Special equipment report intelligent auditing system and method

Abstract

The invention relates to the technical field of special equipment, in particular to an intelligent auditing system and method for a special equipment report, and the system comprises an infrastructure layer, a data layer, a service layer and a display layer. A server platform and an operating system are deployed in the infrastructure layer so as to provide hardware resources, the operating system, middleware and a network environment for system operation, the data layer is connected with the infrastructure layer, a relational database is adopted to store service data, and a memory database is utilized to perform data caching; the service data comprises a test report template, a parameter variable and a rule set; the service layer is connected with the data layer and is used for bearing an auditing rule engine and a business processing module; the display layer is connected with the service layer and comprises a management end and a user end, the management end provides rule configuration and system management functions, and the user end is integrated in an existing service system in a browser plug-in mode. Therefore, the detection efficiency and accuracy of the special equipment report are improved.

Description

Technical Field

[0001] This invention relates to the field of special equipment technology, specifically to an intelligent review system and method for special equipment reports. Background Technology

[0002] Special equipment inspection reports are crucial technical documents for ensuring the safe operation of special equipment, covering various equipment types such as elevators, cranes, and boilers. With the continuous growth in the number and increasing diversification of special equipment in my country, the demand for inspection and testing is constantly increasing, placing higher demands on the accuracy, efficiency, and traceability of report preparation.

[0003] Currently, the review of special equipment inspection reports mainly relies on manual methods. Inspectors record inspection data through on-site surveys, including but not limited to specific parameters such as elevator door opening angles and boiler wall thickness measurement points, and then enter the data into a comprehensive management platform. Reviewers then manually review each item of the report based on their professional experience and relevant laws and regulations. This review method is inefficient and cannot meet the needs of batch testing. Furthermore, the accuracy of the review is limited by the technical level and experience of the reviewers; different personnel may have different understandings of the standard clauses, easily leading to omissions or misjudgments. Therefore, the testing efficiency and accuracy of special equipment reports are relatively low. Summary of the Invention

[0004] To address the technical problem of low inspection efficiency and accuracy in special equipment reports, the present invention aims to provide an intelligent review system and method for special equipment reports. The specific technical solution adopted is as follows:

[0005] In a first aspect, embodiments of the present invention disclose an intelligent auditing system for special equipment reports. The intelligent auditing system for special equipment reports includes: an infrastructure layer, a data layer, a service layer, and a presentation layer. The infrastructure layer deploys a server platform and operating system to provide the hardware resources, operating system, middleware, and network environment for system operation. The data layer is connected to the infrastructure layer, uses a relational database to store business data, and utilizes an in-memory database for data caching. The business data includes inspection report templates, parameter variables, and rule sets. The service layer is connected to the data layer and is used to carry the audit rule engine and business processing module to parse inspection report data, execute audit rules, and return audit conclusions. The presentation layer is connected to the service layer and includes a management terminal and a user terminal. The management terminal provides rule configuration and system management functions, and the user terminal is integrated into the existing business system as a browser plugin.

[0006] Secondly, this invention discloses an intelligent review method for special equipment reports. The method includes: accessing report data from an inspection report template in a smart special equipment inspection management platform via a browser plugin; importing the inspection report template and extracting parameters to generate a template, and creating variables based on the parameters in the template; manually configuring a first rule or automatically generating a second rule based on a large model, and creating and activating a rule set based on the first and second rules; identifying the report type of the inspection report template and matching the rule set, executing the review rules, and returning the review result, which includes pass items, critical items, and general items.

[0007] Through the technical solutions disclosed in this invention, the server platform and network environment at the infrastructure layer provide stable computing power. The data layer caches business data through an in-memory database and, in conjunction with the structured storage of a relational database, enables rapid retrieval and parsing of inspection report data. The service layer's audit rule engine replaces manual verification, automatically executing audit rules and returning conclusions, solving the technical problem of low efficiency in manual auditing and efficiently handling the auditing needs of batch testing reports. The system transforms legal norms and inspection standards into standardized rule sets stored in the data layer and uniformly executes audit logic through the rule engine, avoiding misunderstandings of standards due to differences in technical level and experience during manual auditing. This reduces omissions and misjudgments at the source, ensuring the objectivity and consistency of audit results, thereby improving the accuracy of special equipment inspection report audits. The management end supports rule configuration and system management, flexibly adapting to updates in inspection standards and changes in business needs, thus improving the flexibility of special equipment inspection reports. The user end is integrated into the existing business system as a browser plugin, eliminating the need to reconstruct the original platform, reducing deployment costs, and facilitating quick operation for inspection and audit personnel, further improving the efficiency of special equipment inspection report audits. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the structure of an intelligent review system for special equipment reports provided in an embodiment of the present invention;

[0009] Figure 2 A flowchart illustrating an intelligent review method for special equipment reports provided in an embodiment of the present invention;

[0010] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0011] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a special equipment report intelligent review system and method proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The specific details of a special equipment report intelligent review system and method provided by this invention are described below with reference to the accompanying drawings.

[0013] Please see Figure 1 , Figure 1 This is a schematic diagram of the structure of a special equipment report intelligent review system provided in an embodiment of the present invention. The special equipment report intelligent review system includes: an infrastructure layer, a data layer, a service layer, and a presentation layer.

[0014] The infrastructure layer deploys server platforms and operating systems to provide hardware resources, operating systems, middleware, and network environment for system operation. The data layer connects to the infrastructure layer, uses a relational database to store business data, and utilizes an in-memory database for data caching. Business data includes inspection report templates, parameter variables, and rule sets.

[0015] The service layer connects to the data layer and is used to carry the audit rule engine and business processing module to parse inspection report data, execute audit rules and return audit conclusions. The presentation layer connects to the service layer and includes the management end and the user end. The management end provides rule configuration and system management functions, and the user end is integrated into the existing business system as a browser plugin.

[0016] Specifically, the infrastructure layer in this embodiment provides the underlying support environment. At the hardware and operating system level, the system design supports domestic alternatives and can run stably on mainstream servers and domestic operating systems (such as Kylin and UnionTech UOS). At the software middleware level, this embodiment uses domestically compatible Nginx as the web server and Tomcat as the application container, and supports the HTTPS protocol to ensure data transmission security. Furthermore, this embodiment introduces Docker containerization technology at the infrastructure layer to containerize system service modules, enabling rapid application deployment, isolation, and elastic scaling based on business load.

[0017] Furthermore, in this embodiment of the invention, the data layer is responsible for the persistence and caching management of all system data. Core business data, including inspection report templates, parameter variables, rule sets, etc., are stored in a structured manner using a relational database (such as MySQL). The relational database ensures data integrity and consistency. To meet the response speed requirements of the intelligent auditing system, this embodiment also introduces a Redis in-memory database as a cache to store frequently accessed auditing rules and other data, significantly reducing direct access to the core database and thus improving the efficiency of rule retrieval and execution.

[0018] Furthermore, in this embodiment of the invention, the service layer carries the audit rule engine and various business processing modules. The backend service is implemented using Java technology and is mainly responsible for parsing the incoming inspection report data and driving the audit rule engine to perform the audit. The audit rule engine in this embodiment of the invention adopts a dual-mode execution mechanism: one is the rule engine execution mode, which is suitable for precise and quantifiable rigid rules (such as "door gap < 6mm"), where the system directly performs logical judgments, resulting in fast execution speed and accurate results. The other is the large model execution mode, which is suitable for flexible rules that require semantic understanding (such as judging the standardization of filling), which uses a large language model for analysis and decision-making to handle complex scenarios that traditional rule engines cannot handle. In this way, the dual-mode execution mechanism simultaneously takes into account the efficiency, accuracy, and intelligent flexibility of the audit.

[0019] Furthermore, the presentation layer, serving as the interaction interface between the system and users, is divided into a management terminal and a user terminal. The management terminal is a standalone web application providing a user-friendly graphical interface for system administrators and rule configuration personnel. It is used for report template management, parameter variable definition, configuration and publishing of audit rules / rule sets, and statistical analysis of system-wide audit results. The user terminal adopts a non-intrusive integration solution, embedding itself as a browser plugin into the existing business systems used daily by inspection personnel. This allows inspection personnel to directly invoke the "one-click intelligent audit" function when compiling reports without switching systems. Audit results are provided in real-time via prompts, achieving seamless integration between new functions and existing workflows, greatly improving usability. It is worth noting that the existing business system can be a smart special inspection management platform, etc., and this embodiment of the invention is not limited thereto.

[0020] Furthermore, as an optional embodiment of the present invention, the infrastructure layer also includes: a containerized deployment module for rapid deployment and an elastically scalable service module.

[0021] Specifically, the containerized deployment module in this embodiment of the invention is based on Docker containerization technology, which builds each business module in the service layer (such as the rule engine service, report parsing service, etc.) into an independent Docker image. Each image contains the code, runtime environment, system tools and library files necessary to run the service, thereby ensuring a high degree of consistency between development, testing and production environments.

[0022] Furthermore, to achieve unified orchestration and elastic management of services, the system adopts Kubernetes as the container orchestration engine. In a Kubernetes cluster, each service image is defined as a dynamically replicable Pod instance. Moreover, when concurrent audit requests from users increase, and the system monitoring module detects that service layer resource utilization exceeds a preset threshold, the containerized deployment module will trigger an elastic scaling mechanism. This mechanism can automatically increase the number of Pod instances for relevant services based on real-time load, and automatically reduce the number of instances after the traffic peak, releasing computing resources and thus optimizing costs.

[0023] Furthermore, as an optional embodiment of the present invention, the data layer further includes: a template management module for establishing a one-to-one correspondence between test report templates and report types; and a parameter management module for establishing a one-to-many relationship between test report templates and parameters.

[0024] Specifically, inspection report types include, but are not limited to, periodic inspection reports and supervisory inspection reports, while special equipment types include, but are not limited to, elevators and cranes. The core function of the template management module is to establish a one-to-one correspondence between inspection report templates and report types. When a user downloads raw JSON data of a report from the "Smart Special Inspection Management Platform" via a browser plugin, they can import it into this system. The template management module will parse the structure of the JSON data and automatically create or update a corresponding inspection report template. During this process, the system extracts the keys from the JSON as the core parameters of this type of report and identifies the data type of each key's corresponding value, thereby completing the standardized definition of the template. Once the template is successfully imported, the system establishes and persists a strong association between the report type and this inspection report template in the database.

[0025] Furthermore, the core function of the parameter management module is to establish a one-to-many relationship between the inspection report template and its parameters. After the inspection report template is created, the parameter management module uses all the JSON keys parsed from the template as basic parameters and stores them in the database, establishing a one-to-many association with the template. Each parameter includes not only its name but also metadata such as its data type. More importantly, this module supports creating variables based on the basic parameters. For example, based on the basic parameter "door gap," a variable named "Is the door gap greater than 6mm?" can be created, with the expression defined as door gap >= 6.

[0026] Furthermore, as an optional embodiment of the present invention, the service layer's audit rule engine includes: a manual rule generation module, which creates manually configured rules based on parameter variables, and the manually configured rules support logical and arithmetic operations; an automatic rule generation module, which converts semantic rules into executable rules based on large model capabilities, and the audit rules include manually configured rules and executable rules; and a rule execution module, which executes the audit rules and generates audit conclusions.

[0027] Specifically, in the manual rule generation module of this embodiment, the user selects one or more pre-created variables or parameters from the parameter pool of a specified inspection report template through the management interface of the presentation layer. Pre-created variables or parameters include, for example, "whether the door gap is greater than 6mm". The manual rule generation module provides a graphical interface for users to define rule conditions. The module supports complex logical and arithmetic operations to combine multiple variables into composite conditions. All manually configured rules are persistently stored in the data layer in a structured JSON format for use by the rule execution module.

[0028] Furthermore, the automatic rule generation module leverages the semantic understanding and code generation capabilities of the large language model to automatically convert textual descriptions of regulations and semantic rules into structured rules that the system can execute. The specific implementation process is as follows: The user inputs a semantic rule on the management end. The module first sends this text to the large model, requiring it to identify and extract the key parameters involved in the rule. The system verifies whether these parameters already exist in the parameter pool of the current report template. If the parameters exist, the module further checks whether there are directly related variables, such as "whether the horizontal distance is greater than 35mm". If they exist, the variable is used directly to construct the rule. If they do not exist, the large model is called again to automatically generate a variable expression that meets the requirements. The module guides the large model to generate a complete rule tree with nested structures based on the logic of the semantic rule. During this process, the system prioritizes searching for and reusing logically similar rule nodes in the existing rule library to avoid duplicate creation, thereby improving efficiency and ensuring rule consistency. The generated executable rules are temporarily stored in the test environment.

[0029] Furthermore, the rule execution module is responsible for executing both manually and automatically generated review rules and generating the final review conclusion. The module employs a dual-mode execution strategy to balance efficiency and flexibility: Rule Engine Execution Mode: For deterministic, quantifiable rules, such as "door gap < 6mm," the module directly calls the embedded high-performance expression calculation engine. This mode is extremely fast and suitable for most rigid technical indicators. Large Model Execution Mode: For flexible rules requiring contextual understanding or complex logical judgment, such as "the report filler and reviewer should not be the same person," the module integrates the structured description of the rule with the current report's contextual data into a prompt, which is then sent to the large model for reasoning. The large model returns "pass," "fail," or specific reasons for the judgment based on its understanding.

[0030] During the review process, the module first loads the corresponding rule set based on the report type, then iterates through each rule in the set and intelligently selects the execution mode based on the rule's configuration. The execution results of all rules are aggregated in real time, ultimately generating a structured review conclusion, which is then returned to the user through the presentation layer.

[0031] In this way, through the collaborative work of the above modules, the audit rule engine achieves intelligent management of the entire lifecycle from rule creation and optimization to execution, ensuring the efficiency, accuracy and flexibility of the audit process.

[0032] Furthermore, as an optional embodiment of the present invention, the automatic rule generation module further includes: a parameter detection unit for detecting whether the parameters in the semantic rules exist in the template parameter pool; a variable verification unit for checking whether the variables associated with the parameters in the semantic rules meet the requirements of the semantic rules; a rule generation unit for creating variables and executable rules using a large model and generating a rule tree; and a test and release unit for storing the rule tree and performing test verification.

[0033] Specifically, when a user inputs a semantic rule on the management interface, the parameter detection unit is activated. First, it calls the integrated large language model, leveraging its natural language processing capabilities to perform entity recognition and keyword extraction on the text, accurately extracting the core parameters. Then, the parameter detection unit automatically compares and verifies the extracted parameter list against a pre-defined template parameter pool in the data layer for this type of report template, confirming whether these parameters already exist in the system. This avoids referencing non-existent parameters.

[0034] Furthermore, after the parameter detection passes, the rule generation unit retrieves all created variables associated with these parameters. For example, querying variables related to the parameter "horizontal distance" might include "whether the horizontal distance is greater than 35mm" or "whether the horizontal distance is a number." The unit further analyzes the logical expressions of existing variables to determine if they meet the requirements of semantic rules. For example, if the rule requires "not to be greater than 35mm," the logic is horizontal distance <= 35. If the existing variable "whether the horizontal distance is greater than 35mm" has a logic of horizontal distance > 35, the unit determines that the rule requirement can be met by logical negation, thus identifying the variable as directly reusable. If an existing variable cannot directly or after simple transformation meet the requirements, it is marked as needing to create a new variable.

[0035] Furthermore, for newly created variables or rules, the rule generation unit constructs a structured prompt. This prompt integrates the semantic rules to be converted, relevant parameter information, a list of platform-supported functions and operators, and a few rule examples, providing them to the large model. If the variable validation unit determines that a new variable needs to be created, it guides the large model to generate a compliant variable name and expression. During this process, the system prioritizes searching and reusing rule nodes with the same or similar logic in the existing rule base. Finally, a complete rule structure is generated from the root node to the leaf nodes, providing review conclusions such as "pass" and "fail," and converted into a JSON format that the system can parse and execute.

[0036] Furthermore, the test release unit is responsible for quality verification and lifecycle management of the generated rule tree. The rule tree produced by the rule generation unit is first stored in the test library of the data layer. Subsequently, this unit automatically calls a set of historical report data with known correct audit results as test cases to test and verify the rule tree. The verification process checks whether the execution results of the rule tree are consistent with the expected results and evaluates its execution performance and stability. After the test passes, the rule tree will be visible to users on the management side, and the status will change to "Test passed". The rule administrator can conduct a final review and fine-tune it. After the user confirms that there are no errors, they can publish and activate it through the operation interface, and the rule will be officially incorporated into the rule set of the production environment for auditing.

[0037] Thus, through the collaborative work of the above four units, the automatic rule generation module realizes the automated and intelligent conversion from text rules to executable code, significantly improving the system's speed of adaptation to new regulations and standards and its ability to solidify knowledge.

[0038] Furthermore, as an optional embodiment of the present invention, the special equipment report intelligent review system further includes: a rule set management module, used to establish the correspondence between the report type of the inspection report template and the rule set, and to manage the state transition of the rule set and the dynamic updating and version management of the rule set. The state of the rule set includes test state, online state and offline state.

[0039] Specifically, the core function of the rule set management module is to establish a strict correspondence between report types and rule sets in the inspection report templates. In the system, one report type corresponds to one and only one active rule set. This one-to-one binding ensures that when the system identifies a report type to be reviewed, it can accurately call the unique and valid rule set associated with it, avoiding rule conflicts or misuse. The module manages the entire lifecycle of rule sets, which mainly include three states: testing, online, and offline. Testing State: When an administrator creates a new rule set or modifies an existing one, the rule set is in this state. In this state, the rule set can be used for functional verification by internal testers or specific users. Online State: Only one rule set can be in this state, i.e., the currently active rule set. When a rule set in the testing state passes all verifications, the administrator can publish it as "online." When a new rule set is activated, the system automatically converts the original online rule set to "offline." This operation ensures the smoothness and consistency of the audit baseline switch, avoiding the chaos of two sets of rules being effective simultaneously. Offline Status: The rule set is in a dormant state and no longer participates in the review of any reports. Offline rule sets are retained in the system database and can be used for tracing and reviewing historical report review conclusions or as a reference template for designing new rules.

[0040] In addition, the rule set management module provides dynamic update and version management capabilities. When regulations change or audit logic needs optimization, administrators can create new versions based on existing online rule sets, adding, modifying, or deleting rules. Each modification generates a new rule set version and records the version number, the modifier, the modification time, and a change log. This version-based management ensures that every iteration of the rules is traceable and supports quick rollback to any historical version, enhancing the system's maintainability and agile response to changes in business requirements.

[0041] Corresponding to the intelligent review system for special equipment reports provided in the above embodiments, based on the same technical concept, this embodiment of the invention also provides an intelligent review method for special equipment reports, such as... Figure 2 As shown, Figure 2 A flowchart illustrating an intelligent review method for special equipment reports provided in this embodiment of the invention includes:

[0042] Step S201: Access the report data of the inspection report template of the intelligent special inspection management platform through a browser plugin.

[0043] Specifically, in this embodiment of the invention, the user terminal is pre-installed and embedded in the "Smart Special Inspection Management Platform" web application used daily by inspection personnel, in the form of a browser plugin such as ChromeExtension. When an inspection personnel completes report preparation or opens a report awaiting review on the platform, the plugin provides "Smart Review" or "Extract Data" buttons in the sidebar or toolbar of the page. Clicking this button triggers the data extraction process.

[0044] Furthermore, once the plugin is triggered, it works collaboratively in the following two main ways to retrieve data from the currently opened report page:

[0045] Front-end DOM parsing: The plugin injects scripts into the document object model of the current page, locates the HTML elements that display key data in the report, such as input boxes and table cells, and reads their value or textContent attribute values.

[0046] Network Request Interception: The plugin monitors network communication between the browser and the server. When a user interacts with the platform, such as clicking "View Report," it triggers an API call from the browser requesting report data from the server. The plugin intercepts these API responses and directly obtains the raw, structured data returned.

[0047] Furthermore, the extracted raw data will be encapsulated into a standardized JSON format by the plugin. For example, for elevator door gap data, the plugin will generate a JSON object in the form of {"Door Gap":6,"Rated Speed":2.5,...}. The plugin will ensure that the keys in the JSON are consistent with the report template parameter names defined in the system data layer.

[0048] Finally, the encapsulated JSON data is transmitted via the secure HTTPS protocol as a request payload to the dedicated data receiving interface provided by the service layer, transmitting the report data to the backend service of this intelligent auditing system for further processing.

[0049] Step S202: Import the inspection report template and extract parameters to generate a template, and create variables based on the parameters in the parameter generation template.

[0050] Specifically, the system receives the raw report data (in JSON format) transmitted via a browser plugin in step S201. The system parses the structure of the JSON data and, based on its overall structural characteristics, automatically creates or matches a corresponding report template.

[0051] Furthermore, after the template is created, the system extracts all the keys from the JSON object as the basic parameters for this type of report. For example, it extracts three parameters from the JSON above: "door gap," "rated speed," and "inspection date." Simultaneously, the system automatically identifies the data type of each parameter value. These parameters and their metadata establish a one-to-many relationship with the report template and are stored, forming the parameter pool for this type of report.

[0052] Furthermore, variables are the basic logical units for constructing audit rules; they are Boolean (true / false) values ​​obtained after performing operations or judgments on parameters. The system supports users manually creating variables based on parameters to endow data with logical judgment capabilities. Creation method: Users select target parameters (such as "door gap") from the parameter pool through the graphical interface on the management terminal, and then apply logical or arithmetic operations to them. In addition, the system supports creating more complex variables by processing parameters through built-in functions.

[0053] Step S203: Manually configure the first rule or automatically generate the second rule based on the large model, and create and activate a rule set based on the first and second rules.

[0054] Specifically, users manually configure audit rules (i.e., the first rule) through the rule management interface on the management terminal, based on the parameters and variables created in step S202. The semantic rules are then converted into the second rule based on the large model capability.

[0055] Furthermore, as an optional embodiment of the present invention, manually configuring the first rule or automatically generating the second rule based on a large model includes: receiving external input, the external input including a semantic rule description; configuring the first rule according to the semantic rule description; converting the semantic rule into a second rule based on the capabilities of the large model; nesting the first rule and the second rule using a rule tree structure to obtain a rule tree; and performing rule testing and verification on the rule tree.

[0056] Specifically, the system receives external input from users through a web interface on the management side. There are two input methods: Structured input (for manually configuring the first rule): Users drag and drop or select from a preset list of parameters and variables in the graphical interface, configuring logical operators and thresholds to define rule conditions (IF section) and results (THEN / ELSE section) in a structured manner. Semantic input (for automatically generating the second rule): Users directly enter or paste natural language descriptions of regulatory clauses or review requirements into a text input box. This part of the input is called the semantic rule description.

[0057] Furthermore, after reading and understanding the semantic rule description, the auditing expert or administrator can manually establish logical relationships in the rule configuration interface using the variables created in step S202. For example, configure a nested rule: IF if the rated speed is greater than 2.5 (true), THEN check if the horizontal distance of the sub-rule is greater than 35; if the sub-rule is also true, then report an error.

[0058] Furthermore, the semantic rule description is sent to the large model, requiring it to identify and extract key parameters and verify their existence in the parameter pool of the current report template. Variable validation and creation: Check whether existing related variables can satisfy the rule logic. If not, guide the large model to generate new variable expressions. Rule construction: The large model automatically generates a complete and executable rule based on the semantic logic.

[0059] Furthermore, both the manually configured first rule and the automatically generated second rule will ultimately be organized into a hierarchical structure like a rule tree. During implementation, the system supports adding new rule conditions to the THEN or ELSE branches of a rule, forming multi-level nesting. For example, the above rule can be expanded to: the root node checks if the "equipment type is an elevator"; if so, it then proceeds to the child nodes to check if the "rated speed is greater than 2.5"; if satisfied, it finally checks if the "horizontal distance exceeds the standard".

[0060] Furthermore, newly generated or modified rule trees will not take effect immediately; they must first undergo a testing and verification process. During implementation, the system provides a testing function, allowing users to select a batch of historical report data (whose audit conclusions are known) as test cases to perform automated batch testing of the rule trees. The system compares the automated audit results of the rule trees with the known correct conclusions to verify their accuracy and consistency. After the test passes, the rule tree's status is updated to "Test Passed," and the administrator can review and publish it. If the test fails, the process returns to the above steps for adjustments until verification is passed.

[0061] Furthermore, a rule set is a collection of all rules used to review a specific type of report. Users select a target report template on the management interface and add the configured first rule and the tested, automatically generated second rule to the same rule set. The system ensures that only one rule set for a particular report type can be active (online) at any given time. When the administrator sets the new rule set to "online," the system automatically sets the previously online rule set to "offline," achieving a smooth and seamless switch between review rules and ensuring the uniqueness and consistency of the review criteria.

[0062] Step S204: Identify the report type of the inspection report template and match it with the rule set, execute the audit rules, and return the audit results, which include pass items, critical items, and general items.

[0063] Specifically, when a user triggers "one-click intelligent review" through a browser plugin, the system first obtains the source data of the current report. The business processing module in the service layer analyzes this data, automatically identifying the specific report type by recognizing the embedded report template ID or parsing the characteristics of the data structure. Subsequently, the system queries the rule set management module in the data layer, matching the rule set that is bound to the report type and is currently in an "online" state, based on the principle of "one type of report corresponds to one active rule set".

[0064] Furthermore, after matching a rule set, the rule execution module loads the first and second rules contained in that rule set and typically traverses and executes them according to the rule tree structure. Rule execution employs a dual-mode strategy to balance efficiency and flexibility: Rule Engine Execution Mode: For deterministic, quantifiable, rigid rules, the module calls the built-in high-performance expression calculation engine to directly perform logical judgments on the report data and rule conditions. This mode is extremely fast and produces accurate results. Large Model Execution Mode: For flexible rules requiring semantic understanding or complex contextual judgments, the module integrates the structured description of the rule with the report data into a prompt word, which is then sent to the large language model for inference and judgment.

[0065] Furthermore, each rule execution generates a preliminary conclusion. The rule execution module aggregates the execution results of all rules and automatically categorizes them according to the pre-set severity level of the rule: Pass: Items that fully comply with regulations and standards. Critical: Items that seriously violate mandatory safety specifications and may lead to significant safety hazards. These issues must be modified until passing. General: Items that do not meet specifications or best practices but have low severity. These issues are recommended for modification. Finally, the system generates a structured audit conclusion, clearly displayed to the user through a browser plugin interface. The results are usually presented in list form, clearly listing the specific content, location, and relevant regulatory basis of all "critical" and "general" items, and summarizing the number of "pass" items, thus efficiently guiding inspectors in revising reports.

[0066] Furthermore, as an optional embodiment of the present invention, identifying the report type of the inspection report template and matching it with the rule set, executing the audit rules, and returning the audit results also includes: using a large model to audit the dynamic rules when the inspection report template matches a dynamic rule in the rule set; using a rule engine to calculate and audit the strict rules when the inspection report template matches a strict rule in the rule set; generating audit conclusions and modification suggestions; and displaying the audit results in real time through the user terminal.

[0067] Specifically, when the system traverses the rule set and identifies a rule as a dynamic rule (typically requiring semantic understanding, contextual correlation, or flexible judgment), the rule execution module initiates the large model review process. Implementation: The module integrates the structured description of this dynamic rule with relevant data fields from the current inspection report to construct a context-rich prompt. The large model, leveraging its powerful natural language understanding capabilities, reasons and judges the rule, returning a structured review result.

[0068] Furthermore, when the system identifies a rule as a strict rule, typically referring to a clear and quantifiable technical specification, the rule execution module will invoke the built-in rule engine for computational review. The rule engine will directly parse the logical expression of the strict rule (e.g., door gap <= 6). Then, the engine extracts the values ​​of the corresponding parameters from the current report data, substitutes them into the expression for rapid calculation, and obtains a Boolean value (true / false) result.

[0069] Furthermore, the system automatically categorizes the review results of each rule into approved, critical, and general categories based on predefined severity levels. For example, if the strict rule "door gap > 6" is triggered, it will be marked as a "critical" rule; if the dynamic rule "vague description" is triggered, it may be marked as a "general" rule. Modification suggestions are generated: For items that fail to pass, the system generates specific modification suggestions.

[0070] Furthermore, the final review conclusions and suggested modifications are displayed to the inspectors in real time via the user interface. As soon as the review results are generated, the plugin interface pops up or is embedded next to the report page in a clear and visual manner. Users can intuitively see all the issues and their detailed suggestions without switching pages, enabling them to quickly locate and modify the report content, achieving a real-time and seamless review feedback experience.

[0071] Furthermore, as an optional embodiment of the present invention, the method further includes: collecting audit result data; performing statistical analysis on the problem distribution of the audit result data; and optimizing the rule content and template structure based on the analysis results of the audit result data.

[0072] Through the technical solutions disclosed in this invention, the server platform and network environment at the infrastructure layer provide stable computing power. The data layer caches business data through an in-memory database and, in conjunction with the structured storage of a relational database, enables rapid retrieval and parsing of inspection report data. The service layer's audit rule engine replaces manual verification, automatically executing audit rules and returning conclusions, solving the technical problem of low efficiency in manual auditing and efficiently handling the auditing needs of batch testing reports. The system transforms legal norms and inspection standards into standardized rule sets stored in the data layer and uniformly executes audit logic through the rule engine, avoiding misunderstandings of standards due to differences in technical level and experience during manual auditing. This reduces omissions and misjudgments at the source, ensuring the objectivity and consistency of audit results, thereby improving the accuracy of special equipment inspection report audits. The management end supports rule configuration and system management, flexibly adapting to updates in inspection standards and changes in business needs, thus improving the flexibility of special equipment inspection reports. The user end is integrated into the existing business system as a browser plugin, eliminating the need to reconstruct the original platform, reducing deployment costs, and facilitating quick operation for inspection and audit personnel, further improving the efficiency of special equipment inspection report audits.

[0073] Corresponding to the intelligent review method for special equipment reports provided in the above embodiments, based on the same technical concept, this embodiment of the invention also provides an electronic device for executing the above-described intelligent review method for special equipment reports. Figure 3 This is a schematic diagram of the structure of an electronic device provided in one embodiment of the present invention, as shown below. Figure 3 As shown. Electronic devices can vary considerably due to differences in configuration or performance, and may include one or more processors 301 and memories 302. The memory 302 stores computer programs that can run on the processor 301, and the processor 301 executes the programs stored in the memory 302 to achieve the above. Figure 1 The various steps in the method embodiment are described. The memory 302 can be temporary or persistent storage. The application stored in the memory 302 may include one or more modules (not shown), each module may include a series of computer-executable instructions for the electronic device.

[0074] Furthermore, the processor 301 may be configured to communicate with the memory 302 and execute a series of computer-executable instructions stored in the memory 302 on the electronic device. The electronic device may also include one or more power supplies 303, one or more wired or wireless network interfaces 304, one or more input / output interfaces 305, and one or more keyboards 306.

[0075] Specifically, in this embodiment, the electronic device includes a processor, a communication interface, a memory, and a communication bus; wherein, the processor, the communication interface, and the memory communicate with each other via the bus; the memory is used to store computer programs; and the processor is used to execute the programs stored in the memory to achieve the above. Figure 1 The various steps in the method embodiments are the same as those in the above method embodiments, and have the same beneficial effects. To avoid repetition, the embodiments of the present invention will not be described again here.

[0076] It should be noted that the electronic device provided in this embodiment of the invention and the intelligent review method for special equipment reports provided in this embodiment of the invention are based on the same application concept. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned intelligent review method for special equipment reports, and has the same or similar beneficial effects. Repeated parts will not be described again.

[0077] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0078] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A special equipment report intelligent review system, characterized in that, The intelligent review system for special equipment reports includes: an infrastructure layer, a data layer, a service layer, and a presentation layer; The infrastructure layer deploys a server platform and operating system to provide hardware resources, operating system, middleware and network environment for system operation. The data layer is connected to the infrastructure layer, uses a relational database to store business data and uses an in-memory database for data caching. The business data includes inspection report templates, parameter variables and rule sets. The service layer is connected to the data layer and is used to carry the audit rule engine and business processing module to parse the inspection report data, execute the audit rules and return the audit conclusion. The presentation layer is connected to the service layer and includes a management terminal and a user terminal. The management terminal provides rule configuration and system management functions, and the user terminal is integrated into the existing business system in the form of a browser plugin.

2. The intelligent review system for special equipment reports according to claim 1, characterized in that, The infrastructure layer also includes: Containerized deployment modules enable rapid deployment and elastic scaling of service modules.

3. The intelligent review system for special equipment reports according to claim 1, characterized in that, The data layer also includes: The template management module is used to establish a one-to-one correspondence between the test report templates and report types; The parameter management module is used to establish a one-to-many relationship between the test report template and the parameters.

4. The intelligent review system for special equipment reports according to claim 1, characterized in that, The service layer's audit rule engine includes: The manual rule generation module creates manual configuration rules based on parameter variables. These manual configuration rules support logical and arithmetic operations. The automatic rule generation module, based on large model capabilities, converts semantic rules into executable rules. The audit rules include manually configured rules and executable rules. The rule execution module is used to execute the audit rules and generate audit conclusions.

5. The intelligent review system for special equipment reports according to claim 4, characterized in that, The automatic rule generation module also includes: The parameter detection unit is used to detect whether the parameters in the semantic rules exist in the template parameter pool; the variable validation unit is used to check whether the variables associated with the parameters in the semantic rules meet the requirements of the semantic rules. The rule generation unit uses the large model to create variables and the executable rules, and generates a rule tree; The test release unit is used to store the rule tree and perform test verification.

6. The intelligent review system for special equipment reports according to claim 1, characterized in that, The intelligent review system for special equipment reports also includes: The rule set management module is used to establish the correspondence between the report type of the inspection report template and the rule set, and to manage the status transition of the rule set and the dynamic updates and version management of the rule set. The status of the rule set includes test status, online status and offline status.

7. A method for intelligent review of special equipment reports, characterized in that, include: Report data from inspection report templates accessed through a browser plugin to the intelligent special inspection management platform; Import the test report template and extract parameters to generate a template, and create variables based on the parameters in the generated template; The first rule can be manually configured or the second rule can be automatically generated based on the large model. A rule set can be created and activated based on the first rule and the second rule. Identify the report type of the inspection report template and match it with the rule set, execute the audit rules, and return the audit results, which include pass items, critical items, and general items.

8. The method according to claim 7, characterized in that, The manual configuration of the first rule or the automatic generation of the second rule based on the large model includes: Receive external input, which includes a semantic rule description; Configure the first rule according to the semantic rule description. Based on the capabilities of large models, semantic rules are converted into second rules; The first rule and the second rule are nested using a rule tree structure to obtain a rule tree; The rule tree is then subjected to rule testing and verification.

9. The method according to claim 7, characterized in that, The process of identifying the report type of the inspection report template, matching it with the rule set, executing the audit rules, and returning the audit result also includes: If the inspection report template matches a dynamic rule in the rule set, the dynamic rule is reviewed using a large model. If the inspection report template matches a strict rule in the rule set, the strict rule is calculated and reviewed using a rule engine. Generate review conclusions and modification suggestions; The review results are displayed in real time on the user's device.

10. The method according to claim 7, characterized in that, The method further includes: Collect audit result data; The audit results data were statistically analyzed to determine the distribution of problems. The content of the rules and the structure of the templates are optimized based on the analysis results of the audit results data.

Images