parsing and classification of handover text, intelligent prediction of project timelines, autonomous clustering of defect root causes, intelligent comparison and verification of handover data, and dynamic optimization scheduling of multi-tasks. The joint inspection and feedback item collaborative management module is used to organize joint on-site inspections and to track and manage the issues found during the inspections throughout their entire lifecycle. The intelligent rule verification and execution module is used to support the validity verification of handover packages, the completion verification of the knowledge graph reasoning pre-process, the closed-loop verification of opinion items, the closed-loop verification of document review, and the rules for multi-person countersigning.

[0010] The structured data storage and output module is used to define and persistently store the core business information generated in the business process in a structured manner. The unified workbench and decision support module provides a personalized portal interface for users with different roles within the system, centrally displaying pending tasks, processed transactions, and processes of interest. It also provides multi-dimensional statistical analysis and visualization functions based on the structured data accumulated throughout the entire process, generating various management dashboards and reports on handover progress, status of comments, and review efficiency.

[0011] As an improvement to the above system, the functions of the quintuple knowledge graph and the improved GNN intelligent reasoning module include: Construct a domain knowledge graph consisting of five elements: unit, system, handover phase, process entity, and dependency. Define four types of entities: unit, KKS system, handover phase, and process instance, as well as five types of directed weighted edges: strong precondition, weak precondition, system dependency, professional coupling, and parallel constraint. The message passing mechanism of GNN is improved by assigning high weights to strong pre-dependencies, using weighted feature aggregation, setting constraint gating thresholds to truncate weakly related edges, and performing hierarchical neighborhood aggregation by unit-system-professional to enhance the dominance of reasoning based on strong pre-dependencies. Supports a three-layer multi-hop path inference pruning method that includes professional irrelevant filtering, cross-unit invalid node removal, and weak weighted relational path truncation, automatically reducing irrelevant nodes and redundant paths; Automatically calculate the completeness score from 0 to 100, classify compliance risk levels as high / medium / low based on the score, and output a detailed list of unclosed and missing dependencies.

[0012] As an improvement to the above system, the weighted feature aggregation is represented as follows: h v =∑W uv × h u ; Among them, h v W is the feature vector of the target node, representing the new feature representation after aggregation of the nodes currently being updated; uv h is the edge weight / dependency strength coefficient, representing the weight value of the edge connecting neighbor node u to target node v; u is the feature vector of a neighboring node, representing the feature representation of a certain neighboring node of the current node v.

[0013] This application also provides a collaborative management method for the entire process of nuclear power plant handover and commissioning, implemented based on the above system, including: Step 1: System initialization and configuration; Configure basic data for power plants, generating units, and systems; Define various user roles and permissions; Configure various standard handover process models and their corresponding intelligent verification rules; Construct a standardized five-tuple domain knowledge graph for generating units, systems, handover phases, process entities, and dependencies; Step 2: Handover Plan Creation and Process Initiation Validation; Create a handover package plan based on the project plan, compile a structured handover package code that conforms to the rules, and verify its status as "valid"; Select the handover type to be executed, associate it with a "valid" handover package plan, and initiate a new handover process; Load the handover package information and trigger "handover package validity verification" and "pre-process closure verification"; If the verification passes, the process instance is successfully created and enters the compilation state; Step 3: Online approval and document review; the process flows sequentially through the drafting, internal review and submission stages, reaching the document review node; all assigned reviewers independently review the documents online and fill in structured review comments, and the results are aggregated after all reviewers have completed their reviews; if the review generates comments, it will be transferred to the comments sub-process for processing, and if the "document review closed-loop verification" is met, it will proceed to the next stage; Step 4: Joint Inspection Organization and Comment Processing; After the document review is passed, the process enters the joint inspection stage. The system automatically or the person in charge triggers the "Issue Joint Inspection Notice" to organize a joint inspection online; After the on-site inspection, comments are entered; After the comments are created, the system automatically assigns them to the responsible person for processing, feedback, verification and closure, forming a closed loop; The process engine continuously monitors the closure status of "Category I Items"; Step 5: Certificate Preparation, Signing, Issuance, and Process Traceability Upgrade; When all "Category I Items" are closed, the process progresses to the certificate signing stage; the system automatically marks the unclosed "Category II Items" as legacy items and fills them into the corresponding positions in the certificate template; the certificate goes through compilation, multi-level online review and approval in sequence, and is finally issued by the authorized user; at the same time as issuance, the system automatically generates the certificate file and updates the process status to "closed", completing this handover; for closed processes, the system retains the full-process structured operation records, document versions, and comment history.

[0014] As an improvement to the above method, step 1 includes: Step 1-1: Configure basic master data; enter the basic information of the nuclear power plant and units, import the unified system and KKS system for the entire plant, and build a complete organizational tree; Steps 1-2: Configure the 5-tuple knowledge graph entities and relationships. First, configure the entities, creating unit entities, KKS system entities, handover phase entities, and process instance entities. Unit entities are used for top-level data isolation, KKS system entities are based on standard KKS coding, handover phase entities solidify the standard handover phases, and process instance entities are dynamically created according to the actual handover process. Next, configure the relationships, constructing five types of directed weighted relationships: strong pre-dependency, weak pre-dependency, system dependency, professional coupling, and parallel constraint. All relationships are represented by directed weighted arrows, with the arrow direction clearly indicating the business sequence logic of the predecessor and successor. Weights are assigned according to the business enforcement level: strong pre-dependency is configured with high weight, weak pre-dependency, system dependency, and professional coupling are configured with medium weight, and parallel constraint is configured with low weight. Entities and relationships are stored uniformly in the graph database; static entities are configured once, and dynamic instances are automatically updated according to business needs. Steps 1-3: Initialize the improved GNN network and multi-hop pruning parameters; When initializing the improved GNN network, first load the four types of entities and five types of directed weighted relations of the quintuple knowledge graph to construct a weighted adjacency matrix; configure high transitive weights for strong pre-dependencies, set constraint gating thresholds to block weak associations from participating in aggregation, and implement progressive strengthening of neighborhood features according to unit-system-professional hierarchy to complete the network structure and parameter initialization; Multi-hop pruning parameter initialization includes setting the maximum number of inference hops, configuring three layers of rules: filtering professionally irrelevant nodes, removing invalid nodes across units, and truncating weak weighted relation paths; Steps 1-4: Load BERT / LSTM / SOM models; complete the initial loading and parameter fixing of lightweight BERT, LSTM, and SOM AI models; Steps 1-5: Define the role and permission system; After completing the organization and user configuration, define a refined role and permission model.

[0015] As an improvement to the above method, step 2 includes: Step 2-1: Develop a handover package plan and generate a four-dimensional code for the unit, system, handover phase, and subsystem; Step 2-2: The four-dimensional code is semantically decomposed and structured feature parsed using the encoding parsing engine to extract four core inherent features: unit affiliation, system professional classification, standardized handover stage, and business application scenario. The original encoded features obtained from the decomposition are standardized and normalized to unify the dimension specifications, weight benchmarks, and field storage formats of all features. Redundant, invalid, and interfering field data are cleaned and removed to generate a standardized structured feature dataset that is adapted to the storage specifications of each entity node in the five-tuple knowledge graph. Steps 2-3: Improved GNN to initiate multi-hop path reasoning; The system relies on the improved GNN graph neural network to initiate intelligent multi-hop path reasoning. Combining the hierarchical constraints and strong and weak dependencies of nuclear power plant handover, it introduces hierarchical path pruning, weighted gating filtering and multi-hop progressive aggregation mechanism to achieve in-depth mining and deduction of implicit pre-dependencies and cross-professional coupling relationships among units, KKS systems, handover stages and process instances. Steps 2-4: Handover process initiation and pre-process intelligent verification; After the handover process is initiated, the system obtains the full set of pre-process constraints of the target handover package through multi-hop path reasoning of the five-tuple knowledge graph, assigns differentiated weight parameters to each constraint based on the improved GNN model, calculates the pre-process completeness quantitative score in the range of 0 to 100 through the weighted normalization algorithm, and divides the compliance risk levels into high, medium and low levels by combining the closed-loop state of strong pre-process constraints and the score range, and outputs a structured list of missing dependencies simultaneously; Steps 2-5: Create or intercept a process instance; if all pre-process smart checks pass successfully, create a new process instance and initialize its status to "In Development"; if any check fails, issue a reminder and prevent the creation of the current process instance.

[0016] As an improvement to the above method, step 3 includes: Step 3-1: Initiate document review task assignment; When the process completes internal review and enters the document review stage, the system sends a pending task notification to the reviewer and assigns the review task to the corresponding reviewer. Step 3-2: Implement a multi-person review mechanism; reviewers review the documents online and fill in their comments in the structured review comment form provided by the system; if all reviewers approve the document, the process continues. Steps 3-4: Generate and track review comments; During the countersigning process, for each specific comment filled in by the reviewer, the system will automatically create a corresponding "Document Review Comment Item" record. This record is initially classified as "Category 1 Item" and associated with the current process instance, the submitter, and the submission time information; These comments then enter an independent comment item processing sub-process. Steps 3-5: Handling and Closing-Loop Verification of Comments; For each comment, the person handling it submits the handling measures or explanations online, and supports uploading supporting documents; After submission, the person who submitted the comment will automatically receive a verification task and confirm online whether the processing result meets the requirements; If the verification is successful, the status of the comment will be updated to "Closed"; The system continuously tracks the status of all related comments. Steps 3-6: Document review closed-loop verification drives the process forward; when all comments generated by this document review have reached the "closed" state, it marks the completion of the document review stage; at this time, the system's built-in "document review closed-loop verification" rule is satisfied. The satisfaction of this rule is the basis for the process engine to decide whether to allow the current process to exit the "document review" stage and move to the next stage.

[0017] As an improvement to the above method, step 4 includes: Step 4-1: On-site joint inspection and entry of comments; After the process enters the on-site joint inspection stage, the joint inspection organizer issues a joint inspection notice through the system and organizes the on-site inspection; After the inspection, the person in charge of the joint inspection centrally enters the problems found on-site into the system to form "joint inspection comments"; Each comment includes a description, responsible unit and required removal deadline, and is initially classified; The classification includes "Category I items" and "Category II items"; Step 4-2: Assigning comments and tracking the processing; The system automatically assigns the entered comments to the corresponding responsible units or individuals; The responsible persons receive the tasks online and provide feedback on the processing plan and implementation status; For complex issues, the system supports the responsible persons to initiate a "classification application" to request that "Category 1 items" that cannot be completely closed in the short term be converted to "Category 2 items"; Step 4-3: Online approval of the classification change application; After the classification change application is initiated, the system will process it according to the predefined approval path, and finally confirm it by the original proposer of the opinion item; After the approval is approved, the system will update the current classification of the opinion item to "Category II item", but the initial classification information will still be retained in its history for traceability; Step 4-4: Mandatory Closed-Loop Verification of the Signature Process; When all comments have been processed and the process is ready to enter the certificate signing stage, the person in charge of the transfer triggers the "Prepare for Signature" operation, and the system forcibly executes "Mandatory Closed-Loop Verification of Comments"; The "Mandatory Closed-Loop Verification of Comments" is as follows: Scan all comments associated with the current process, filter out records initially classified as "Category 1 Items", and check their status. Only when the status of all comments initially classified as "Category 1 Items" is "Closed" can the verification pass; If any such item is not closed, the system will pop up a prompt and intercept the process, prohibiting it from entering the signing state; Steps 4-5: Legacy item information is automatically imported into the certificate; for "Category II items" that are still in the "not closed" state after verification, the system automatically extracts the information of these items and fills them into the "Legacy Item List" section of the certificate template when generating the handover certificate, ensuring complete information transfer.

[0018] As an improvement to the above method, step 5 includes: Step 5-1: Generate and publish the handover certificate; After the handover certificate has completed all online approval and digital signature processes, the authorized user performs the final "publish" operation; The structured data storage and output module extracts all the structured data finally determined in this process from the database, including: basic process information, approved equipment boundary list, signatures of all parties, signature timestamps, and list of legacy items; Step 5-2: Generate templated files; The system will automatically generate certificates from the extracted structured data according to the templates of each factory, fill in the core list and legacy items, complete multi-level digital signing, and bind the signature and timestamp; The document processing engine will fill the data into the corresponding field positions of the template, dynamically generate a PDF format official certificate file that can be directly used for archiving, and provide a download link; Step 5-3: Full-process data aggregation and statistical analysis; The unified workbench and decision support module run data aggregation tasks, and based on the structured data accumulated throughout the process, build a multi-dimensional data analysis model to support the generation and display of various statistical data.

[0019] Compared with existing technologies, the advantages of this application are: 1. Fully structured: Enables end-to-end online and integrated collaboration—integrating previously scattered handover activities via paper, email, and meetings into a unified digital platform, achieving real-time online collaboration and information synchronization across organizations and regions, completely breaking down information silos and significantly improving communication and collaboration efficiency; at the same time, the handover scope, boundaries / keys / drawings list, and legacy items are all online, automatically imported into certificates, and zero human error; 2. Intelligent Handover Management: This invention features a unique five-tuple domain knowledge graph modeling approach, which combines unit-system-handover phase-process entity-dependency relationship. This approach decomposes the implicit and unstructured pre- and post-handover constraints, system subordination, and professionally coupled business dependencies that originally existed only in institutional norms and engineers' experience into standardized entity nodes and directed weighted semantic relationships. This constructs a digital semantic network that can be parsed, associated, and subjected to multi-hop reasoning. It completely abandons the traditional method of manually hard-coding rules one by one, and relies on the inherent business semantics of the graph to automatically associate and deduce pre-dependencies, thereby making handover constraints explicit, knowledge structured, and reasoning intelligent. 3. Enables structured, storable, and reusable business knowledge, allowing new units and systems to automatically correlate and deduce without reconfiguring the front-end logic; 4. Improved GNN: Strengthens strong pre-constraints at the network mechanism level, and the inference results are in line with the actual management rules of nuclear power plants; realizes fully automated intelligent inference before and after handover, without the need for manual hard coding configuration, which greatly reduces maintenance workload and reduces the risk of omissions; 5. The underlying technological innovation, which upgrades the traditional process to online processing, combines semantic modeling, intelligent reasoning, and rigid rule execution, has prominent substantive features and significant progress. 6. Refined Management: Enhanced data governance and decision support capabilities – Structured storage of end-to-end business data (plans, processes, feedback items, certificates) makes data querying, statistics, and analysis efficient and accurate. Based on this, various real-time reports and dashboards provide project managers with intuitive and quantifiable decision-making support, facilitating refined management. 7. Reporting Automation: Significantly improves work efficiency and quality – through functions such as automatic data import, template generation, and one-click generation of report PPTs, it reduces a lot of repetitive manual work and document processing time; 8. Multi-plant and multi-unit adaptation: Supports parallel handover management of construction projects for multiple power plants and multiple units, data isolation and unified aggregation, meeting the needs of large-scale nuclear power construction. Attached Figure Description

[0020] Figure 1 The diagram shows the architecture of the collaborative management system for the entire process of nuclear power plant handover and commissioning. Figure 2 The diagram shows the overall flowchart of the collaborative management method for the entire process of nuclear power plant handover and commissioning. Figure 3 The diagram shown is a flowchart of system initialization and basic data configuration; Figure 4 The diagram shown is a flowchart of the handover package plan creation and process initiation verification. Figure 5 The diagram shows the flowchart for document review collaboration and comment processing; Figure 6 The diagram shows the flowchart for processing on-site joint inspection comments and mandatory control measures before signing. Figure 7 The diagram shows the flowchart for structured data output and multidimensional decision analysis. Detailed Implementation

[0021] The technical solution of this application will be described in detail below with reference to the accompanying drawings.

[0022] This invention's system adopts a microservice architecture and a front-end / back-end separation design, using a hybrid storage of relational database and graph database, and is deployed on a server cluster. During the system initialization phase, master data configuration for power plants, generating units, KKS systems, and organizational roles is completed; entity and relation input for the five-tuple knowledge graph is completed; and the parameters of the improved GNN network are initialized and configured.

[0023] This invention's five-tuple knowledge graph refers to a knowledge graph specific to the nuclear power plant handover and production transfer domain, constructed with five core components: generating unit, KKS (Knowledge, Skills, and Systems) system, handover phase, process entity, and dependencies. Specifically, it defines four types of nodes: generating unit, KKS system, handover phase, and process instance; and five types of directed weighted edges: strong prerequisite, weak prerequisite, system dependency, professional coupling, and parallel constraint. Different weights are configured according to business importance, constructing a three-layer graph architecture: a basic entity layer, a process instance layer, and a logical constraint layer.

[0024] The improved GNN strong dependency reinforcement implementation process includes: using a directed weighted adjacency matrix, configuring high weights (0.9~1.0) for strong pre-dependencies EESR→TOB and TOB→TOTO, medium weights for ordinary associations, and low weights for weak couplings; using weighted feature aggregation in message passing, significantly increasing the proportion of strong dependency features; setting a 0.5 weight constraint gate, directly truncating weak association edges below the threshold and not participating in aggregation; updating features layer by layer according to the same group, system, and profession, gradually amplifying the priority of strong dependency propagation, so that rigid pre-constraints always dominate in the inference results.

[0025] The specific implementation process of multi-hop path pruning includes: starting from the target handover process, carrying out multi-hop traversal, sequentially performing professional irrelevant node filtering, non-target unit entity elimination, low-weight weak path truncation, and automatically pruning invalid inference nodes, thereby significantly reducing the computational load while ensuring the integrity of inference.

[0026] When a process is initiated, the system automatically invokes a knowledge graph and an improved GNN to perform multi-hop reasoning, calculates the completeness score of prerequisites, assesses compliance risks, and outputs a list of missing dependencies. If the verification fails, the process is directly blocked. After document review and joint inspection are completed, the system enforces a closed-loop verification of all comments in a certain category. If the loop is not closed, the process is prohibited from proceeding to the signing stage. Certificates are automatically populated with various lists and legacy items according to the power plant template, bound with digital signatures and timestamps, and automatically generated archiveable and effective certificates. The system automatically calculates progress on a periodic basis, issues warnings for overdue tasks, and generates standardized reporting PPTs with one click, achieving intelligent closed-loop management of the entire process.

[0027] Core principles of quintuple knowledge graph modeling: This invention features a unique five-tuple domain knowledge graph, constructing a standardized five-tuple model of "unit-system-transfer phase-process entity-dependency relationship".

[0028] The five-tuple model contains four core entities: unit entity, KKS system entity, handover phase entity, and process instance entity; The quintuple model defines five types of directed weighted dependencies: strong prerequisite dependency, weak prerequisite dependency, system dependency, professional coupling, and parallel constraint.

[0029] The core principle of the quintuple model is to transform the originally scattered, implicit business constraints that rely solely on human memory into standardized entity nodes and computable directed semantic edges, thereby making implicit dependencies explicit, structuring experiential knowledge, and enabling business rules to be reasoned by machines.

[0030] Traditionally, rules are memorized and configured manually. This invention, on the other hand, uses machines to store rules and push dependencies, eliminating the need for manual writing of pre-login. New systems and units are automatically associated and deduced, thus fundamentally solving the technical pain points of tedious and easily overlooked manual configuration.

[0031] The graph is divided into a three-layer architecture: the basic entity layer, the business process instance layer, and the logical constraint layer, which realizes the deep integration of static standard ontology and dynamic process instance.

[0032] Improved GNN strong dependency reinforcement principle: This application abandons the general GNN equal weight transfer mode and improves it through four aspects: differentiated edge weight assignment, weighted feature aggregation, constraint gating truncation, and unit-system-professional hierarchical neighborhood aggregation. This ensures that strong pre-dependencies such as EESR→TOB always occupy the dominant weight in graph reasoning, while weakly correlated and irrelevant nodes are automatically suppressed, accurately adapting to the rigid closed-loop rules of nuclear power.

[0033] 1. Introduce a directed weighted adjacency matrix and assign values ​​separately to strong dependencies; In the quintuple knowledge graph, weights are pre-fixed for each type of dependency: Strong pre-dependencies (EESR→TOB, TOB→TOTO) weight = 0.9~1.0 Weak pre-dependency weight = 0.6~0.8 System subordination and professional coupling weights = 0.3~0.5 Weak association (parallel constraint) weight = 0.1~0.2 In a typical GNN, the adjacency matrix only has 0 / 1, while this invention uses continuous weight values ​​to differentiate between strong and weak dependencies from the bottom layer.

[0034] 2. When passing messages, aggregate them by weight; strong dependencies contribute more. Standard GNN aggregation formula: h v =∑ h neighbor This invention is an improved weighted aggregation: h v =∑W uv × h u Among them, h v W is the feature vector of the target node, representing the new feature representation after aggregation of the node currently being updated (i.e., the "center node" or "target node"); uv h is the edge weight / dependency strength coefficient, representing the weight value of the edge connecting neighbor node u to target node v; u is the feature vector of the neighboring nodes, representing the feature representation of a certain neighboring node (i.e., node u connected to v) of the current node v.

[0035] A new state h of a node v It is determined by the states h of all its neighboring nodes. u According to the closeness of their relationship W uv The weighted sum is obtained. Strongly dependent edge W. uv Larger and significantly higher in feature aggregation, weakly related edges are automatically reduced in contribution, directly strengthening the discourse power of strong dependencies from the message weighting level.

[0036] 3. Add constraint gating thresholds to directly block weak dependencies; Set a fixed threshold (e.g., 0.5): Edge weight ≥ 0.5: allowed to participate in message passing; Edge weight < 0.5: gated truncation, not participating in aggregation.

[0037] Constraint gating thresholds can allow only strong and medium dependencies to enter inference, while weakly related and irrelevant systems are directly shielded, thus preventing weak dependencies from interfering with strong-constraint inference from a mechanism perspective.

[0038] Example 1 like Figure 1 As shown, the collaborative management system for the entire process of nuclear power plant handover and commissioning provided in this application includes the following core functional modules: The Master Data and Access Control module is used to configure and maintain the baseline data system and unified access control mechanism for system operation. Baseline data includes nuclear power plant and unit identifiers, system and equipment codes, plant codes, information on all participating units, organizational structure, and user and role definitions. This module establishes a refined role-based access control model to achieve domain-specific and differentiated access control for different users on function menus, business process instances, and specific business data, ensuring the security of business processes and data isolation.

[0039] The handover package planning management module, in conjunction with the master data and access control module, is used to create, maintain, and manage the full lifecycle status of handover package plans. Its core lies in defining and managing unique and structured handover package coding rules, which typically employ a multi-segment structure of "unit-system / plant-handover direction-handover stage-subsystem". Handover package plans have clearly defined states such as "in preparation", "valid", and "invalid". Furthermore, the initiation of all subsequent specific handover processes must and can only be associated with a handover package plan in the "valid" state, thereby ensuring strong binding force of the plan on execution.

[0040] The business process engine module, as the central driving module of the system, has a built-in standardized process model library covering all scenarios of nuclear power plant handover and commissioning. These models include, but are not limited to: the scope definition process for commissioning and production handover, the preparation and review process for installation completion status reports, various handover processes such as isolation handover, maintenance handover and temporary operation handover, plant handover processes, and reverse handover processes. This module is responsible for the automated creation of process instances, the precise push of task nodes to designated personnel, the intelligent control of the flow path, and the complete recording of the process status change history. It also supports the suspension, rollback, cancellation, and version upgrade operations of processes.

[0041] The core intelligent module is the quintuple knowledge graph and the improved GNN intelligent reasoning module. It is used to construct the quintuple knowledge graph and set differentiated weights for different dependencies. The improved GNN message passing strengthens strong pre-dependencies through weighted aggregation, gating truncation, and hierarchical aggregation. After multi-hop path pruning, it intelligently outputs completeness, risk level, and missing list, providing intelligent decision-making basis for process initiation and node release. It is deeply integrated into the key decision nodes of the business process engine module and runs in the form of pre-service or automatic verifier to enforce business compliance logic and ensure the rigid implementation of management requirements. This module also includes three types of AI models: lightweight BERT, LSTM, and SOM. Through multi-model division of labor and collaborative reasoning, it can respectively achieve semantic parsing and classification of handover text, intelligent prediction of project timeline, autonomous clustering of defect root causes, intelligent comparison and verification of handover data, and dynamic optimization and scheduling of multi-tasks. Combined with the five-tuple knowledge graph and improved GNN intelligent reasoning capabilities, it can build a multi-dimensional, full-scenario nuclear power handover and production AI intelligent management and control system, realizing the upgrade from traditional manual experience management to data-driven and model-empowered intelligent management.

[0042] The Joint Inspection and Comment Item Collaboration Management module is used to efficiently organize joint on-site inspections and track and manage the issues discovered during the inspections throughout their entire lifecycle. This module includes a Joint Inspection Notification Management submodule and a Comment Item Lifecycle Management submodule. The Joint Inspection Notification Management submodule can automatically or manually create and publish joint inspection notifications based on the process status. The Comment Item Lifecycle Management submodule provides convenient functions for creating, assigning, processing, providing feedback, verifying, and closing comments, ensuring that every issue is recorded, tracked, and forms a closed loop. Specifically, when the associated handover process enters the certificate preparation stage, the system will automatically mark comments that have not yet been closed and are classified as "Category II Items" as "Legacy Items" and automatically associate them with the final generated handover certificate.

[0043] The intelligent rule verification and execution module supports the verification of handover package validity, the completion verification of the knowledge graph inference pre-process, the closed-loop verification of opinion items, the closed-loop verification of document review, and the multi-person countersigning rules. This module abandons the traditional manual verification and fixed rule hard-coding mode, using improved GNN multi-hop inference as its core and supported by a quantified topology of a five-tuple knowledge graph to construct a quantifiable and adaptive intelligent verification mechanism for handover and production. Relying on the improved GNN to uncover implicit business constraints and dynamically output compliance indicators, it achieves full-process intelligent verification of handover package validity, pre-process compliance, opinion and document closed-loop, and multi-person countersigning, solving the shortcomings of traditional static verification in adapting to dynamic implicit constraints. The module consists of a rule configuration layer, a data access layer, a hierarchical verification engine, and a process execution layer, using GNN inference results as the core verification basis throughout the process. The specific implementation is as follows: The rule configuration layer only sets basic access thresholds and control levels; all complex implicit pre-constraints are dynamically generated by the improved GNN multi-hop inference, enabling flexible rule iteration. The data access layer aggregates GNN inference results, weighted graph topology, and dynamic node features, combined with lightweight BERT and twin network-assisted data, providing quantitative input for intelligent verification. The hierarchical verification engine uses GNN inference results as the highest judgment criterion, verifying and intercepting unqualified conditions at each level. The process execution layer implements differentiated control based on GNN risk levels, achieving fully automated closed-loop control of process locking, rectification push, and permission start / stop.

[0044] Intelligent verification of handover package validity: Relying on the improved GNN topology reasoning capabilities and combining the entity relationships of units, systems, and stages in the five-tuple graph, it not only verifies the compliance of handover package coding and the validity of filing, but also intelligently verifies the rationality of topology attribution and identifies hidden anomalies such as cross-unit and cross-professional mismatches. Failure to pass verification prevents filing and entry into the database; successful verification generates process instance nodes and writes them into the graph, providing valid topology data for GNN deep reasoning.

[0045] Intelligent Compliance Verification of Pre-process: This step relies entirely on improved GNN multi-hop weighted reasoning to achieve dynamic intelligent verification, overcoming the limitations of traditional static verification with fixed lists. The improved GNN, based on a directed weighted topology of the graph, automatically mines explicit pre-process constraints and implicit coupling constraints through multi-hop traversal, path pruning, and weighted gating aggregation, dynamically generating sets of strong and weak constraints, completeness scores, and risk levels. The system establishes a three-level admission mechanism: high-weight strong pre-process constraints that are not closed-loop or have a score <60 are forcibly blocked; strong pre-process constraints that are fully closed-loop, weak pre-process constraints with minor deficiencies and a score of 60-85 are given a warning and allowed to proceed; a score >85 and fully constrained closed-loop results in automatic compliance approval. The system updates graph features according to operating conditions, driving iterative GNN reasoning to achieve dynamic adaptive verification.

[0046] Intelligent closed-loop verification of feedback items: Based on GNN compliance status, and combined with lightweight BERT semantic classification, a tiered verification system is implemented. The system automatically distinguishes between Category I mandatory defects and Category II general issues, determines the risk level based on GNN constraints, rigidly intercepts Category I defects that are not closed-loop, and locks the signing and certification process; Category II issues only trigger a ledger warning. The closed-loop status updates the graph nodes in real time, and reverse-optimizes the accuracy of subsequent GNN inference.

[0047] Document review closed-loop intelligent verification: Based on the GNN's pre-topology compliance verification, intelligent document comparison is completed through a twin network to identify issues such as missing information, parameter deviations, and specification conflicts. The deviation level is differentiated by the GNN risk rating, with fatal deviations being forcibly blocked and general deviations subject to time-limited closed-loop verification. After rectification and compliance, the graph features are updated to assist the GNN in making the final compliance determination.

[0048] Intelligent verification of multi-signature rules: Relying on the professional coupling topology of the graph to verify the qualifications of signatories, and linking and improving the GNN's pre-compliance results to achieve joint control of permissions. The system only grants signing permissions when the GNN determines that the pre-process, opinion items, and document review are all compliant; at the same time, it verifies the professional matching degree of the signatories and the number of valid signatories, filters invalid signatures, and realizes intelligent linkage control of the signing process.

[0049] The core of this module is an intelligent verification architecture based on improved GNN multi-hop inference and quintuple graph quantification constraints, which overturns the traditional manual and static rule verification mode. By dynamically mining implicit constraints, quantifying compliance levels, and adaptively adapting to handover scenarios through GNN, it achieves intelligent, dynamic, and quantifiable control over handover and production verification. This solves the shortcomings of traditional verification rules being rigid, implicit risks being missed, and lacking quantitative basis, effectively improving the accuracy and intelligence level of nuclear power handover and production compliance control.

[0050] The structured data storage and output module is responsible for defining and persistently storing the core business information generated in the business process. Based on this structured data foundation, this module, through a template engine, can automatically populate data according to predefined document template formats and generate standardized handover application documents and formal handover certificates that can be directly used for printing and archiving.

[0051] The unified workbench and decision support module provides personalized portal interfaces for users with different roles within the system, centrally displaying pending tasks, processed transactions, and processes of interest. Furthermore, based on structured data accumulated throughout the entire process, this module offers powerful multi-dimensional statistical analysis and visualization capabilities, generating various management dashboards and reports on handover progress, comment status, review efficiency, etc., providing intuitive and quantitative data support for project managers' decision-making.

[0052] Example 2 like Figure 2 As shown, this application also provides a collaborative management method for the entire process of nuclear power plant handover and commissioning, which is implemented based on the above system and includes the following steps: S1: System Initialization and Configuration. Administrators configure basic data such as power plant, generating units, and systems in the master data and permission management module, defining various user roles and permissions. In the business process engine module, various standard handover process models (such as EESR processes, TOB processes, etc.) and their corresponding intelligent verification rules are pre-configured and configured. A standardized five-tuple domain knowledge graph is constructed using a five-tuple knowledge graph and an improved GNN intelligent reasoning module, representing the generating unit-system-handover phase-process entity-dependency relationship. Unit entity: Stores unit number, model, base location, and construction period attributes; System Entities: Each entity is uniquely identified by a KKS code, which records system hierarchy, professional classification, and upstream and downstream physical relationships. Handover Phase Entities: Define the standardized phase nodes for EESR, TOB, TOM, TOTO, and BHO; Process Entity: Corresponds to each handover process instance, storing process status, approval progress, and closure rate of comments; Dependency relationships: Define five types of directed weighted edges: strong prerequisite dependency, weak prerequisite dependency, system dependency, professional coupling, and parallel constraint. Configure the constraint strength, effective conditions, and enforcement level for each relationship.

[0053] The graph is divided into a three-layer architecture: the basic entity layer, the business process instance layer, and the logical constraint layer, which realizes the integration of static ontology and dynamic process.

[0054] S2: Handover Plan Creation and Process Initiation Verification. In the handover package plan management module, the planning engineer creates a handover package plan based on the project plan, compiles a structured handover package code that conforms to the rules, and verifies its status as "valid." Subsequently, users (such as commissioning engineers and contractor engineers) select the handover type to be executed (such as TOB) on the workbench, associate it with a "valid" handover package plan, and initiate a new handover process. The system automatically loads the handover package information and immediately triggers "handover package validity verification" and "pre-process closure verification" (such as verifying whether the corresponding EESR for TOB is completed). If the verification passes, the process instance is successfully created and enters the compilation state.

[0055] S3: Online Process Approval and Document Review. The process flows sequentially through stages such as drafting, internal review, and submission, reaching the document review node. Document review is typically divided into preliminary review and final review: the preliminary review is assigned to reviewers by the production engineer; the final review is a "multi-person countersigning" node, where all assigned reviewers independently review the document online and fill in structured review comments. The system automatically aggregates the results after all reviewers have completed their work. If comments are generated during the review, the process moves to the comment sub-process, and must meet the "document review closed-loop verification" before proceeding to the next stage.

[0056] S4: Joint Inspection Organization and Comment Processing. After document review and approval, the process enters the joint inspection stage. The system automatically or the responsible person triggers the "Issue Joint Inspection Notice" to organize a joint inspection online. After the on-site inspection, the joint inspection organizer "enters comments" in the system. Once a comment is created, the system automatically assigns it to the responsible person for processing, feedback, verification, and closure, forming a closed loop. The process engine continuously monitors the closure status of "Category A Items".

[0057] S5: Certificate Preparation, Signing, Issuance, and Process Traceability Upgrade. Once all "Category I Items" are closed (through "mandatory verification of closed-loop feedback items"), the process progresses to the certificate signing stage. The system automatically marks any remaining "Category II Items" as legacy items and populates them into the corresponding positions in the certificate template. Subsequently, the certificate undergoes drafting, multi-level online review (countersigning), and approval, and is finally issued by the authorized user. Simultaneously with issuance, the system automatically generates a formal PDF certificate file and updates the process status to "Closed," completing the handover. For closed processes, the system retains a fully structured operation record, document versions, and feedback item history. As needed, authorized users can initiate an "upgrade" operation for closed processes. The system will create a new version instance based on the original process, update the original process status to "Revised," and the new process inherits necessary historical information and restarts the workflow.

[0058] For example, a specific implementation of step S1 of the method of the present invention Figure 3 As shown, the system's technical architecture and basic data configuration include the following sub-steps: S301: System Deployment and Architecture Initialization. This system is built using a microservice architecture based on Spring Cloud to achieve high cohesion and low coupling between business modules. The front-end application is developed using the Vue.js framework, providing a responsive user interface. Data storage adopts a hybrid design. Core structured business data, including user information, process instance data, master data encoding, etc., is stored in a relational database (such as DM Database) to ensure transactional consistency and performance for complex queries. Unstructured documents, drawings, and other attachments are stored in a distributed file system or object storage service, with metadata indexes recorded in the database for efficient access.

[0059] S302: Master Data Configuration and Management for Units, Systems, KKS, and Plants. The first step in system implementation is configuring the basic master data. The system administrator, through a dedicated management backend, first enters the basic information of the nuclear power plant and its units, such as configuring the plant code as "SPN" and creating "SPN1" and "SPN2" to represent Unit 1 and Unit 2 respectively. Next, the system imports the plant-wide unified system and equipment coding (KKS) system, for example, configuring "SPN1-ABC10" to represent a specific pump system of Unit 1. Then, a complete organizational tree is built in the system, covering all participating parties such as the owner company, general contractor, various construction contractors, and commissioning units, and corresponding departments and user accounts are created under each organizational node.

[0060] S303: Configuration of Entities and Relationships in the Five-Tuple Knowledge Graph. First, entity configuration is performed, establishing entities for the unit, KKS system, handover phase, and process instance. The unit entity is used for top-level data isolation; the KKS system entity uses standard KKS coding as its core; the handover phase entity uses standard phases such as EESR, TOB, TOM, TOTO, and BHO; and the process instance entity is dynamically created according to the actual handover process. Next, relationship configuration is performed, constructing five types of directed weighted relationships: strong pre-dependency, weak pre-dependency, system dependency, professional coupling, and parallel constraint. All relationships are represented by directed weighted arrows, with the arrow direction clearly indicating the business sequence logic of the predecessor and successor. Weights are assigned according to the level of business enforcement: strong pre-dependency is configured with high weight (0.9~1.0), weak pre-dependency, system dependency, and professional coupling are configured with medium weight, and parallel constraint is configured with low weight. Entities and relationships are stored uniformly in a graph database. Static entities are configured once, while dynamic instances are automatically updated as business needs change. Relationships and weights support visual configuration and dynamic expansion. New units and systems can be quickly adapted without hard coding, ultimately transforming implicit business constraints into a reasonable, computable, and iterative digital semantic network.

[0061] S304: Initialization of the Improved GNN Network and Multi-hop Pruning Parameters. When initializing the improved GNN network, the system first loads the four types of entities and five types of directed weighted relations from the quintuple knowledge graph, constructing a weighted adjacency matrix. High transitive weights are configured for strong prerequisite dependencies, and constraint gating thresholds are set to block weak associations from participating in aggregation. Neighborhood features are progressively strengthened according to the unit-system-professional hierarchy, completing the network structure and parameter initialization. Multi-hop pruning parameter initialization includes setting the maximum number of inference hops, configuring three layers of rules: filtering professionally irrelevant nodes, removing invalid nodes across units, and truncating weakly weighted relation paths. This significantly reduces computational redundancy while ensuring inference integrity, enabling the model to output prerequisite completeness scores, compliance risk levels, and missing dependency lists in real time.

[0062] S305: Loading various AI models including BERT, LSTM, and SOM. The system completes the initial loading and parameter fixing of three types of lightweight AI models: BERT, LSTM, and SOM. Through multi-model collaboration and cooperative reasoning, it achieves semantic parsing and classification of handover text, intelligent prediction of project timeline, autonomous clustering of defect root causes, intelligent comparison and verification of handover data, and dynamic optimization and scheduling of multiple tasks. Combined with a five-tuple knowledge graph and improved GNN intelligent reasoning capabilities, it constructs a multi-dimensional, full-scenario AI-powered intelligent management and control system for nuclear power handover and production, realizing the upgrade from traditional manual experience management to data-driven, model-empowered intelligent management.

[0063] S306: Role and Permission System Definition. After completing the organization and user configuration, a refined role and permission model needs to be defined. This system predefines a series of roles such as "Handover Planning Engineer," "Commissioning System Engineer," "Commissioning Team Leader," "Owner's Takeover Engineer," "Supervising Engineer," and "Contractor's Construction Manager." For each role, the permission model configures the system function menus they can access, the types of business processes they are authorized to initiate and process (e.g., only handling TOB / TOTO processes, or only handling BHO processes), and the scope of data they can manipulate (e.g., restricting commissioning engineers to only viewing and manipulating handover data for their assigned unit or the system they are responsible for). This configuration ensures the isolation of business data and the security of operations.

[0064] For example, a specific implementation of step S2 of the method of the present invention Figure 4 As shown, the handover package plan creation and process initiation verification includes the following sub-steps: S401: Handover Package Planning. The handover plan engineer enters the handover package planning management module in the system and starts the preparation of a new plan.

[0065] S402: Generate a four-dimensional code for the unit-system-handover phase-subsystem. Taking the TOB handover of the Unit 1 conventional island circulating water system (assuming KKS code XYA10) as an example, the engineer needs to create a corresponding handover package plan. In the creation interface, the engineer enters a handover package code that conforms to preset rules, such as "SPN1-XYA10-GJT-5301". The system backend uses a built-in regular expression validation engine to verify in real time whether the code strictly conforms to the structured rules of "unit-system-handover direction-handover phase-subsystem", and parses out "53" to represent the TOB handover phase. After the engineer supplements the plan description, responsible unit, and other information, they submit it for approval.

[0066] S403: Synchronous Writing of Encoded Features into the Five-Tuple Knowledge Graph. For newly created handover packages, the system automatically generates globally unique four-dimensional codes based on their corresponding unit dimension, KKS system dimension, handover stage dimension, and sub-business dimension attributes. The built-in encoding parsing engine performs refined semantic decomposition and structured feature parsing of the four-dimensional codes, accurately extracting four core inherent features: unit affiliation, system professional classification, standardized handover stage, and business application scenario. Based on this, the original encoded features obtained from the decomposition undergo standardization and normalization processing, unifying the dimension specifications, weight benchmarks, and field storage formats of all features. Redundant, invalid, and interfering field data are cleaned and removed, generating a standardized structured feature dataset that fully adapts to the storage specifications of each entity node in the five-tuple knowledge graph. This provides compliant, organized, and unified underlying data source support for subsequent multi-dimensional feature fusion, graph-based point writing, and intelligent model inference.

[0067] S404: Improved GNN-based multi-hop path reasoning. The system leverages an improved GNN graph neural network to initiate intelligent multi-hop path reasoning. Unlike traditional GNNs that rely on fixed neighborhood aggregation and indiscriminate message passing, this system incorporates hierarchical constraints and strong / weak dependencies inherent in nuclear power plant handover. It introduces hierarchical path pruning, weighted gating filtering, and multi-hop progressive aggregation mechanisms to deeply mine and accurately deduce implicit pre-dependencies and cross-professional coupling relationships among units, KKS systems, handover phases, and process instances. This provides core algorithmic support for pre-completion completeness scoring, compliance risk rating, and missing dependency screening.

[0068] S405: Handover Process Initiation and Pre-process Intelligent Verification. When a debugging system engineer needs to initiate a TOB handover for the system, they select the "Initiate TOB Handover" function in the system workbench and choose the effective handover package plan "SPN1-XYA10-GJT-5301" from the drop-down list. The moment the user confirms the creation of the process instance, the system's background intelligent rule verification and execution module is automatically triggered, executing a series of verification logic.

[0069] S406: Calculate the prerequisite completeness score and assess the compliance risk level. The system obtains all prerequisite constraints of the target handover package through multi-hop path reasoning using a quintuple knowledge graph. It assigns differentiated weight parameters to each constraint based on an improved GNN model, and calculates the prerequisite completeness quantitative score in the range of 0-100 using a weighted normalization algorithm. The specific implementation method is as follows: First, all valid prerequisite constraints obtained from the graph reasoning of the current handover package are counted, and the total weight of all constraints is summed to obtain the global weighted total score. Then, the actual closed-loop state of all prerequisite constraints is traversed, and the corresponding weights of the closed-loop completed constraints are accumulated to obtain the completed weighted total score. Finally, a normalization algorithm is used to calculate the prerequisite completeness score, with a score range of 0-100. The core calculation formula is: Completeness Score = (Weighted Total Score of Closed-Loop Prerequisite Constraints ÷ Weighted Total Score of All Prerequisite Constraints) × 100. By combining the closed-loop status of strong pre-constraints with the scoring range, the system automatically classifies compliance risks into three levels: high, medium, and low. It also simultaneously and accurately outputs a structured list of missing dependencies, enabling quantitative assessment of the completeness of pre-transfer conditions for nuclear power plant handover and intelligent, hierarchical, and rigid management of compliance risks.

[0070] S407: Process Instance Creation or Interception. If all the above checks pass successfully, the system creates a new TOB process instance, initializes its status to "In Progress," and automatically imports key information such as the handover package code and system name into the application form for engineers to fill in details. If any check fails, such as if the preceding EESR process is not closed, the system will immediately display a clear warning message on the front-end interface, such as "Check Failed: Unable to initiate TOB process; the EESR handover corresponding to this handover package (SPN1-XYA10-GJT-5201) is incomplete," and prevent the creation of the current process instance, thereby ensuring the compliance of the process logic at the source.

[0071] For example, a specific implementation of step S3 of the method of the present invention Figure 5 As shown, document review collaboration and comment processing include the following sub-steps: S501: Document Review Task Assignment and Initiation. When the TOB (Transfer-to-Build) application process completes internal review and enters the document review stage, the owner's takeover engineer, as the person responsible for the initial document review, will receive a pending task notification from the system. After reviewing the application documents online, this engineer can assign subsequent detailed review tasks to multiple professional engineers in the system, such as electrical engineers and instrumentation engineers, based on professional needs.

[0072] S502: Multi-person review mechanism implemented. Assigned electrical and instrumentation engineers will simultaneously receive independent review tasks. They will review the documents online and fill in their comments in the structured review comment form provided by the system. In this embodiment, the review node is configured in a "unanimous agreement" review mode. This means the system will wait for all assigned reviewers to submit their review conclusions before automatically aggregating the results. If all reviewers select "Review Passed" or "Agree but with Comments," the node passes overall; if any reviewer selects "Reject," the node fails, and the process may be reversed.

[0073] S503: Generation and Tracking of Review Comments. During the review process, for each specific comment submitted by the reviewer, the system automatically creates a corresponding "Document Review Comment Item" record. This record is initially categorized as "Category 1 Item" and associated with information such as the current TOB process instance, the submitter, and the submission time. These comment items then enter an independent comment item processing sub-process.

[0074] S504: Comment Item Processing and Closed-Loop Verification. The process initiator (or designated handler) will see newly generated pending comment items in their workbench. For each comment item, the handler must submit processing measures or explanations online and can upload supporting documents. After submission, the person who initiated the comment item will automatically receive a verification task and can confirm online whether the processing result meets the requirements. If the verification passes, the comment item's status will be updated to "Closed". The system continuously tracks the status of all related comment items.

[0075] S505: Document review closed-loop verification drives the process forward. Once all comments generated from this document review reach the "closed" state, the document review phase is truly complete. At this point, the system's built-in "document review closed-loop verification" rule is satisfied. Satisfying this rule is the key criterion for the process engine to determine whether to allow the current TOB process to exit the "document review" stage and automatically proceed to the next stage (such as "issuing a joint inspection notice"), thus achieving mandatory closed-loop management of review quality.

[0076] Specific implementation of step S4 of the method of the present invention, for example Figure 6 As shown, the processing of on-site joint inspection comments and mandatory control before signing include the following sub-steps: S601: On-site Joint Inspection and Comment Entry. After the process enters the on-site joint inspection stage, the joint inspection organizer issues a joint inspection notice through the system and organizes the on-site inspection. After the inspection, the person in charge of the joint inspection centrally enters the problems found on-site into the system, forming "Joint Inspection Comments". Each comment must be clearly described, specify the responsible unit, require a deadline for removal, and be initially categorized, for example, into "Category 1 Items" that must be removed before signing, or "Category 2 Items" that can be left behind but need to be recorded.

[0077] S602: Feedback Item Assignment and Processing Tracking. The system automatically assigns entered feedback items to the corresponding responsible units or individuals. The responsible individuals receive the tasks online and provide feedback on the handling plan and implementation status. For complex issues, the responsible individuals can initiate a "classification application" in the system to request that "Category I items" that cannot be completely closed in the short term be converted to "Category II items."

[0078] S603: Online Approval of Classification Change Requests. After a classification change request is initiated, the system follows a predefined approval path (e.g., requiring online approval sequentially by the transfer manager, professional department manager, and debugging team leader). Ultimately, the original submitter of the request must provide final confirmation. Upon approval, the system updates the current classification of the request to "Category II," but retains the initial classification information in its history for future reference.

[0079] S604: Mandatory Closed-Loop Verification in the Signature Process. When all comments have been processed and the process is ready to proceed to the certificate handover signing stage, the person in charge of the handover triggers the "Prepare for Signature" operation. At this time, the system will forcibly execute "Mandatory Closed-Loop Verification of Comments." The core logic of this verification is: scan all comments associated with the current process, filter out records whose "initial classification" is "Category 1," and check their status. This verification is only successful when all comments initially classified as "Category 1" (including those approved for conversion to Category 2) are in the "Closed" state. If any such item is not closed, the system will display a strong warning and intercept the process, absolutely prohibiting entry into the signing state, thus firmly establishing a security and quality defense line.

[0080] S605: Legacy item information is automatically imported into the certificate. For "Category II items" that are still in the "not closed" state after verification, the system will automatically extract the information of these items (such as description, responsible party, etc.) and fill them into the "Legacy Item List" section of the certificate template when generating the handover certificate, ensuring complete information transfer.

[0081] For example, a specific implementation of step S5 of the method of the present invention Figure 7 As shown, structured data output and multidimensional decision analysis include the following sub-steps: S701: Structured Generation and Issuance of Transfer Certificates. Once the transfer certificate has completed all online approval and digital signature (stamp) processes, the authorized user performs the final "issuance" operation. At this moment, the system's structured data storage and output module is activated. This module extracts all structured data finalized in this process from the database, including but not limited to: basic process information, approved equipment boundary list, signatures and timestamps of all parties, and a list of legacy items.

[0082] S702: Automatic generation of templated documents. The system automatically generates certificates from the extracted structured dataset according to the templates of each factory; populates the core list and legacy items; completes multi-level digital signing, and binds signatures and timestamps. Through the backend document processing engine (such as integrating iText libraries), the data is automatically and accurately filled into the corresponding field positions of the template, dynamically generating a formal PDF certificate file that is formatted correctly, contains authoritative content, and can be directly used for archiving, and provides a download link.

[0083] S703: End-to-end data aggregation and statistical analysis. Simultaneously, the system's unified workbench and decision support module continuously run data aggregation tasks in the background. This module constructs a multi-dimensional data analysis model based on the structured data accumulated throughout the entire process.

[0084] S704: Management View and Decision Dashboard Display. After logging into the system, project managers can access various statistical analysis dashboards. For example, on the "Overall Project Handover Dashboard," they can intuitively view the monthly handover plan completion rate, the total number of comments for each unit or contractor, and the closure trend. The system supports data drill-down; clicking on a chart allows further viewing of which systems have overdue comments, along with details of these comments and the current responsible party.

[0085] S705: Data-driven management decision support. Through the real-time, accurate, and visualized data presentation described above, project managers can quickly grasp the overall health of project handover, accurately locate bottlenecks (such as lags in specific contractors or professional fields), and thus make targeted resource coordination, schedule expediting, and decision adjustments, achieving a shift from experience-based management to data-driven management.

[0086] Through the system deployment and application of the above specific implementation methods, the present invention successfully transforms the complex and rigorous handover and management system of nuclear power plants into a stable, efficient, intelligent, and traceable digital collaborative work platform, effectively solving the various problems mentioned in the background technology and achieving all the objectives of the present invention.

[0087] This application may also provide a computer device, including: at least one processor, memory, at least one network interface, and a user interface. The various components in this device are coupled together via a bus system. It is understood that the bus system is used to implement communication between these components. In addition to a data bus, the bus system also includes a power bus, a control bus, and a status signal bus.

[0088] The user interface can include a display, keyboard, or clicking device. Examples include a mouse, trackball, touchpad, or touchscreen.

[0089] It is understood that the memory in the embodiments disclosed in this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memories described herein are intended to include, but are not limited to, these and any other suitable types of memory.

[0090] In some implementations, the memory stores elements such as executable modules or data structures, or subsets thereof, or extended sets thereof: operating systems and applications.

[0091] The operating system includes various system programs, such as the framework layer, core library layer, and driver layer, used to implement various basic business functions and handle hardware-based tasks. The application programs include various applications, such as media players and browsers, used to implement various application functions. Programs implementing the methods of the embodiments of this disclosure can be included in the application programs.

[0092] In the above embodiments, the processor can also invoke programs or instructions stored in memory, specifically programs or instructions stored in an application program, for the following purposes: Follow the steps described above.

[0093] The above methods can be applied to or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above methods can be completed by integrated logic circuits in the processor's hardware or by software instructions. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the disclosed methods, steps, and logic block diagrams. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the disclosed methods can be directly implemented by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0094] It is understood that the embodiments described in this application can be implemented using hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described in this application, or combinations thereof.

[0095] For software implementation, the technology of this application can be implemented by executing the functional modules (e.g., procedures, functions, etc.) of this application. The software code can be stored in memory and executed by a processor. The memory can be implemented in the processor or outside the processor.

[0096] This application may also provide a non-volatile storage medium for storing a computer program. When the computer program is executed by a processor, it can implement the steps in the above method embodiments.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit it. Although this application has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of this application do not depart from the spirit and scope of the technical solutions of this application, and should all be covered within the scope of the claims of this application.

Claims

1. A collaborative management system for the entire process of nuclear power plant handover and commissioning, characterized in that, include: The master data and access control module is used to configure and maintain the baseline data for system operation and a unified access control mechanism. The handover package plan management module is used to create, maintain, and manage the entire lifecycle status of handover package plans. The business process engine module has a built-in standardized process model library covering the entire process of nuclear power plant handover and production. It is used to create process instances, push task nodes to designated personnel and intelligently control the flow path, and record the history of process status changes. It also supports process suspension, rollback, cancellation and version upgrade operations. The core intelligent module is the quintuple knowledge graph and the improved GNN intelligent reasoning module, which are used to construct the quintuple knowledge graph and set differentiated weights for different dependencies; the improved GNN message passing strengthens strong pre-dependencies through weighted aggregation, gated truncation and hierarchical aggregation. After multi-hop path pruning, the system outputs completeness, risk level, and missing item list, providing intelligent decision-making basis for process initiation and node release. The decision node, deeply integrated into the business process engine module, runs in the form of a front-end service or automatic verifier to enforce business compliance logic. The module also includes three types of AI models: lightweight BERT, LSTM, and SOM, which respectively realize semantic parsing and classification of handover text, intelligent prediction of project timeline, autonomous clustering of defect root causes, intelligent comparison and verification of handover data, and dynamic optimization and scheduling of multi-tasks. The joint inspection and feedback item collaborative management module is used to organize joint on-site inspections and to track and manage the issues found during the inspections throughout their entire lifecycle. The intelligent rule verification and execution module is used to support the validity verification of handover packages, the completion verification of the knowledge graph reasoning pre-process, the closed-loop verification of opinion items, the closed-loop verification of document review, and the rules for multi-person countersigning. The structured data storage and output module is used to define and persistently store the core business information generated in the business process in a structured manner. The unified workbench and decision support module provides a personalized portal interface for users with different roles within the system, centrally displaying pending tasks, processed transactions, and processes of interest. It also provides multi-dimensional statistical analysis and visualization functions based on the structured data accumulated throughout the entire process, generating various management dashboards and reports on handover progress, status of comments, and review efficiency.

2. The system according to claim 1, characterized in that, The functions of the quintuple knowledge graph and the improved GNN intelligent reasoning module include: Construct a domain knowledge graph consisting of five elements: unit, system, handover phase, process entity, and dependency. Define four types of entities: unit, KKS system, handover phase, and process instance, as well as five types of directed weighted edges: strong precondition, weak precondition, system dependency, professional coupling, and parallel constraint. The message passing mechanism of GNN is improved by assigning high weights to strong pre-dependencies, using weighted feature aggregation, setting constraint gating thresholds to truncate weakly related edges, and performing hierarchical neighborhood aggregation by unit-system-professional to enhance the dominance of reasoning based on strong pre-dependencies. Supports a three-layer multi-hop path inference pruning method that includes professional irrelevant filtering, cross-unit invalid node removal, and weak weighted relational path truncation, automatically reducing irrelevant nodes and redundant paths; Automatically calculate the completeness score from 0 to 100, classify compliance risk levels as high / medium / low based on the score, and output a detailed list of unclosed and missing dependencies.

3. The system according to claim 2, characterized in that, The weighted feature aggregation is represented as follows: h v =∑W uv × h u ; Among them, h v W is the feature vector of the target node, representing the new feature representation after aggregation of the nodes currently being updated; uv h is the edge weight / dependency strength coefficient, representing the weight value of the edge connecting neighbor node u to target node v; u is the feature vector of a neighboring node, representing the feature representation of a certain neighboring node of the current node v.

4. A collaborative management method for the entire process of nuclear power plant handover and commissioning, implemented based on the system described in any one of claims 1-3, comprising: Step 1: System initialization and configuration; Configure basic data for the power plant, generating units, and system, and define various user roles and permissions; Configure various standard handover process models and their corresponding intelligent verification rules; construct a standardized five-tuple domain knowledge graph of units, systems, handover phases, process entities, and dependencies; Step 2: Handover plan creation and process initiation verification; Create a handover package plan based on the project plan, compile a structured handover package code that conforms to the rules, and approve its status as "valid"; select the handover type to be executed, associate it with a "valid" handover package plan, and initiate a new handover process; load the handover package information and trigger "handover package validity check" and "pre-process closure check"; if the check passes, the process instance is successfully created and enters the compilation state; Step 3: Online approval and document review of the process; The process flows sequentially through the drafting, internal review, and submission stages, reaching the document review node. All assigned reviewers independently review the document online and fill in structured review comments. The results are aggregated after all reviewers have completed their reviews. If comments are generated during the review, the process moves to the comments sub-process. If the "document review closed-loop verification" is met, the process proceeds to the next stage. Step 4: Joint inspection organization and handling of comments; After the document review is approved, the process enters the joint inspection stage. The system automatically or the person in charge triggers the "issue joint inspection notice" to organize a joint inspection online. After the on-site inspection, comments are entered. After the comments are created, the system automatically assigns them to the responsible person for processing, feedback, verification and closure, forming a closed loop. The process engine continuously monitors the closure status of "Category A items". Step 5: Certificate preparation, signing, issuance, and process traceability upgrade; once all "Category I items" are closed, the process proceeds to the certificate signing stage; the system automatically marks any unclosed "Category II items" as legacy items and fills them into the corresponding positions in the certificate template; The certificate goes through a process of compilation, multi-level online review and approval, and is finally issued by the authorized user. At the same time as the certificate is issued, the system automatically generates the certificate file and updates the process status to "closed", thus completing the handover. For closed processes, the system retains a full-process structured operation record, document version, and comment history.

5. The collaborative management method for the entire process of nuclear power plant handover and commissioning according to claim 4, characterized in that, Step 1 includes: Step 1-1: Configure basic master data; enter the basic information of the nuclear power plant and units, import the unified system and KKS system for the entire plant, and build a complete organizational tree; Steps 1-2: Configure the 5-tuple knowledge graph entities and relationships. First, configure the entities, creating unit entities, KKS system entities, handover phase entities, and process instance entities. Unit entities are used for top-level data isolation, KKS system entities are based on standard KKS coding, handover phase entities solidify the standard handover phases, and process instance entities are dynamically created according to the actual handover process. Next, configure the relationships, constructing five types of directed weighted relationships: strong pre-dependency, weak pre-dependency, system dependency, professional coupling, and parallel constraint. All relationships are represented by directed weighted arrows, with the arrow direction clearly indicating the business sequence logic of the predecessor and successor. Weights are assigned according to the business enforcement level: strong pre-dependency is configured with high weight, weak pre-dependency, system dependency, and professional coupling are configured with medium weight, and parallel constraint is configured with low weight. Entities and relationships are stored uniformly in the graph database; static entities are configured once, and dynamic instances are automatically updated according to business needs. Steps 1-3: Initialize the improved GNN network and multi-hop pruning parameters; When initializing the improved GNN network, first load the four types of entities and five types of directed weighted relations of the quintuple knowledge graph to construct a weighted adjacency matrix; configure high transitive weights for strong pre-dependencies, set constraint gating thresholds to block weak associations from participating in aggregation, and implement progressive strengthening of neighborhood features according to unit-system-professional hierarchy to complete the network structure and parameter initialization; Multi-hop pruning parameter initialization includes setting the maximum number of inference hops, configuring three layers of rules: filtering professionally irrelevant nodes, removing invalid nodes across units, and truncating weak weighted relation paths; Steps 1-4: Load BERT / LSTM / SOM models; complete the initial loading and parameter fixing of lightweight BERT, LSTM, and SOM AI models; Steps 1-5: Define the role and permission system; After completing the organization and user configuration, define a refined role and permission model.

6. The collaborative management method for the entire process of nuclear power plant handover and commissioning according to claim 4, characterized in that, Step 2 includes: Step 2-1: Develop a handover package plan and generate a four-dimensional code for the unit, system, handover phase, and subsystem; Step 2-2: The four-dimensional code is semantically decomposed and structured feature parsed using the encoding parsing engine to extract four core inherent features: unit affiliation, system professional classification, standardized handover stage, and business application scenario. The original encoded features obtained from the decomposition are standardized and normalized to unify the dimension specifications, weight benchmarks, and field storage formats of all features. Redundant, invalid, and interfering field data are cleaned and removed to generate a standardized structured feature dataset that is adapted to the storage specifications of each entity node in the five-tuple knowledge graph. Steps 2-3: Improved GNN to initiate multi-hop path reasoning; The system relies on the improved GNN graph neural network to initiate intelligent multi-hop path reasoning. Combining the hierarchical constraints and strong and weak dependencies of nuclear power plant handover, it introduces hierarchical path pruning, weighted gating filtering and multi-hop progressive aggregation mechanism to achieve in-depth mining and deduction of implicit pre-dependencies and cross-professional coupling relationships among units, KKS systems, handover stages and process instances. Steps 2-4: Handover process initiation and pre-process intelligent verification; After the handover process is initiated, the system obtains the full set of pre-process constraints of the target handover package through multi-hop path reasoning of the five-tuple knowledge graph, assigns differentiated weight parameters to each constraint based on the improved GNN model, calculates the pre-process completeness quantitative score in the range of 0 to 100 through the weighted normalization algorithm, and divides the compliance risk levels into high, medium and low levels by combining the closed-loop state of strong pre-process constraints and the score range, and outputs a structured list of missing dependencies simultaneously; Steps 2-5: Create or intercept a process instance; if all pre-process smart checks pass successfully, create a new process instance and initialize its status to "In Development"; if any check fails, issue a reminder and prevent the creation of the current process instance.

7. The collaborative management method for the entire process of nuclear power plant handover and commissioning according to claim 4, characterized in that, Step 3 includes: Step 3-1: Initiate document review task assignment; When the process completes internal review and enters the document review stage, the system sends a pending task notification to the reviewer and assigns the review task to the corresponding reviewer. Step 3-2: Implement a multi-person review mechanism; reviewers review the documents online and fill in their comments in the structured review comment form provided by the system; if all reviewers approve the document, the process continues. Steps 3-4: Generate and track review comments; During the review process, for each specific comment filled in by the reviewer, the system will automatically create a corresponding "Document Review Comment Item" record. This record is initially classified as "Category 1 Item" and associated with the current process instance, the submitter, and the submission time information; These comments then enter an independent comment item processing sub-process. Steps 3-5: Handling and Closing-Loop Verification of Comments; For each comment, the person handling it submits the handling measures or explanations online, and supports uploading supporting documents; After submission, the person who submitted the comment will automatically receive a verification task and confirm online whether the handling result meets the requirements; If the verification is successful, the status of the comment will be updated to "Closed"; The system continuously tracks the status of all related comments. Steps 3-6: Document review closed-loop verification drives the process forward; when all comments generated by this document review reach the "closed" state, it marks the completion of the document review stage; at this time, the system's built-in "document review closed-loop verification" rule is satisfied. The satisfaction of this rule is the basis for the process engine to decide whether to allow the current process to exit the "document review" stage and move to the next stage.

8. The collaborative management method for the entire process of nuclear power plant handover and commissioning according to claim 4, characterized in that, Step 4 includes: Step 4-1: On-site joint inspection and entry of comments; After the process enters the on-site joint inspection stage, the joint inspection organizer issues a joint inspection notice through the system and organizes the on-site inspection; After the inspection, the person in charge of the joint inspection centrally enters the problems found on-site into the system to form "joint inspection comments"; Each comment includes a description, responsible unit and required removal deadline, and is initially classified; The classification includes "Category I items" and "Category II items"; Step 4-2: Assigning comments and tracking the processing; The system automatically assigns the entered comments to the corresponding responsible units or individuals; The responsible persons receive the tasks online and provide feedback on the processing plan and implementation status; For complex issues, the system supports the responsible persons to initiate a "classification application" to request that "Category 1 items" that cannot be completely closed in the short term be converted to "Category 2 items"; Step 4-3: Online approval of the classification change application; After the classification change application is initiated, the system follows the predefined approval path and is finally confirmed by the original proposer of the opinion item; After the approval is approved, the system updates the current classification of the opinion item to "Category II item", but the initial classification information is still retained in its history for traceability; Step 4-4: Mandatory Closed-Loop Verification of the Signature Process; When all comments have been processed and the process is ready to enter the certificate signing stage, the person in charge of the transfer triggers the "Prepare for Signature" operation, and the system forcibly executes "Mandatory Closed-Loop Verification of Comments"; The "Mandatory Closed-Loop Verification of Comments" is as follows: Scan all comments associated with the current process, filter out records initially classified as "Category 1 Items", and check their status. Only when the status of all comments initially classified as "Category 1 Items" is "Closed" can the verification pass; If any such item is not closed, the system will pop up a prompt and intercept the process, prohibiting it from entering the signing state; Steps 4-5: Legacy item information is automatically imported into the certificate; for "Category II items" that are still in the "not closed" state after verification, the system automatically extracts the information of these items and fills them into the "Legacy Item List" section of the certificate template when generating the handover certificate, ensuring complete information transfer.

9. The collaborative management method for the entire process of nuclear power plant handover and commissioning according to claim 4, characterized in that, Step 5 includes: Step 5-1: Generate and publish the handover certificate; After the handover certificate has completed all online approval and digital signature processes, the authorized user performs the final "publish" operation; The structured data storage and output module extracts all the structured data finally determined in this process from the database, including: basic process information, approved equipment boundary list, signatures of all parties, signature timestamps, and list of legacy items; Step 5-2: Generate templated files; The system will automatically generate certificates from the extracted structured data according to the templates of each factory, fill in the core list and legacy items, complete multi-level digital signing, and bind the signature and timestamp; The document processing engine will fill the data into the corresponding field positions of the template, dynamically generate a PDF format official certificate file that can be directly used for archiving, and provide a download link; Step 5-3: Full-process data aggregation and statistical analysis; The unified workbench and decision support module run data aggregation tasks, and based on the structured data accumulated throughout the process, build a multi-dimensional data analysis model to support the generation and display of various statistical data.

Images