Automatic correction method for special equipment archives based on multi-modal analysis

Abstract

A kind of special equipment archive automatic proofreading method based on multi-modal analysis, the method comprises: obtaining engineering project archive, and the engineering project archive is analyzed to construct system candidate unit. The equipment in system candidate unit is merged according to structure, to determine system boundary, and the equipment archive field is mapped to the subunit in system candidate unit. Constraint rule is introduced in system candidate unit, and the equipment archive field is proofread according to constraint rule, to obtain field tolerance proofreading result. The field tolerance proofreading result is compared with system test record for consistency, to generate consistency determination conclusion. Based on consistency determination conclusion, the allowed difference between equipment archive field and system level compliance requirement is marked and shielded. The present application avoids the equipment false exception phenomenon caused by the mutual influence of different systems, eliminates the influence for the determination of equipment false exception in single system, and improves the accuracy of the determination of equipment false exception in single system.

Description

Technical Field

[0001] This invention relates to the field of equipment testing technology, and in particular to an automatic verification method for special equipment files based on multimodal analysis. Background Technology

[0002] Industrial equipment condition monitoring and management methods are an important component of the development of the Industrial Internet. They leverage Internet and IoT technologies to enable real-time data interaction and information sharing between devices and between devices and systems, thereby achieving industrial digital transformation. The research and development of industrial equipment condition monitoring and management methods aims to improve production efficiency and product quality by monitoring equipment operating status in real time, promptly identifying potential problems, and taking preventative measures. However, when monitoring the status of industrial equipment, external factors such as the environment in which the equipment operates can lead to a certain rate of misjudgment in the monitored data.

[0003] Currently, in existing equipment testing processes, multiple interconnected and related devices typically exist across different systems. If a device malfunctions, the system automatically reports the malfunctioning device to the front end to notify staff for attention and handling. However, it often happens that during manual verification, devices that trigger alarms are not actually malfunctioning; this is known as "false alarms." To avoid this, it is necessary to propose a method to detect "false alarms" and mark and block such devices, thereby saving manpower and resources. Summary of the Invention

[0004] Therefore, it is necessary to propose an automatic verification method for special equipment files based on multimodal analysis to address the aforementioned technical problems.

[0005] The present invention adopts the following technical solution.

[0006] The first aspect of this invention discloses an automatic verification method for special equipment files based on multimodal analysis, the method comprising:

[0007] Obtain project files and parse them to construct system candidate units;

[0008] The devices in the system candidate units are structurally merged to determine the system boundary, and the device file field is mapped to the sub-units in the system candidate units.

[0009] Constraint rules are introduced into the system candidate units, and the tolerance of the equipment file fields is checked according to the constraint rules to obtain the field tolerance check results;

[0010] The field tolerance calibration results are compared with the system inspection records to generate a consistency judgment conclusion.

[0011] Based on the consistency determination conclusion, the permissible differences between the device file fields and the system-level compliance requirements are marked and masked.

[0012] Furthermore, the step of acquiring project files and parsing the project files to construct system candidate units includes:

[0013] Define the system correlation degree of the project files, and perform system correlation screening on the project files according to the system correlation degree to obtain candidate files with a system correlation degree not lower than a set threshold;

[0014] Field elements are extracted from the candidate files, and confidence levels are set for the extracted field elements to construct a structured element table based on the field elements and their corresponding confidence levels.

[0015] Furthermore, the step of acquiring project files and parsing the project files to construct system candidate units also includes:

[0016] Based on the structured element table, determine the various representations of each device in different engineering project files, and calculate the similarity between different representations, so as to merge the various representations with a similarity of not less than a set threshold into the primary key of the corresponding device.

[0017] The candidate files associated with the primary key are linked to obtain the element primary key table, and the connection relationship between different devices is determined. The connection relationship between different devices includes the primary key of the connection starting device, the primary key of the connection ending device, and the connection type.

[0018] Using the system name as the aggregation clue, and combining the element primary key table and the connection relationships between different devices, system candidate units under the system name are constructed.

[0019] Furthermore, the step of structurally merging the devices in the system candidate units to determine the system boundary and mapping the device file field to the sub-units in the system candidate units includes:

[0020] Calculate the device overlap between different system candidate units, and merge the different system candidate units into the same system candidate unit when the overlap is not lower than a set threshold;

[0021] The connection relationships of different devices in the candidate units of the system are verified, and the boundary consistency coefficient is calculated based on the number of device connection relationships within the system and the number of device connection relationships between the system and outside the system, so as to prune the primary keys of devices with a boundary consistency coefficient lower than a set threshold.

[0022] Furthermore, the step of structurally merging the devices in the system candidate units to determine the system boundary and mapping the device file field to the sub-units in the system candidate units also includes:

[0023] Calculate the connectivity of each device primary key in the system candidate units with other devices within the system, and determine the type of the device primary key based on the range of connectivity.

[0024] Based on the device primary key corresponding to the device file field, and combined with the type of the device primary key and the system candidate unit, a mapping relationship between the device file field and the system candidate unit is generated.

[0025] Furthermore, the step of introducing constraint rules into the system candidate units and performing tolerance calibration on the device file fields according to the constraint rules to obtain field tolerance calibration results includes:

[0026] Based on the mapping relationship between the device file fields and the system candidate units, a suitable system constraint template is selected, and the system constraint template is instantiated using the actual fields of the current system to obtain a system constraint instance;

[0027] The initial tolerance threshold is determined based on the system constraint instance, and the initial tolerance threshold is then corrected in layers according to the equipment type weight and system mechanism residual to obtain the field tolerance threshold.

[0028] Calculate the relative deviation of each equipment file field, compare the relative deviation with the field tolerance threshold, and make a secondary judgment to aggregate the relative deviation comparison results of all equipment file fields and output the field tolerance calibration result.

[0029] Furthermore, the step of comparing the field tolerance calibration results with the system inspection records to generate a consistency judgment conclusion includes:

[0030] Extract filing elements from the system filing files corresponding to the system candidate units, and convert the filing elements into a structured filing element set. Align the structured filing element set with the equipment file fields and the system candidate units to obtain a system filing structure mapping table.

[0031] The inspection elements are extracted from the system inspection records, and the corresponding inspection coverage equipment, equipment file fields and system candidate units in the inspection elements are compared to obtain the system inspection consistency results.

[0032] Based on the system filing time, system inspection time, and system structure confirmation time, calculate the system time consistency deviation and inspection validity deviation, and determine the system time consistency verification result based on the system time consistency deviation and inspection validity deviation.

[0033] The consistency determination conclusion is a result obtained by comprehensively evaluating the system filing structure mapping table, system inspection consistency results, system time consistency verification results, and field tolerance verification results.

[0034] Furthermore, based on the consistency determination conclusion, the allowed differences between the device file fields and system-level compliance requirements are marked and masked, including:

[0035] The system-level non-compliance index is determined based on the field tolerance verification results, and a system verification status quantity is generated in combination with the consistency judgment conclusion. The system verification status quantity is the difference between the score corresponding to the consistency judgment conclusion and the system-level non-compliance index.

[0036] When the system calibration status quantity is in the specified state, the system impact degree of the corresponding device file field is determined, and the corresponding device is marked as abnormal based on the range of the system impact degree.

[0037] The second aspect of this invention discloses an automatic verification device for special equipment files based on multimodal analysis, used to implement the automatic verification method for special equipment files based on multimodal analysis as described in any one of the first aspects, the device comprising:

[0038] The file parsing module is used to obtain engineering project files and parse the engineering project files to construct system candidate units;

[0039] The field mapping module is used to perform structural merging of devices in the system candidate units to determine the system boundary, and to map the device file field to the sub-units in the system candidate units.

[0040] The tolerance calibration module is used to introduce constraint rules into the system candidate units and perform tolerance calibration on the equipment file fields according to the constraint rules to obtain the field tolerance calibration result.

[0041] The consistency determination module is used to compare the field tolerance calibration results with the system inspection records to generate a consistency determination conclusion.

[0042] An anomaly marking and masking module is used to mark and mask the permissible differences between the device file fields and system-level compliance requirements based on the consistency determination conclusion.

[0043] A third aspect of the present invention discloses a terminal, including a processor and a storage medium;

[0044] The storage medium is used to store instructions;

[0045] The processor is configured to operate according to the instructions to perform the steps of the method described in the first aspect.

[0046] A fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0047] The present invention has the following advantages:

[0048] (1) This invention analyzes the files of multimodal special equipment involved in the same engineering project, extracts information such as equipment number, equipment type, connection relationship description, system name, and system schematic diagram annotation, and constructs a set of candidate system units based on the association description between the equipment. Then, the equipment members in the set are structurally merged to clarify the system boundary and form a system-level structural mapping relationship, mapping a single equipment file to a component unit within the system. This effectively avoids the occurrence of false equipment anomalies caused by mutual influence between different systems, and at the same time eliminates the influence on the determination of false equipment anomalies within a single system, thereby improving the accuracy of the determination of false equipment anomalies within a single system.

[0049] (2) Based on system-level structure mapping, this invention introduces system-level constraint rules to perform tolerance verification on single device fields. It only determines anomalies when field deviations exceed the system's allowable range, thereby distinguishing between "reasonable differences within the system" and "system-level non-compliance." Subsequently, the system field tolerance verification results are compared as a whole with the system filing information and system inspection records to determine whether the system-level filing objects, inspection objects, and system structure are consistent, forming a system-level consistency judgment conclusion. This further improves the accuracy of determining false anomalies of equipment within a single system and can effectively shield the impact of false anomaly equipment on industrial projects. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating the automatic verification method for special equipment files based on multimodal analysis provided by the present invention.

[0052] Figure 2 This is a schematic diagram of the structure of the automatic verification device for special equipment files based on multimodal analysis provided by the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] like Figure 1 As shown, in one embodiment, an automatic verification method for special equipment files based on multimodal analysis includes the following steps:

[0055] Step S110: Obtain the project file and parse the project file to construct system candidate units.

[0056] In some embodiments, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes the following steps in step S110:

[0057] Step S111: Define the system correlation degree of the project files, and perform system correlation screening on the project files according to the system correlation degree to obtain candidate files with a system correlation degree not lower than the set threshold.

[0058] Step S112: Extract field features from candidate files and set confidence levels for the extracted field features to construct a structured feature table based on the field features and their corresponding confidence levels.

[0059] In some embodiments, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention further includes the following steps in step S110:

[0060] Step S113: Determine the multiple representations of each device in different project files based on the structured element table, and calculate the similarity between different representations, so as to merge multiple representations with a similarity of not less than a set threshold into the primary key of the corresponding device.

[0061] Step S114: Attach the candidate files associated with the primary key to obtain the element primary key table, and determine the connection relationship between different devices. The connection relationship between different devices includes the primary key of the connection starting device, the primary key of the connection ending device, and the connection type.

[0062] Step S115: Using the system name as the aggregation clue, and combining the element primary key table and the connection relationship between different devices, construct the system candidate unit under the system name.

[0063] In a specific embodiment, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes steps 1 to 5:

[0064] Step 1: Analysis of related elements in multimodal systems and construction of candidate units for the system.

[0065] The process involves parsing the files of multimodal special equipment involved in the same engineering project, extracting information such as equipment number, equipment type, connection relationship description, system name, and system schematic diagram annotations, and constructing a set of candidate system units based on the relationship descriptions between the equipment. This includes the following sub-steps:

[0066] Sub-step 1.1: Construct a candidate file list.

[0067] Specifically, the process begins by acquiring a collection of project files, consisting of several files. Each file includes its format and content, categorized into four types: text, tables, scanned images, and photographs. Next, the project file collection is filtered based on "system association clues." These clues include system name keywords, system-level descriptions such as linkage / integrated / group control, and diagram / flowchart titles. A system association score is assigned to each file, and a candidate file list is generated using a threshold-based selection process.

[0068] The expression for the system correlation score is as follows:

[0069] ;

[0070] In the formula, This represents the system association score corresponding to the k-th file; This represents the number of system name keywords that were matched in the k-th file, with a value ranging from 0 to 50. Indicates the number of connective words involved, such as "connection, parallel, series, linkage, branch, loop", with a value range of 0 to 80; This indicates the number of keywords associated with the matched drawings or diagrams, such as "system diagram, flowchart, loop diagram, P&ID, schematic diagram", with a value range of 0 to 30. This indicates the number of generic descriptive terms that are independent of the system, such as "description, overview, background", with a value ranging from 0 to 200. , , All are weighting coefficients, with values ​​ranging from 0.3 to 0.5, 0.3 to 0.5, and 0.1 to 0.4, respectively, and the sum of the three is 1.

[0071] Sub-step 1.2 generates a structured feature table.

[0072] Specifically, for each file in the candidate file list, elements are extracted to form a unified set of fields: equipment number, equipment type, system name, connection relationship description, and schematic diagram annotation. To ensure that the same meaning of a field can be calculated across different media, a confidence level is defined for each element record, and the element and its confidence level are written together into a structured element table.

[0073] The expression for the confidence level is as follows:

[0074] ;

[0075] In the formula, This represents the number of times the element appears repeatedly within the same file, with a value ranging from 0 to 10. This indicates the number of times the element is cross-referenced across different files, with a value ranging from 0 to 20. This indicates the number of times the element has ambiguous occurrences, such as the same number corresponding to multiple types or the same system name appearing in multiple ways. The value range is 0 to 15. The value ranges from 0 to 20, but in engineering practice it usually falls between 0 and 5. This applies to elements extracted from scanned images and photographs. Count by "multiple annotations on the same page" Count by “text-image cross-reference or table-image cross-reference”.

[0076] Sub-step 1.3: Construct the primary key table of equipment elements and the set of connection relationships.

[0077] Specifically, the elements in the structured element table are standardized across different media, establishing two intermediate products: device primary keys and connection relationships. The device primary key table is used to merge multiple ways of writing the same device in different files into a single primary key; the connection relationship set is used to record connection pairs between devices. In this process, the connection of any two device number elements is first calculated. The similarity score is used to determine whether the keys are merged into the same device primary key.

[0078] The expression for similarity calculation is as follows:

[0079] ;

[0080] In the formula, This represents the length of the longest common substring of the two numbered strings, with a value ranging from 0 to 40. This represents the maximum length of the two numbered strings, ranging from 1 to 40. This is a smoothing constant, ranging from 0.5 to 2. If... If the value is greater than or equal to a set threshold (ranging from 0.80 to 0.92), the results are merged. After merging, a primary key is formed, and the associated device type and system name candidate set are also attached.

[0081] Next, a set of connection relationships is generated, and each connection relationship is described by extracting the primary key of the starting device, the primary key of the ending device, and the connection type. The connection type can be one or more of the following: "series, parallel, linkage, branch, loop, interlock, shared medium pipeline".

[0082] Sub-step 1.4 filters out fake systems that are "nominal systems but have no engineering connections".

[0083] Specifically, system candidate units are constructed using system names as the aggregation thread and connection relationships as a supplement. For each system name candidate value, a set of device primary keys that appear in the device element primary key table for that system name candidate value is collected. This set is then expanded within the connection relationship set by adding device primary keys that have direct connections to the device primary key set, forming an extended member set. The number of members and connection density of the system candidate units are defined to filter out pseudo-systems that are "nominal systems but have no engineering connections".

[0084] Among them, connection density The expression is:

[0085] ;

[0086] In the formula, This represents the number of device primary keys for candidate units in the system, with a value ranging from 2 to 500. This indicates the number of valid connections within the candidate unit, ranging from 1 to 2000. If... Greater than or equal to a set threshold (ranging from 0.3 to 2.0) and If the number of device primary keys is greater than or equal to the lower limit (ranging from 2 to 5), the candidate unit is retained and added to the system candidate unit set. For cases where there is no explicit system name but the connection relationships are dense, anonymous system candidate units are allowed to be generated based on "connection density priority". The names of anonymous candidate units can be generated by adding a serial number to the project number.

[0087] Step S120: Perform structural merging of devices in the system candidate units to determine the system boundary, and map the device file field to the sub-units in the system candidate units.

[0088] In some embodiments, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes the following steps in step S120:

[0089] Step S121: Calculate the device overlap between different system candidate units, and merge different system candidate units into the same system candidate unit when the overlap is not lower than a set threshold.

[0090] Step S122: Verify the connection relationships of different devices in the candidate units of the system, and calculate the boundary consistency coefficient based on the number of device connection relationships within the system and the number of device connection relationships between the system and outside the system, so as to prune the primary keys of devices with boundary consistency coefficients lower than the set threshold.

[0091] In some embodiments, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention further includes the following steps in step S120:

[0092] Step S123: Calculate the connectivity of each device primary key in the system candidate unit with other devices in the system, and determine the type of device primary key based on the range of connectivity.

[0093] Step S124: Based on the device primary key corresponding to the device file field, and combined with the type of the device primary key and the system candidate unit, generate the mapping relationship between the device file field and the system candidate unit.

[0094] In a specific embodiment, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes step 2, system-level structural consistency merging and unit mapping. Based on the system candidate unit set obtained in step 1, the equipment members are structurally merged to clarify the system boundaries and form a system-level structural mapping relationship, mapping a single equipment file to a constituent unit within the system. This includes the following sub-steps:

[0095] Sub-step 2.1: Calculation and merging determination of the overlap of candidate units in the system.

[0096] Specifically, within the system candidate unit set, there may be multiple candidate units with highly overlapping device members, requiring system-level merging to be performed first. For any two system candidate units, the overlap degree of their device members is calculated to determine whether they should be merged into the same system unit. The expression is:

[0097] ;

[0098] In the formula, Candidate units for the system and system candidate units The degree of overlap between equipment members; The number of device primary keys shared by both, ranging from 0 to the lower limit of the number of primary keys for both. , They are the respective system candidate units and system candidate units The number of device primary keys, ranging from 2 to 500.

[0099] When the above-mentioned equipment members overlap If the value exceeds a set threshold (ranging from 0.6 to 0.85), the system candidate unit will be... and system candidate units When merged into the same system unit, the set of devices in the merged system unit is taken as the union, and the set of connections is taken as the union.

[0100] Sub-step 2.2: System boundary integrity verification and boundary trimming.

[0101] Specifically, for each system unit, the closure degree of its internal device connection relationships is checked, and device primary keys that have been mistakenly incorporated only through weak associations are removed. The boundary consistency coefficient of the system unit is calculated, and the expression is:

[0102] ;

[0103] In the formula, For system unit Boundary consistency coefficient; This indicates the number of connections between devices within a system unit, ranging from 1 to 2000. This represents the number of connections between devices within the system unit and external devices, ranging from 1 to 2000; the constant 1 is used to avoid the denominator being zero.

[0104] When the above boundary consistency coefficient If the value is less than the set threshold (range 0.55-0.75), it indicates that the system boundary is unclear. It is necessary to prune the device primary keys corresponding to the number of connection relationships between devices in the system unit and devices outside the system. During pruning, priority should be given to removing devices that only have unidirectional or descriptive connections and do not appear in the system schematic diagram or system name association.

[0105] Sub-step 2.3: Identification and hierarchical division of internal system devices.

[0106] Specifically, in a system-level architecture, different devices play different engineering roles, such as core devices, subordinate devices, and auxiliary devices. This applies to system units. For each device primary key in the system, calculate its intra-system connectivity. To distinguish device roles, the expression is:

[0107] ;

[0108] In the formula, Device primary key Connectivity within the system; To represent the device primary key The number of connections with other devices in the system, ranging from 1 to 100; To represent system units The number of device primary keys in the data ranges from 2 to 500.

[0109] Then, based on the connectivity of the device within the system The device is categorized by size into three roles: when its connectivity is greater than or equal to a first threshold, it is classified as a core system device; when its connectivity is greater than or equal to a second threshold but less than the first threshold, it is classified as a subordinate system device; and when its connectivity is less than the second threshold, it is classified as an auxiliary system device. The first threshold ranges from 0.4 to 0.7, and the second threshold ranges from 0.15 to 0.3.

[0110] Sub-step 2.4: Generation of mapping from single device file to system structure.

[0111] Specifically, for each single device file, based on its corresponding device primary key, its system unit and device role are located in the system structure set, and a mapping record is generated. Each mapping record contains at least: system identifier, device primary key, device role, and index location within the system.

[0112] To verify the reliability of the mapping, a mapping consistency score is defined. The expression is:

[0113] ;

[0114] In the formula, Device primary key The number of times it is explicitly referenced in the system diagram or system description, ranging from 0 to 20; This represents the number of times the device is simultaneously hit across different system units, ranging from 0 to 5. The mapping consistency score... When the value exceeds the set threshold (range 0.6-1.5), the mapping is confirmed to be successful.

[0115] Step S130: Introduce constraint rules into the system candidate units, and perform tolerance calibration on the equipment file fields according to the constraint rules to obtain the field tolerance calibration results.

[0116] In some embodiments, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes the following steps in step S130:

[0117] Step S131: Based on the mapping relationship between the equipment file fields and the system candidate units, select the appropriate system constraint template, and instantiate the system constraint template using the actual fields of the current system to obtain a system constraint instance.

[0118] Step S132: Determine the initial tolerance threshold based on the system constraint instance, and perform hierarchical correction on the initial tolerance threshold according to the equipment type weight and system mechanism residual to obtain the field tolerance threshold.

[0119] Step S133: Calculate the relative deviation of each equipment file field, compare the relative deviation with the field tolerance threshold and make a secondary judgment, so as to aggregate the relative deviation comparison results of all equipment file fields and output the field tolerance calibration result.

[0120] In a specific embodiment, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes step 3, field tolerance verification under system constraints. Based on the system-level structure mapping formed in step 2, system-level constraint rules are introduced to perform tolerance verification on individual equipment fields. Anomalies are only identified when field deviations exceed the system's allowable range, thereby distinguishing between "reasonable differences within the system" and "system-level non-compliance." This includes the following sub-steps:

[0121] Sub-step 3.1: System-level constraint template selection and constraint parameter instantiation.

[0122] Specifically, first, based on the equipment type combination and connection type of the system unit, select the corresponding system constraint template, and then instantiate the template parameters using the actual fields of this system to obtain a set of computable system constraint instances. The system constraint template includes at least three categories: "conservation constraints," "consistency constraints," and "complementarity constraints." Taking a complete pressure system as an example, the conservation constraints use "flow conservation residual" as the system mechanism constraint; based on the inlet and outlet equipment sets within the system boundary, for any time period or any operating condition (if the file does not contain a time period, then the rated operating condition is used), the calculated residual is... The expression is:

[0123] ;

[0124] In the formula, This represents the sum of the rated flow rates of the inlet equipment, with units of m³ / h and a value range of 0.1-50000. This represents the sum of the rated flow rates of the outlet equipment, with a value range of 0.1-50000; the constant 1 is used to avoid the denominator being zero. and It is obtained by extracting the "rated flow rate" field from the single device file field set and aggregating it according to the system topology.

[0125] Subsequently, for linked elevator groups or combined lifting equipment, "matching capacity constraint residuals" can be used. For example, the total rated load of the group-controlled elevators is consistent with the coverage of the service area. The residuals are defined in the same way and instantiated into the system constraint instance set.

[0126] Sub-step 3.2: Generate field tolerance thresholds based on role weights and system residuals.

[0127] Specifically, for each system and each type of field (e.g., rated pressure, rated flow, media type, manufacturing license number, inspection validity period, etc.), a uniform tolerance threshold is generated within the system, allowing for tiered correction based on equipment role. In this process, a reference value for the field within the system is first calculated. This reference value uses a "role-weighted average" to ensure that core equipment contributes more to the reference value; that is, it is the weighted result of the sum of the products of all role weights and the corresponding equipment's field value.

[0128] Subsequently, system mechanism residuals are introduced to adaptively scale the tolerance. For conservation constraints, if the mechanism residuals... A large tolerance indicates a lack of consistency in the system at the rated parameter level, in which case the tolerance of key fields should be tightened. A small tolerance allows for slightly more reasonable variations within the system, ultimately leading to the tolerance threshold. The expression is:

[0129] ;

[0130] In the formula, The basic tolerance for each field is set, and the value range is determined by field type: pressure field: 0.02-0.08; flow field: 0.03-0.12; power field: 0.03-0.10; time field: 1-30 (unit: days). This is the scaling factor, with a value ranging from 0.2 to 2; This represents the residual value of the system constraint, ranging from 0 to 100. It prioritizes the residual value that is most relevant to the current field in the system constraint instance set (e.g., the mechanism residual corresponding to the flow field).

[0131] Sub-step 3.3: Field deviation quantification and system-level tolerance calibration judgment.

[0132] Specifically, for each field of each device within the system unit, its relative deviation is calculated and compared with the tolerance threshold. To avoid misjudgment caused by relying solely on the "difference magnitude," a "tolerance-normalized deviation score" is introduced, expressed as follows:

[0133] ;

[0134] In the formula, This indicates the value that the device can take in the current field; This represents a system reference value; if the current field is a time-related field, then... X_ref is represented as "number of days"; if the current field is not a numeric type, then... and Replace with "Category Matching Flag", assigning 0 for a match and 1 for no match, and substitute into the same formula.

[0135] The judgment rule adopts a two-level judgment: when the value of the device in the current field is less than or equal to 1, it is judged as "reasonable difference within the system"; when 1 < the value of the device in the current field ≤ the set threshold (value range 1.5-4), it is judged as "suspicious deviation"; when the value of the device in the current field > the set threshold (value range 1.5-4), it is judged as "abnormal deviation". For core devices, the lower limit of the set threshold (value range 1.5-4) is used; for auxiliary devices, the upper limit of the set threshold (value range 1.5-4) is used.

[0136] Sub-step 3.4: System-level non-compliant aggregation and fake anomaly masking output.

[0137] Specifically, the system aggregates all field verification results at the system unit level to form a system-level non-compliance index and outputs a "list of maskable single-device false anomalies". The system-level non-compliance index is obtained by weighted summation by role, including the sum of the number of fields judged as abnormal deviations in core devices, the sum of the number of abnormal fields in subordinate devices, the sum of the number of abnormal fields in auxiliary devices, and the number of devices of the three types of roles in the system.

[0138] When the aforementioned system-level non-compliance index is greater than or equal to the set threshold (range 0.05-0.20), the system is deemed to have a system-level non-compliance risk. Meanwhile, if a device only shows a "suspicious deviation" in auxiliary or non-critical fields, and the representative value of the system mechanism residual is less than the residual threshold that allows the system to be considered self-consistent (slightly tightening as the scale increases), then this deviation is classified as a "maskable false anomaly."

[0139] Step S140: Compare the field tolerance calibration results with the system inspection records to generate a consistency judgment conclusion.

[0140] In some embodiments, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes the following steps in step S140:

[0141] Step S141: Extract filing elements from the system filing archive corresponding to the system candidate unit, and convert the filing elements into a structured filing element set, so as to align the structured filing element set with the equipment archive fields and the system candidate unit to obtain the system filing structure mapping table.

[0142] Step S142: Extract inspection elements from the system inspection records, and compare the corresponding inspection coverage equipment and equipment file fields in the inspection elements with the system candidate units to obtain the system inspection consistency results.

[0143] Step S143: Based on the system filing time, system inspection time, and system structure confirmation time, calculate the system time consistency deviation and inspection validity deviation, and determine the system time consistency verification result based on the system time consistency deviation and inspection validity deviation.

[0144] The consistency determination conclusion is a result obtained by comprehensively evaluating the system filing structure mapping table, system inspection consistency results, system time consistency verification results, and field tolerance verification results.

[0145] In a specific embodiment, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes step 4, a comprehensive verification of system-level filing and inspection consistency. The system field tolerance verification results obtained in step 3 are compared as a whole with the system filing information and system inspection records to determine whether the system-level filing object, inspection object, and system structure are consistent, thus forming a system-level consistency judgment conclusion. This includes the following sub-steps:

[0146] Sub-step 4.1: Extract system filing objects and align them with the system structure.

[0147] Specifically, for each system unit, information about the registration object is extracted from the system registration archive, including the registration system name, registration equipment list, registration system type, and scope of application, and then transformed into a structured set of registration elements. This structured set of registration elements is then aligned with the system structure in the system-level structural mapping relationship, with a focus on comparing the consistency between the "set of equipment covered by the registration object" and the "actual set of equipment in the system structure." To quantify this consistency, a registration coverage rate metric is introduced. The expression is:

[0148] ;

[0149] In the formula, The number of primary keys of devices in the system architecture that are explicitly covered by the filing list; This represents the total number of device primary keys in the system architecture, with a value range of 2-500.

[0150] When the above filing coverage indicators If the value is greater than or equal to the set threshold (ranging from 0.7 to 0.95), the system registration object and system structure are considered consistent at the coverage level. If the registration object only describes the system without listing the equipment, then... Replaced by the number of devices whose system name matches the system type.

[0151] Sub-step 4.2: System verification of the consistency between the object being verified and the system structure.

[0152] Specifically, for a system unit, inspection object information is extracted from the system inspection record set, including inspection report number, inspection covered equipment, inspection type (system inspection or single-machine inspection), inspection conclusion, etc., to form a system inspection element set. Then, the inspection covered equipment set in the system inspection element set is compared with the equipment set in the system structure for consistency, and the inspection coverage consistency coefficient is calculated. The expression is:

[0153] ;

[0154] In the formula, This indicates the number of device primary keys covered by the inspected record in the system architecture; when the inspection record is a system-level inspection but no single device is listed. Take directly .

[0155] Simultaneously, a system key equipment integrity factor is introduced. This is used to prevent situations where only auxiliary equipment is inspected while core equipment is not. The expression is:

[0156] ;

[0157] In the formula, The number of core devices covered by the inspection records; This indicates the number of core devices in the system, with a value ranging from 1 to 50.

[0158] When the above test coverage consistency coefficient The system's critical equipment inspection integrity factor is greater than or equal to the set threshold (range 0.7-1.0). When the value is greater than or equal to the set threshold (range 0.8-1.0), the system is determined to be consistent with the system structure.

[0159] Sub-step 4.3: Logical consistency verification of filing and inspection time.

[0160] Specifically, at the system level, logical verification is performed on the filing time, initial inspection time, periodic inspection time, and system architecture formation time. This involves defining the system filing time, the system's most recent valid inspection time, and the system architecture confirmation time, and then calculating the system time consistency deviation. The expression is:

[0161] ;

[0162] In the formula, This indicates the commissioning time of the latest-formed core equipment in the system architecture; Indicates the system registration time; This indicates the allowed time for filing, ranging from 30 to 180 days.

[0163] Simultaneously calculate the test validity bias. The expression is:

[0164] ;

[0165] In the formula, Indicates the current verification time; This indicates the legally mandated inspection period for the system, expressed in days, with a range of 90-720. This indicates the time of the system's most recent valid verification.

[0166] When the above system time consistency deviation Less than or equal to 1 and the test validity bias When the value is less than or equal to 1, it is considered that the system filing and verification are consistent with the system structure in terms of time logic, and the system time consistency verification result is the maximum value of the two deviations.

[0167] Sub-step 4.4: System-level filing and inspection consistency determination.

[0168] Specifically, at the system unit level, the consistency results of the above-mentioned filing, inspection, and time consistency, as well as the system field tolerance, are comprehensively scored to form a system consistency score. This score is the weighted sum of all the above consistency results and their corresponding weight coefficients, with each weight coefficient ranging from 0.2 to 0.4, and the sum not exceeding 2. When the system consistency score is greater than or equal to a set threshold (ranging from 0.7 to 0.9), the system-level filing and inspection are determined to be consistent with the system structure; otherwise, it is determined to be inconsistent at the system level.

[0169] Step S150: Based on the consistency determination conclusion, mark and mask the permissible differences between the device file fields and the system-level compliance requirements.

[0170] In some embodiments, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes the following steps in step S150:

[0171] Step S151: Determine the system-level non-compliance index based on the field tolerance verification results, and generate the system verification status quantity in combination with the consistency judgment conclusion. The system verification status quantity is the difference between the score corresponding to the consistency judgment conclusion and the system-level non-compliance index.

[0172] Step S152: When the system calibration status quantity is in the specified state, determine the system impact degree of the corresponding device file field, and mark the corresponding device as abnormal based on the range of the system impact degree.

[0173] In a specific embodiment, the automatic verification method for special equipment files based on multimodal analysis provided by the present invention includes step 5, which generates a system-level file verification conclusion and masks false anomalies of individual devices. Based on the system-level consistency judgment result output in step 4, an automatic file verification conclusion centered on the system is generated. Cases where the system is compliant at the system level but there are reasonable differences in individual device fields are uniformly identified and masked, thereby completing the automatic verification of system-level special equipment files. This includes the following sub-steps:

[0174] Sub-step 5.1: System-level calibration of state determination and state quantity generation.

[0175] Specifically, for each system unit, a discretized system verification state quantity is generated based on the consistency score and non-compliance index, used for subsequent unified conclusion output. The system verification state quantity equals the difference between the system consistency score and the system-level non-compliance index. When the system verification state quantity is ≥ the third threshold, the system is judged as "system-level compliant"; when the fourth threshold is ≤ the system verification state quantity < the third threshold, the system is judged as "system-level requiring attention"; when the system verification state quantity < the fourth threshold, the system is judged as "system-level non-compliant". The third threshold ranges from 0.65 to 0.85, and the fourth threshold ranges from 0.35 to 0.55.

[0176] Sub-step 5.2: Judgment of the rationality of single device anomalies and identification of false anomalies.

[0177] Specifically, false anomaly identification is performed only for single-device anomalies when the system is in a "system-level compliance" or "system-level attention" state. For any device's anomaly field, its system impact is calculated. The expression is:

[0178] ;

[0179] In the formula, This represents the field normalization bias score calculated in step 3; This indicates the connectivity of the device in the system, with a value ranging from 0 to 1.

[0180] When the above system influence If the value is less than or equal to the set threshold (range 0.8-2.0), the anomaly is marked as "reasonable difference within the system"; otherwise, it is retained as "anomaly requiring review".

[0181] Sub-step 5.3: Generation of system-level verification conclusions and output of anomaly masking.

[0182] Specifically, the system unit summarizes the system status and anomaly labeling results, generating a system-level verification conclusion, including the system verification status, system compliance status, a list of masked false anomalies for individual devices, and a list of anomalies requiring manual review. The system anomaly retention rate is defined. To verify the validity of the blocking effect, the expression is:

[0183] ;

[0184] In the formula, Indicates the number of exceptions that were retained; This indicates the number of false anomalies that were masked.

[0185] When the above system abnormal retention rate If the result falls within the range of 0.1-0.4, the shielding result is considered to meet the expectations of the system-level calibration.

[0186] The automatic verification device for special equipment files based on multimodal analysis provided by the present invention will be described below. The automatic verification device for special equipment files based on multimodal analysis described below and the automatic verification method for special equipment files based on multimodal analysis described above can be referred to in correspondence.

[0187] like Figure 2 As shown, in one embodiment, a special equipment file automatic verification device based on multimodal parsing includes a file parsing module, a field mapping module, a tolerance verification module, a consistency determination module, and an anomaly marking and masking module.

[0188] The file parsing module is used to obtain project files and parse them to construct system candidate units.

[0189] The field mapping module is used to structurally merge devices in the system candidate units to determine the system boundary and map device profile fields to sub-units in the system candidate units.

[0190] The tolerance calibration module is used to introduce constraint rules into the candidate units of the system and perform tolerance calibration on the equipment file fields according to the constraint rules to obtain the field tolerance calibration results.

[0191] The consistency determination module is used to compare the field tolerance calibration results with the system inspection records to generate a consistency determination conclusion.

[0192] The anomaly marking and masking module is used to mark and mask permissible differences between device profile fields and system-level compliance requirements based on consistency judgment conclusions.

[0193] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0194] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for automatic verification of special equipment files based on multimodal analysis, characterized in that, The method includes: Obtain project files and parse them to construct system candidate units; The devices in the system candidate units are structurally merged to determine the system boundary, and the device file field is mapped to the sub-units in the system candidate units. Constraint rules are introduced into the system candidate units, and the tolerance of the equipment file fields is checked according to the constraint rules to obtain the field tolerance check results; The field tolerance calibration results are compared with the system inspection records to generate a consistency judgment conclusion. Based on the consistency determination conclusion, the permissible differences between the device file fields and the system-level compliance requirements are marked and masked. The process of acquiring project files and parsing them to construct system candidate units includes: Define the system correlation degree of the project files, and perform system correlation screening on the project files according to the system correlation degree to obtain candidate files with a system correlation degree not lower than a set threshold; Field elements are extracted from the candidate files, and confidence levels are set for the extracted field elements to construct a structured element table based on the field elements and corresponding confidence levels. The step of acquiring project files and parsing them to construct system candidate units further includes: Based on the structured element table, determine the various representations of each device in different engineering project files, and calculate the similarity between different representations, so as to merge the various representations with a similarity of not less than a set threshold into the primary key of the corresponding device. The candidate files associated with the primary key are linked to obtain the element primary key table, and the connection relationship between different devices is determined. The connection relationship between different devices includes the primary key of the connection starting device, the primary key of the connection ending device, and the connection type. Using the system name as the aggregation clue, and combining the element primary key table and the connection relationships between different devices, system candidate units under the system name are constructed.

2. The automatic verification method for special equipment files based on multimodal analysis according to claim 1, characterized in that, The step of structurally merging the devices in the system candidate units to determine the system boundary and mapping the device file fields to the sub-units in the system candidate units includes: Calculate the device overlap between different system candidate units, and merge the different system candidate units into the same system candidate unit when the overlap is not lower than a set threshold; The connection relationships of different devices in the candidate units of the system are verified, and the boundary consistency coefficient is calculated based on the number of device connection relationships within the system and the number of device connection relationships between the system and outside the system, so as to prune the primary keys of devices with a boundary consistency coefficient lower than a set threshold.

3. The method for automatic verification of special equipment files based on multimodal analysis according to claim 2, characterized in that, The step of structurally merging the devices in the system candidate units to determine the system boundary and mapping the device file field to the sub-units of the system candidate units further includes: Calculate the connectivity of each device primary key in the system candidate units with other devices within the system, and determine the type of the device primary key based on the range of connectivity. Based on the device primary key corresponding to the device file field, and combined with the type of the device primary key and the system candidate unit, a mapping relationship between the device file field and the system candidate unit is generated.

4. The method for automatic verification of special equipment files based on multimodal analysis according to claim 1, characterized in that, The process of introducing constraint rules into the system candidate units and performing tolerance calibration on the equipment file fields according to the constraint rules to obtain field tolerance calibration results includes: Based on the mapping relationship between the device file fields and the system candidate units, a suitable system constraint template is selected, and the system constraint template is instantiated using the actual fields of the current system to obtain a system constraint instance; The initial tolerance threshold is determined based on the system constraint instance, and the initial tolerance threshold is then corrected in layers according to the equipment type weight and system mechanism residual to obtain the field tolerance threshold. Calculate the relative deviation of each equipment file field, compare the relative deviation with the field tolerance threshold, and make a secondary judgment to aggregate the relative deviation comparison results of all equipment file fields and output the field tolerance calibration result.

5. The method for automatic verification of special equipment files based on multimodal analysis according to claim 1, characterized in that, The step of comparing the field tolerance calibration results with the system inspection records to generate a consistency judgment conclusion includes: Extract filing elements from the system filing files corresponding to the system candidate units, and convert the filing elements into a structured filing element set. Align the structured filing element set with the equipment file fields and the system candidate units to obtain a system filing structure mapping table. The inspection elements are extracted from the system inspection records, and the corresponding inspection coverage equipment, equipment file fields and system candidate units in the inspection elements are compared to obtain the system inspection consistency results. Based on the system filing time, system inspection time, and system structure confirmation time, calculate the system time consistency deviation and inspection validity deviation, and determine the system time consistency verification result based on the system time consistency deviation and inspection validity deviation. The consistency determination conclusion is a result obtained by comprehensively evaluating the system filing structure mapping table, system inspection consistency results, system time consistency verification results, and field tolerance verification results.

6. The method for automatic verification of special equipment files based on multimodal analysis according to claim 5, characterized in that, Based on the consistency determination conclusion, the allowable differences between the device file fields and system-level compliance requirements are marked and masked, including: The system-level non-compliance index is determined based on the field tolerance verification results, and a system verification status quantity is generated in combination with the consistency judgment conclusion. The system verification status quantity is the difference between the score corresponding to the consistency judgment conclusion and the system-level non-compliance index. When the system calibration status quantity is in the specified state, the system impact degree of the corresponding device file field is determined, and the corresponding device is marked as abnormal based on the range of the system impact degree.

7. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-6.

Images