Method for scoring the risk of steel structure connection based on the bolt loosening characteristics of the converter station

Abstract

The application discloses a steel structure connection risk scoring method based on bolt loosening characteristics of a converter station, and particularly relates to the field of steel structure monitoring, and aims to solve the problem that bolt loosening characteristics are difficult to be converted into connection risk scores; bolt loosening characteristic data of a steel structure connection part of the converter station are acquired, a connection unit is divided based on a basic boundary of a connection plate or a node plate and combined with structure role information to form a node atlas; the bolt loosening characteristic data are converted into loosening state values and collected into a loosening state set of the connection unit according to the node atlas; continuous anomalies are extracted, force transmission direction anomaly through degrees and boundary direction anomaly fragmentation degrees of the continuous anomalies across the boundary of the connection plate or the node plate are calculated, and abnormal co-occurrence segments are reselected; then, a connection unit risk value is formed according to the concentration degree of the abnormal co-occurrence segments, and a connection risk score result is output after being verified in combination with historical disposal results.

Description

Technical Field

[0001] This invention relates to the field of steel structure monitoring, and more specifically, to a method for risk scoring of steel structure connections based on the characteristics of loose bolts in converter stations. Background Technology

[0002] The steel structure connections in converter stations typically include the valve hall steel structure, equipment supports, busbar supports, and connecting plates or node plates in the maintenance platform. These connections and force transmission rely on groups of bolts. During long-term operation, equipment vibration, environmental changes, and continuous stress can easily cause some bolts to show signs of loosening, such as misalignment of marking lines, changes in appearance angle, changes in exposed condition, changes in local response, changes in the condition of connection points, or abnormal maintenance records. Existing inspection and monitoring methods can usually detect changes in bolt condition from inspection images, vibration records, strain records, or maintenance records, and based on this, determine whether there is a risk of loosening in individual bolts.

[0003] However, the bolts in the steel structure connections of the converter station do not work in isolation. The same connection plate or node plate may simultaneously house valve hall steel structure connection bolts, equipment support connection bolts, busbar support connection bolts, or maintenance platform connection bolts. The bolt hole arrangement, adjacent relationships, boundary positions, and structural roles directly affect the expression of risk to the connection unit from loosening signs. Existing processing methods often treat bolt loosening characteristic data from inspection images, vibration records, strain records, and maintenance records as single-point anomalies, lacking a unified process for mapping loosening characteristics from different sources to the associated bolts, connection units, and bolt connection relationships. When adjacent bolts show suspected anomalies or abnormal states, it is also difficult to determine whether these anomalies occur continuously along the bolt connection relationship within the same connection unit or are discrete anomalies located on either side of the boundaries of different connection functions or different connection plates / node plates.

[0004] Furthermore, in converter station steel structure connections, bolts often cross the boundaries of connecting plates or node plates. Loosening signs near the same location may appear as a continuous pattern along the force transmission direction, or they may simply be abnormal parallel occurrences of different connection units on either side of the boundary. If risk assessment is based solely on spatial distance or the condition of a single bolt, loosening signs under different structural roles can easily be merged into the same risk object, or continuous anomalies in the same force transmission direction can be treated separately, leading to inconsistencies between the risk value of the connection unit and the actual connection relationship. Therefore, existing methods, after detecting bolt loosening signs, struggle to generate a connection risk score that simultaneously reflects the bolt connection relationship, structural role information, cross-boundary anomaly morphology, and historical handling results of similar connection units.

[0005] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

[0006] To overcome the aforementioned deficiencies in the prior art, embodiments of the present invention provide a steel structure connection risk scoring method based on the bolt loosening characteristics of converter stations. This method acquires bolt loosening characteristic data at the connection points of the converter station's steel structure, divides connection units using connection plates or node plates as basic boundaries and combining structural role information to form a node map. Based on the node map, the bolt loosening characteristic data is converted into loosening state values ​​and aggregated into a loosening state set for the connection units. Continuous anomalies are extracted, and for continuous anomalies crossing the boundaries of connection plates or node plates, the force transmission direction anomaly continuity and boundary direction anomaly fragmentation are calculated, and anomaly co-occurrence segments are reselected. Finally, based on the concentration of anomaly co-occurrence segments, a connection unit risk value is formed, and after verification with historical handling results, the connection risk scoring result is output to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: S1. Obtain bolt loosening characteristic data of the steel structure connection parts of the converter station, divide the connection unit based on the connection plate or node plate as the basic boundary and combined with the structural role information, determine the bolt connection relationship between bolts in the connection unit, and form a node map containing bolt connection relationship and structural role information. S2. Based on the bolt connection relationships in the node diagram, convert the bolt loosening characteristic data into the corresponding bolt loosening state values, and aggregate the loosening state values ​​by connection unit to form a loosening state set of connection unit. S3. Extract continuous anomalies from the set of loose states of the connection unit. When a continuous anomaly crosses the boundary of the connection plate or node plate, calculate the force transmission direction anomaly penetration degree and the boundary direction anomaly separation degree. Based on the direction results of the force transmission direction anomaly penetration degree and the boundary direction anomaly separation degree, reselect the continuous anomaly as an anomaly co-occurrence segment; otherwise, determine the continuous anomaly as an anomaly co-occurrence segment. S4. Calculate the associated risk value of the connection unit based on the concentration of abnormal co-occurring segments in the connection unit, and then map the associated risk value of the connection unit back to the structural role information in the node graph to form the risk value of the connection unit. S5. Perform consistency verification between the risk value of the connection unit and the historical handling results of similar connection units to obtain the verification difference. When the verification difference falls within the range of the historical handling results of similar connection units, output the connection risk score result; otherwise, update the boundary of the connection board or node board according to the verification difference and return to S2 until the verification difference falls within the range of the historical handling results of similar connection units, or the abnormal co-occurrence fragment is converted into the state within a single connection unit and then the connection risk score result is output.

[0008] Furthermore, the bolt loosening characteristic data are generated separately according to inspection images, vibration records, strain records and maintenance records, and are indexed by bolt number, name of connecting plate or node plate, and collection location or maintenance location. The bolt loosening feature data generated from the inspection images include the relative position of the bolt marking line, the position of the bolt outer contour, the appearance angle of the bolt head, and the exposed state of the bolt. The bolt loosening characteristic data generated by vibration recordings include the acquisition location, acquisition time, vibration amplitude variation, and characteristic frequency band response value variation. The bolt loosening characteristic data generated by strain records include strain acquisition location, strain acquisition time, location of changes in the state of the connection, and direction of strain change. The bolt loosening characteristic data generated in the maintenance record includes bolt number, name of connecting plate or node plate, maintenance location, record of abnormality upon review, description of abnormality before tightening, and confirmation record after tightening.

[0009] Furthermore, when bolt loosening feature data falls into multiple candidate bolt positions simultaneously, it is first determined whether the data source contains the name of the connecting plate or node plate. If the name of the connecting plate or node plate is included, only candidate bolts with the same name as the connecting plate or node plate will be retained. If the name of the connecting plate or node plate is not included, all candidate bolt entry positions will be compared. The bolts assigned to a particular bolt are determined by comparing their positions in the following order: minimum sum of bidirectional position differences, minimum position difference in the direction of the foundation boundary, and minimum position difference in the direction perpendicular to the foundation boundary. If the candidate bolts cannot be distinguished by location comparison, the structural role information is read, and the candidate bolts whose structural role information matches the data source label are identified as the assigned bolts; if the structural role information still cannot distinguish the candidate bolts, the corresponding bolt loosening feature data is marked as feature data to be reviewed, and it is used to determine the loosening state value after the manual review result is formed.

[0010] Furthermore, in the set of loose connection unit states, continuous anomalies within the connection unit are first extracted on a unit-by-unit basis; Bolts with a loosening status of abnormal or suspected abnormal are designated as abnormal candidate bolts, while bolts with a loosening status of normal are designated as interrupted bolts. When the bolts belonging to the connection plate or node plate on both sides of the boundary are connected through the boundary adjacency relationship, and both bolts belonging to the connection plate on both sides of the boundary are abnormal candidate bolts, the continuous abnormalities inside the corresponding connection unit on both sides of the boundary are merged and marked as cross-boundary continuous abnormalities. When there is no boundary adjacency relationship between the two sides of the boundary, or when the bolt on either side of the boundary is in a normal state, no cross-boundary continuous anomaly is formed, and the continuous anomaly within the connection unit is identified as an anomaly co-occurrence segment.

[0011] Furthermore, structural role information is read from the converter station steel structure design drawings, as-built drawings, equipment support ledgers, busbar support ledgers, maintenance platform ledgers, or 3D models; When the same connecting plate or node plate corresponds to multiple connection functions in the ledger, the bolts are assigned to the corresponding structural role information according to the bolt number, hole row, and plate position. When the structural role information of a single bolt is not directly recorded in the ledger, the structural role information is determined from the design annotations of the corresponding connection parts by using the name of the connection plate or node plate as the primary index and the hole column and plate position as the secondary index.

[0012] Furthermore, historical records of unloose connections are obtained from commissioning and acceptance images, the most recent re-tightening confirmation record, inspection images confirmed by manual review as unloose, or confirmation records after handling of the same connection unit. When the first execution is performed and there is no historical record of loosening, the image of the same type of connection unit that is confirmed to be loose during this manual review is used as a temporary record of loosening; When the re-tightening confirmation record is generated, the temporary unloosening record is replaced with the re-tightening confirmation record; If the maintenance record only contains a confirmation record after tightening, or if the confirmation record after tightening is later than the record of abnormality during review and the description of abnormality before tightening, the maintenance record does not indicate any signs of loosening.

[0013] Furthermore, when the range of historical processing results is empty, the verification difference is recorded as the range of historical processing results that does not fall into the same type of connection unit, and the source record of the empty range is retained; If the abnormal co-occurrence fragment is not converted to a state within a single connection unit, the node graph is updated according to the boundary position of the connection plate or node plate corresponding to the abnormal co-occurrence fragment and S2 is returned. If the abnormal co-occurrence fragment has been converted to a state within a single connection unit, the connection risk score result is output, retaining the verification difference and empty range source record.

[0014] Furthermore, the risk value of the connection unit includes the segment risk level, concentration degree, abnormal co-occurrence segment number, and the boundary position of the connection plate or node plate. The connection risk scoring results are output according to the segment risk level. When the segment risk level is low segment risk, the connection risk score output is level 1; When the segment risk level is median segment risk, the connectivity risk score output is level 2; When the segment risk level is high segment risk, the connectivity risk score output is level 3; The connection risk scoring results also record the connection unit number, structural role information, connection unit risk value, abnormal co-occurrence segment number, verification difference, and historical handling result range identifier.

[0015] Furthermore, a boundary version record is generated each time the connection board or node board boundary is updated; The boundary version record includes at least the boundary position before the update, the boundary position after the update, the bolt number of the reassigned connection unit, and the verification difference type that triggered the update. If the same check difference points to the same boundary position again after returning to S2, the same boundary update will not be repeated, and the boundary assignment will be checked along the other side of the abnormal co-occurrence segment connecting the board or node board boundary.

[0016] Furthermore, the scope of historical treatment results consists of closed treatment records, which simultaneously include the risk value of the connection unit before treatment, the co-occurring abnormal fragments before treatment, the risk value of the connection unit after treatment, the changes in the co-occurring abnormal fragments after treatment, and the confirmation records after treatment. The closed-loop disposal records are compiled into verifiable records. Verifiable records include at least the fragment risk level before disposal, the concentration level before disposal, the boundary position of the connecting plate or node plate before disposal, whether the abnormal co-occurring fragments before disposal were obtained by cross-boundary continuous abnormal reselection, the changes in the abnormal co-occurring fragments after disposal, and the changes in the risk value of the connecting unit after disposal.

[0017] The technical effects and advantages of this invention, which uses a steel structure connection risk scoring method based on the characteristics of loose bolts in converter stations, are as follows: This invention addresses the problem of bolts with different structural roles and loosening characteristics from different sources on the same connecting plate or node plate easily being mixed into the same judgment object. It first uniformly assigns inspection images, vibration records, strain records, and maintenance records to specific bolts, and then divides the connection units according to foundation boundaries and structural role information. Therefore, even if equipment bracket connecting bolts, busbar bracket connecting bolts, and maintenance platform connecting bolts are located on the same connecting plate, they will not be merged into the same connection risk object due to spatial proximity; only one loosening state value is retained for the same assigned bolt within the same analysis period, reducing duplicate or conflicting judgments when image anomalies, vibration anomalies, and post-tightening confirmation records coexist.

[0018] This invention addresses the problem of distinguishing between continuous loosening within the same connection unit and parallel anomalies across boundaries when adjacent bolts exhibit anomalies. It extracts continuous anomalies by utilizing bolt connection relationships and simultaneously assesses the force transmission direction anomaly continuity and the boundary direction anomaly fragmentation degree when a continuous anomaly crosses the boundary of a connection plate or node plate. In this way, anomalies that continue intact along the force transmission direction are retained as co-occurring segments of the same anomaly, while anomalies whose state levels change abruptly on both sides of the boundary are separated and reselected according to the boundary of the connection plate or node plate. This avoids merging anomalies under different connection effects simply because bolt distances are close, and also avoids mistakenly splitting anomalies that actually unfold continuously along the connection relationship into multiple isolated points.

[0019] This invention addresses the problem that existing single-point loosening assessments struggle to generate connection risk scores. It combines the concentration of co-occurring abnormal segments, segment risk level, structural role information, and the range of historical handling results to form the connection risk score. The score not only reflects whether a particular bolt is abnormal but also retains the connection unit number, structural role information, abnormal co-occurring segment number, verification discrepancies, and the range of historical handling results. When verification discrepancies indicate a deviation from the boundary attribution, boundary version records restrict repeated updates to the same boundary location, ensuring that the score corresponds to a specific connection function, the source of the abnormal segment, and the historical handling relationship. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the risk scoring method for steel structure connections based on the characteristics of loose bolts in converter stations, as presented in this invention. Figure 2 This is a schematic diagram illustrating the formation of a loose state set of the connection unit in this invention; Figure 3 This is a schematic diagram illustrating the formation of risk values ​​and the reflection of structural role information in the connection unit of the present invention. Detailed Implementation

[0021] 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, and 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.

[0022] Please see Figures 1 to 3 This invention provides a method for risk assessment of steel structure connections based on the characteristics of loose bolts in converter stations, including: S1. Obtain bolt loosening characteristic data of the steel structure connection parts of the converter station, divide the connection unit based on the connection plate or node plate as the basic boundary and combined with the structural role information, determine the bolt connection relationship between bolts in the connection unit, and form a node map containing bolt connection relationship and structural role information. S2. Based on the bolt connection relationships in the node diagram, convert the bolt loosening characteristic data into the corresponding bolt loosening state values, and aggregate the loosening state values ​​by connection unit to form a loosening state set of connection unit. S3. Extract continuous anomalies from the set of loose states of the connection unit. When a continuous anomaly crosses the boundary of the connection plate or node plate, calculate the force transmission direction anomaly penetration degree and the boundary direction anomaly separation degree. Based on the direction results of the force transmission direction anomaly penetration degree and the boundary direction anomaly separation degree, reselect the continuous anomaly as an anomaly co-occurrence segment; otherwise, determine the continuous anomaly as an anomaly co-occurrence segment. S4. Calculate the associated risk value of the connection unit based on the concentration of abnormal co-occurring segments in the connection unit, and then map the associated risk value of the connection unit back to the structural role information in the node graph to form the risk value of the connection unit. S5. Perform consistency verification between the risk value of the connection unit and the historical handling results of similar connection units to obtain the verification difference. When the verification difference falls within the range of the historical handling results of similar connection units, output the connection risk score result; otherwise, update the boundary of the connection board or node board according to the verification difference and return to S2 until the verification difference falls within the range of the historical handling results of similar connection units, or the abnormal co-occurrence fragment is converted into the state within a single connection unit and then the connection risk score result is output.

[0023] This invention focuses on processing bolt loosening characteristic data at the connection points of converter station steel structures. First, it maps state changes in inspection images, vibration records, strain records, and maintenance records to the associated bolts. Then, within the foundation boundary formed by the connection plate or node plate, connection units are divided based on structural role information, constructing a node map containing bolt connection relationships and structural role information. Subsequently, bolt loosening characteristic data from different sources are uniformly converted into normal, suspected abnormal, or abnormal states, forming a loosening state set for each connection unit. Based on this, continuous anomalies are extracted according to bolt connection relationships. For continuous anomalies crossing the boundaries of the connection plate or node plate, the force transmission direction and boundary direction are distinguished, forming anomaly co-occurrence segments. Then, the risk values ​​of the connection units are formed by arranging the anomaly co-occurrence segments according to their concentration and loosening state values. Consistency verification is performed using the historical treatment results range of similar connection units, ultimately outputting the connection risk score.

[0024] Detailed implementation of step S1: Risk assessment of steel structure connections in converter stations requires addressing the issue of data attribution. Although the input bolt loosening characteristic data comes from inspection images, vibration records, strain records, or maintenance records, it must ultimately be attributed to specific bolts, connection units, and structural roles. However, the same connection plate or node plate in a converter station often contains bolts with different connection functions. If a clear attribution boundary is not established in S1, the loosening state value formed in S2 will be mixed with incorrect connection units.

[0025] Bolt loosening characteristic data are generated separately according to their data sources. Bolt loosening characteristic data from inspection images include the relative position of bolt marking lines, the position of the bolt's outer contour, the appearance angle of the bolt head, and the exposed state of the bolt. Bolt loosening characteristic data from vibration records include the acquisition location, acquisition time, vibration amplitude changes, and characteristic frequency band response value changes. Bolt loosening characteristic data from strain records include the strain acquisition location, strain acquisition time, location of changes in the connection's state, and direction of strain change. Bolt loosening characteristic data from maintenance records include the bolt number, name of the connecting plate or gusset plate, maintenance location, verification anomaly record, description of anomalies before retightening, and confirmation record after retightening. All of the above data uses the bolt number, name of the connecting plate or gusset plate, and acquisition location or maintenance location as the index for subsequent mapping to the assigned bolt.

[0026] First, address the differences in the sources of bolt loosening characteristic data to avoid inconsistent data entries formed by inspection images, vibration records, strain records, and maintenance records. Specifically, first establish a bolt location list for the steel structure connection parts of the converter station. The bolt location list should at least record the bolt number, the connecting plate or node plate to which it belongs, the plate position, the hole row affiliation, and the structural role information. The bolt number is used to distinguish different bolts, the connecting plate or node plate to which it belongs is used to determine the foundation boundary, the plate position is used to compare the positions when there are multiple candidate bolts, the hole row affiliation is used to establish the bolt connection relationship, and the structural role information is used to divide the connection unit.

[0027] The bolt location list is established from the converter station steel structure design drawings, as-built drawings, 3D models, equipment ledgers, or on-site verification records. When records from different sources are inconsistent, the bolt number and plate position determined by the on-site verification records shall be used as the current version of the bolt location list.

[0028] After acquiring bolt loosening characteristic data, first read the information from the data source that points to the bolt location. For bolt loosening characteristic data from inspection images, match the bolt location to the bolt location list based on the plate location, bolt outline location, or marker line location in the image; for bolt loosening characteristic data from vibration records, match the bolt location to the bolt location list based on the connection plate or node plate location corresponding to the acquisition location; for bolt loosening characteristic data from strain records, match the bolt location list based on the connection part to which the strain record belongs; for bolt loosening characteristic data from maintenance records, match the bolt location list based on the bolt number, connection plate or node plate name, and maintenance location in the record.

[0029] When a single bolt loosening feature data point corresponds to only one bolt location, the attribution relationship between the bolt loosening feature data and the bolt location is directly established. When a single bolt loosening feature data point falls into multiple candidate bolt locations simultaneously, it is first determined whether the data source contains the name of a connecting plate or node plate. If it does, only candidate bolts with the same connecting plate or node plate name are retained; otherwise, all candidate bolts are retained for location comparison. During location comparison, the foundation boundary direction is determined by the extension direction of the hole array in the connecting plate or node plate, and the direction perpendicular to the foundation boundary is the direction within the plate surface that is perpendicular to the foundation boundary direction. After the bolt loosening feature data forms feature location points within the plate surface, these feature location points are projected onto both the foundation boundary direction and the direction perpendicular to the foundation boundary direction to obtain the bidirectional position of the feature location points. Similarly, the plate surface positions of the candidate bolts are projected onto the above two directions to obtain the bidirectional positions of the candidate bolts. For each candidate bolt, the positional difference in the direction of the foundation boundary and the positional difference in the direction perpendicular to the foundation boundary are obtained respectively, and compared in the following order: minimum sum of bidirectional positional differences, minimum positional difference in the direction of the foundation boundary, and minimum positional difference in the direction perpendicular to the foundation boundary. The candidate bolt that first meets the minimum value condition in the above comparison order is determined as the assigned bolt. If there are still candidate bolts that cannot be distinguished, the structural role information is read. The candidate bolt whose structural role information is consistent with the data source label is determined as the assigned bolt. If the structural role information still cannot be distinguished, the corresponding bolt loosening feature data is marked as feature data to be reviewed, and it is used in the determination of loosening state value after the manual review result is formed.

[0030] In one embodiment, an inspection image identifies a marker line offset. The image annotation shows that the offset location is on the same node plate in the valve hall steel structure, close to two adjacent bolts. During processing, the positions of the two candidate bolts on the node plate are first read. Then, the position differences between the offset location and the two candidate bolts in the foundation boundary direction, and the position differences between the offset location and the two candidate bolts in the direction perpendicular to the foundation boundary, are obtained respectively. If the sum of the two-way position differences of one candidate bolt is less than that of the other candidate bolt, the bolt loosening feature data from the image source is assigned to the candidate bolt with the smaller sum of the two-way position differences. If the sum of the two-way position differences cannot be distinguished, the position differences in the foundation boundary direction and the position differences in the direction perpendicular to the foundation boundary are compared sequentially, and the candidate bolt that first meets the minimum value condition is determined as the assigned bolt. If it still cannot be distinguished, the node plate name and structural role information in the image annotation are read. Candidate bolts with the same node plate name and the same structural role information are assigned as the assigned bolt.

[0031] Once the loosening characteristic data of each bolt is clearly assigned to a specific bolt, it is no longer stored separately according to the data source, but is instead used as a unified object to divide the connection units according to the connection plate or node plate and structural role information.

[0032] After obtaining the assigned bolts, the bolts are divided into connection units. The connection units are not simply determined by spatial location, but are constrained by both foundation boundaries and structural role information. The foundation boundaries are determined by the connection plates or node plates, representing the physical connection range of the bolts. The structural role information indicates the connection function of the bolt connection in the converter station steel structure, such as valve hall steel structure connection, equipment support connection, busbar support connection, or maintenance platform connection.

[0033] Structural role information is retrieved from the converter station steel structure design drawings, as-built drawings, equipment support ledgers, busbar support ledgers, maintenance platform ledgers, or 3D models. If the same connecting plate or node plate corresponds to multiple connection functions in the ledger, the bolts are assigned to the corresponding structural role information according to the bolt number, hole row affiliation, and plate position. If the ledger does not directly record the structural role information of a single bolt, the structural role information is determined from the design annotations of the corresponding connection parts, using the name of the connecting plate or node plate as the primary index and the hole row affiliation and plate position as the secondary index.

[0034] In specific processing, first assign the bolts to the corresponding foundation boundary according to the connecting plate or node plate to which they belong; for bolts located within the same foundation boundary, read the structural role information; if several bolts are located within the same foundation boundary and have the same structural role information, they are assigned to the same connection unit; if several bolts are located within the same foundation boundary but have different structural role information, they are assigned to different connection units; if several bolts have the same structural role information but are located within the foundation boundary formed by different connecting plates or node plates, they are not assigned to the same connection unit.

[0035] To ensure the verifiability of the partitioning results, the connection element record must include at least the connection element number, foundation boundary name, structural role information, assigned bolt number, and plate surface location of the assigned bolt. The connection element number is jointly determined by the foundation boundary name and the structural role information. The foundation boundary name is used to trace the connection plate or node plate, the structural role information is used to distinguish different connection functions on the same connection plate or node plate, the assigned bolt number is used to determine the range of bolts included in the connection element, and the plate surface location is used to determine whether a bolted connection relationship is formed between the bolts.

[0036] If both equipment bracket connecting bolts and maintenance platform connecting bolts exist on the same connecting plate, the process should first confirm that both types of bolts belong to the same foundation boundary, and then read the structural role information. Equipment bracket connecting bolts should be assigned to the connection unit corresponding to the equipment bracket connection, and maintenance platform connecting bolts should be assigned to the connection unit corresponding to the maintenance platform connection. Even if the two types of bolts are close to each other on the connecting plate, they will not be assigned to the same connection unit because the structural role information of the two types of bolts is different.

[0037] Once a connection element has both basic boundary constraints and structural role information constraints, different connection actions on the same connection plate or node plate will not be merged into the same risk scoring object.

[0038] After obtaining the connection unit, the bolt connection relationships within the connection unit are further determined. The bolt connection relationships are used to indicate which bolts within the connection unit are adjacent on the plate surface; not all bolts located in the same connection unit directly form a bolt connection relationship.

[0039] In specific processing, first read the hole column affiliation and plate position of each bolt in the connection unit; the hole column affiliation indicates the arrangement line of the bolt on the connection plate or node plate, and the plate position indicates the order of the bolt in its respective hole column; for two bolts belonging to the same hole column in the same connection unit, if the two bolts are arranged in adjacent order in the hole column, a bolt connection relationship is established; if there are other bolts between the two bolts, a bolt connection relationship is not established directly.

[0040] For two bolts belonging to different hole rows within the same connection unit, first determine whether the hole rows to which the two bolts belong constitute adjacent hole rows in the same stress-bearing part; if the two hole rows are adjacent and the arrangement positions of the two bolts in their respective hole rows correspond, then a bolt connection relationship spanning hole rows is established; if the two hole rows are not adjacent, or the arrangement positions do not correspond, then a bolt connection relationship spanning hole rows is not established. Corresponding arrangement positions mean that the two bolts are in the same sequential position in their respective hole rows, or are located in adjacent sequential positions and there are no other bolts separating them.

[0041] After determining the bolt connection relationships, a node diagram is generated. The node diagram uses bolts as nodes and bolt connection relationships as the connections between nodes. The node diagram also records the connection unit to which each bolt belongs, the foundation boundary, the structural role information, the hole row affiliation, the plate position, and the bolt loosening characteristic data. For two bolts with established bolt connection relationships, the node diagram records the connection direction as adjacent in the same hole row or adjacent across hole rows. For two bolts without established bolt connection relationships, the node diagram does not record the connection relationship.

[0042] The node diagram also records the boundary adjacency relationships on both sides of the connection plate or node plate boundary. Boundary adjacency relationships are used to indicate the assigned bolts located on both sides of adjacent foundation boundaries, corresponding to each other on the plate surface, and having a force transfer relationship in the structural role information. Boundary adjacency relationships do not change the connection element division results already formed in S1, nor do they merge different connection elements into the same connection element. Boundary adjacency relationships are only used by S3 as the basis for determining whether continuous anomalies cross the boundary of the connection plate or node plate. Therefore, S3 uses bolt connection relationships when extracting continuous anomalies within the connection element, and reads boundary adjacency relationships when determining cross-boundary continuation.

[0043] The node map can not only indicate which bolt corresponds to the bolt loosening feature data, but also which connection unit the bolt belongs to, which structural role information the connection unit corresponds to, and whether bolts form a bolt connection relationship, thus meeting the input requirements of S2 to collect loosening state values ​​according to bolt connection relationships.

[0044] After S1 processing, the bolt loosening feature data is first mapped to the assigned bolt, then the connection units are divided according to the foundation boundary and structural role information, and finally the bolt connection relationship and node map are formed within the connection unit. The node map formed in this way can exclude the mixed aggregation of bolts with different connection functions and provide clear data boundaries for the loosening state set of the connection unit.

[0045] Detailed implementation of step S2: S2 follows the node map formed by S1. The input objects include bolt loosening feature data, connection units, and bolt connection relationships that have been assigned to the bolts. The processing goal is to form a unique loosening state value for each assigned bolt within the same analysis period. However, different data sources express loosening signs in different ways. If time period merging and source judgment are not performed first, the loosening state set of the connection unit will have the problem of multiple states of the same bolt coexisting.

[0046] First, determine the analysis period corresponding to the current connection risk assessment result. The analysis period ends at the trigger time of the current connection risk assessment and begins at the time of the first bolt loosening characteristic data after the output of the previous connection risk assessment result. When executing this method for the first time, the earliest bolt loosening characteristic data time already acquired is taken as the starting point. After the analysis period is determined, read the associated bolts, connection units, and bolt connection relationships in the node map, and group the bolt loosening characteristic data according to the associated bolts. The inspection images, vibration records, strain records, and maintenance records of the same associated bolt within the analysis period are retained as status records under different data sources.

[0047] When multiple status records exist for the same bolt from the same data source, the status records are first arranged according to the collection time. Then, each status record is checked to see if it corresponds to the same bolt and the same connection unit. Status records that correspond to both the same bolt and the same connection unit are retained; status records that do not correspond to both are not included in the determination of the loosening status value. If multiple status records are retained from the same data source, the status record closest to the end of the analysis period is selected as the current status record for this data source.

[0048] Maintenance records need to be organized separately. Verification of anomaly records and descriptions of anomalies before retightening are retained as anomaly descriptions in the maintenance records; confirmation records after retightening are retained as handling confirmation records. Handling confirmation records do not directly eliminate loosening signs from inspection images, vibration records, or strain records. Instead, they are used to determine whether the maintenance records still indicate loosening when determining loosening status values. Through the above processing, bolts belonging to the same category form current status records differentiated by data source within the same analysis period. These status records still retain the acquisition time, data source, and original status description.

[0049] After obtaining the current status record, determine whether each data source points to signs of loosening, and then convert the determination results into a loosening status value for the assigned bolt. The loosening status value is categorized as normal, suspected abnormal, and abnormal. A normal status indicates that the current status record does not show any signs of loosening; a suspected abnormal status indicates that a single data source shows signs of loosening and there is a lack of other data sources pointing to it; an abnormal status indicates that at least two types of data sources point to signs of loosening within the same analysis period, or that the maintenance record contains a verification anomaly and one of the inspection images, vibration records, or strain records simultaneously points to signs of loosening.

[0050] Historical records of unloose bolts are obtained from commissioning and acceptance images, the most recent re-tightening confirmation record, inspection images confirmed as unloose by manual review, or confirmation records after handling the same connection unit. When performing a new operation and no historical records of unloose bolts exist, an image of a similar connection unit confirmed as unloose by manual review in the current operation is used as a temporary unloose bolt record; this temporary record is replaced after subsequent re-tightening confirmation records are generated. Historical unloose bolt records must include at least the bolt number, relative position of the marking line, appearance angle, exposed condition, collection time, and confirmation source.

[0051] The process for determining the source of the inspection image is as follows: First, read the relative position of the marker line, the position of the bolt's outer contour, the appearance angle of the bolt head, and the exposed state of the bolt from the current status record. Then, read the historical records of the same bolt that were not loosened. The relative position of the marker line is determined with the center of the bolt head as the reference point. If the current relative position of the marker line shifts angularly relative to the historical records of not loosened bolts along the bolt loosening direction, the relative position of the marker line points to signs of loosening. The appearance angle of the bolt head is determined with the direction of the bolt head edge in the historical records of not loosened bolts as the reference direction. If the current direction of the bolt head edge changes relative to the reference direction along the bolt loosening direction, the appearance angle of the bolt head points to signs of loosening. The exposed state of the bolt is determined with the exposed thread section or the exposed range of the bolt end in the historical records of not loosened bolts as the reference object. If the current exposed thread section or the exposed range of the bolt end increases relative to the historical records of not loosened bolts, the exposed state of the bolt points to signs of loosening. If any of the above image contents points to signs of loosening, the inspection image points to signs of loosening; if none of the above image contents point to signs of loosening, the inspection image does not point to signs of loosening. The process for determining the source of vibration records is as follows: First, read the vibration amplitude change or characteristic frequency response value change at the sampling location of the assigned bolt in the current state record. Then, read the vibration amplitude change or characteristic frequency response value change at adjacent sampling locations within the same connection unit. If both the sampling location of the assigned bolt and the adjacent sampling locations show an increase relative to their respective historical non-loosening records, the vibration record indicates a loosening; otherwise, the vibration record does not indicate a loosening. The process for determining the source of strain records is as follows: Read the correspondence between the location of the change in the connection state and the location of the assigned bolt in the current state record. If the location of the change in the connection state falls within the connection unit where the assigned bolt is located and covers the hole row where the assigned bolt is located or an adjacent hole row, the strain record indicates a loosening; otherwise, the strain record does not indicate a loosening. The process for determining the source of maintenance records is as follows: if the maintenance record contains a record of abnormality upon review or a description of abnormality before tightening, and there is no confirmation record after tightening within the same analysis period, then the maintenance record indicates signs of loosening; if the maintenance record only contains a confirmation record after tightening, or the confirmation record after tightening is later than the record of abnormality upon review and the description of abnormality before tightening, then the maintenance record does not indicate signs of loosening.

[0052] Local response changes in vibration records include at least the changes in vibration amplitude or characteristic frequency band response values ​​at the connection location of the assigned bolt; the changes in vibration amplitude are obtained from the differences between the vibration records at the same acquisition location during the analysis period and the historical records of no loosening; the changes in characteristic frequency band response values ​​are obtained from the changes in response values ​​at the same acquisition location within the characteristic frequency band of structural vibration; adjacent acquisition locations are determined by the bolt connection relationship in the nodal graph; when both the acquisition location of the assigned bolt and the adjacent acquisition locations show an increase relative to their respective historical records of no loosening, it is considered a local response change that increases in the same direction.

[0053] The changes in the state of the connection points in the strain record include at least the strain changes near the boundary of the connection plate or node plate, and the strain change trends of adjacent acquisition locations within the same connection unit. The strain acquisition location is determined based on the strain record number, installation location, name of the connection plate or node plate, and corresponding connection point; when the coverage area of ​​the strain record includes the hole row where the assigned bolt is located or an adjacent hole row, and the location of the state change falls within the connection unit where the assigned bolt is located, the state change of the connection point is considered to cover the location of the assigned bolt.

[0054] In one embodiment, a bolt belonging to a valve hall steel structure is located within a connection unit formed by the same node plate. During the analysis period, inspection images, vibration records, and maintenance records exist. The inspection images show that the marker line has shifted towards the bolt loosening direction relative to historical records of no loosening. The vibration records do not show a local response change that increases in the same direction at adjacent acquisition locations. The maintenance records only record confirmation after tightening. During processing, the inspection images point to signs of loosening, while the vibration records and maintenance records do not. Therefore, the loosening status value of this bolt is determined to be a suspected abnormal state. If the same bolt simultaneously has a marker line shifting towards the bolt loosening direction and a record of abnormality upon verification, and there is no record of confirmation after tightening, then the loosening status value of this bolt is determined to be an abnormal state.

[0055] After completing the above judgment, if at least two data sources indicate loosening of the same bolt within the same analysis period, the loosening status value is determined to be an abnormal state; if only one data source indicates loosening, the loosening status value is determined to be a suspected abnormal state; if none of the available data sources indicate loosening, the loosening status value is determined to be a normal state. Only one loosening status value is retained for each bolt within the same analysis period.

[0056] After obtaining the loosening state value, the loosening state value is organized according to the connection unit and bolt connection relationship in the node diagram to form a loosening state set of connection unit. During processing, the belonging bolts contained in each connection unit are read first, and then the bolt connection relationship between the belonging bolts is read. The loosening state value of each belonging bolt is attached to the corresponding belonging bolt and arranged according to the hole column order and the adjacent relationship of the cross-hole column within the same connection unit.

[0057] The loosening state set of the connection unit records the connection unit, the assigned bolt, the loosening state value, the data source of the current state record, the collection time, and the bolt connection relationship. Assigned bolts with a loosening state value of abnormal or suspected abnormal are used to indicate locations within the connection unit where loosening may be present. Assigned bolts with a loosening state value of normal are also retained to indicate locations within the connection unit where the abnormal continuation may be interrupted. If there is no bolt connection between two assigned bolts, even if both assigned bolts have abnormal loosening state values, the connection unit loosening state set does not record their adjacent continuation relationship. If there is a bolt connection between two assigned bolts, and both have abnormal loosening state values, or one is abnormal and the other is suspected, the connection unit loosening state set records that they meet the candidate conditions for continuous abnormality.

[0058] After the loose state set of the connection unit is formed, the loose state value in each connection unit is no longer an isolated record, but is presented as an ordered state sequence according to the bolt connection relationship in the node diagram. The normal state position, the suspected abnormal state position and the abnormal state position are all retained, so that S3 can extract continuous abnormalities from the loose state set of the connection unit and determine whether the continuous abnormality crosses the boundary of the connection plate or the node plate.

[0059] After S2 processing, the bolt loosening feature data is first grouped into the current state record of the assigned bolt according to the analysis period. Then, based on the pointing relationship of loosening signs from different data sources, a unique loosening state value is formed. Finally, a loosening state set of the connection unit is formed according to the connection unit and the bolt connection relationship. The loosening state set of the connection unit output in this step can clearly express the abnormal position, suspected abnormal position, normal interruption position and adjacent continuation condition within the connection unit.

[0060] Detailed implementation of step S3: S3 inherits the set of loose connection unit states formed by S2. The input objects are connection units, associated bolts, loose state values, and bolt connection relationships. The processing goal is to identify the co-occurring abnormal segments that can express grouped looseness from within the connection units. However, continuous abnormalities that cross the boundary of the connection plate or node plate may be a continuation in the same force transmission direction or abnormalities from different connection parts may be linked together. Therefore, this step needs to distinguish the force transmission direction and boundary direction of the cross-boundary part after extracting continuous abnormalities.

[0061] First, the connection units, assigned bolts, loosening status values, and bolt connection relationships in the loosening status set of the connection units are read, and continuous anomalies within the connection units are extracted on a unit-by-unit basis. Within each connection unit, the arrangement order of the assigned bolts is read according to the adjacent relationships of the same hole column and the adjacent relationships of cross-hole columns recorded in the node diagram, and then the loosening status values ​​of adjacent assigned bolts are checked one by one. Assigned bolts with loosening status values ​​of abnormal or suspected abnormal states are designated as candidate bolts for anomalies, and assigned bolts with loosening status values ​​of normal states are designated as interrupted bolts. When there is a bolt connection relationship between two adjacent assigned bolts, and both adjacent assigned bolts are candidate bolts for anomalies, the two adjacent assigned bolts are included in the continuous anomalies within the same connection unit. When there is no bolt connection relationship between two adjacent assigned bolts, or when there is an interrupted bolt between two adjacent assigned bolts, the continuous anomalies within the connection unit terminate at the corresponding position. After extracting continuous anomalies within the connection unit, the boundary adjacency relationships in the node graph are read. If the bolts on both sides of the boundary of the connection plate or node plate are connected through boundary adjacency relationships, and both bolts on both sides of the boundary are candidate bolts for anomalies, then the continuous anomalies within the connection unit corresponding to both sides of the boundary are merged and marked as cross-boundary continuous anomalies. If there are no boundary adjacency relationships on both sides of the boundary, or if the bolts on either side of the boundary are in a normal state, then cross-boundary continuous anomalies are not formed. Continuous anomalies within the connection unit that do not form cross-boundary continuous anomalies are directly identified as anomaly co-occurrence segments. When cross-boundary continuous anomalies are formed, they are processed by calculating the force transmission direction anomaly penetration degree and the boundary direction anomaly fragmentation degree.

[0062] For each cross-boundary continuous anomaly, structural role information and bolt connection relationships are first read from the node map, and the target bolt chain segment is determined within the region where the cross-boundary continuous anomaly is located. The target bolt chain segment refers to a sequence of bolts that are continuously connected along the force transmission direction corresponding to the structural role information, including the starting and ending bolts of the cross-boundary continuous anomaly, and linked by bolt connection relationships. When multiple candidate bolt chains exist, the candidate bolt chain segment with the most overlapping bolts with the cross-boundary continuous anomaly is selected; if the number of overlapping bolts is the same, the candidate bolt chain segment passing through the boundary of the connecting plate or node plate is selected.

[0063] The direction of force transmission is read from the steel structure design drawings, construction details, or stress-bearing part markings corresponding to the structural role information; if the direction of force transmission is not directly marked in the design drawings, the extension direction of the main hole column in the connection unit shall be taken as the direction of force transmission.

[0064] The direction of force transmission anomaly continuity is determined by comparing the number of connection relationships. Specifically, the number of all adjacent bolt connections that the target bolt chain segment must traverse from the starting bolt to the ending bolt is first counted. Then, the number of adjacent bolt connections actually covered by the cross-boundary continuous anomaly is counted. If both numbers are the same, and there are no interrupted bolts in the target bolt chain segment, the force transmission anomaly continuity indicates continuous development. If the number of actually covered adjacent bolt connections is less than the total number of adjacent bolt connections the target bolt chain segment must traverse, or if there are interrupted bolts in the target bolt chain segment, the force transmission anomaly continuity indicates discontinuous development.

[0065] The force transmission anomaly penetration is a discrete state quantity, and the calculation results include continuous and discontinuous expansion. The force transmission anomaly penetration does not represent physical length or mechanical strength, but rather whether the cross-boundary continuous anomaly completely covers the adjacent bolt connection relationships that the target bolt chain segment should pass through.

[0066] The degree of boundary anomaly is obtained by comparing the state levels of cross-boundary state segments and same-side state segments. A cross-boundary state segment refers to a state segment formed by two adjacent state bolts located on both sides of the boundary of a connecting plate or node plate in a continuous anomaly and connected by bolts. A same-side state segment refers to a state segment formed by adjacent state bolts located on both sides of a cross-boundary state segment within the same foundation boundary and connected to the cross-boundary state segment. Normal state, suspected anomaly state, and anomaly state are ordered into state levels in ascending order.

[0067] When calculating the boundary-to-abnormal cut, first compare the loose state values ​​at both ends of the cross-boundary state segment; if the ordered state levels at both ends of the cross-boundary state segment are different, it is recorded as a jump in the cross-boundary state; if the ordered state levels at both ends of the cross-boundary state segment are the same, it is recorded as no jump in the cross-boundary state. Then, read the available same-side state segments; if same-side state segments exist on both sides of the boundary, and if a jump occurs in the cross-boundary state, and both same-side state segments on both sides maintain the same ordered state level, then the boundary-to-abnormal cut indicates that the boundary is cut off; if no jump occurs in the cross-boundary state, even if both same-side state segments on both sides maintain the same ordered state level, the boundary-to-abnormal cut still indicates that the boundary is not cut off. When there is a state segment on the same side only on one side of the boundary, if the state across the boundary changes abruptly and the existing state segments on the same side maintain the same ordered state level, then the boundary is cut off to the boundary along the abnormal separation degree. When there are no state segments on the same side on either side of the boundary, the boundary is not cut off to the boundary along the abnormal separation degree. When there is no state change across the boundary, the boundary is not cut off to the boundary along the abnormal separation degree. After the above processing, the continuous anomaly across the boundary obtains the continuation result in the force transmission direction and the segmentation result in the boundary direction.

[0068] In one embodiment, a cross-boundary continuous anomaly exists near the node plate of the valve hall steel structure. This continuous anomaly includes four assigned bolts: the first two are located on one side of the boundary of the connecting plate or node plate, and the latter two are located on the other side. The target bolt chain segment, from the starting bolt to the ending bolt, requires passing through three adjacent bolt connections. The continuous anomaly actually covers these three adjacent bolt connections, and there are no assigned bolts in the target bolt chain segment in a normal state. Therefore, the force transmission direction towards the anomaly continuity indicates continuous development. The two ends of the cross-boundary state segment represent an abnormal state and a suspected abnormal state, respectively, with different levels of ordered state. The state segments on the same side of the boundary maintain the continuity of the abnormal state and the continuity of the suspected abnormal state, respectively. Therefore, the boundary direction towards the anomaly discontinuity indicates boundary cutoff. At this point, the cross-boundary continuous anomaly simultaneously exhibits continuity in the force transmission direction and discontinuity in the boundary direction, requiring reselection of the conflict segment.

[0069] The system reads the force transmission direction anomaly continuity and the boundary direction anomaly severance, and reselects continuous anomalies based on these two directional results. When the force transmission direction anomaly continuity points to continuous expansion, and the boundary direction anomaly severance points to the boundary without cutting off, the continuous anomaly is retained as an anomaly co-occurrence segment. The anomaly co-occurrence segment inherits the bolt number, loosening state value, starting bolt, and ending bolt of the continuous anomaly.

[0070] When the force transmission direction is continuously extended in the direction of abnormal continuity and the boundary is cut off in the direction of abnormal separation, the cross-boundary state segment is identified as a conflict segment. A conflict segment represents a state segment that can be continuously extended in the direction of force transmission, but the state level on both sides of the boundary of the connecting plate or node plate is divided. During processing, the state segments on the same side of the conflict segment are read separately, and the assigned bolts that maintain the same ordered state level as the state segments on the same side are retained in the co-occurrence segment of the same side. The co-occurrence segments of the abnormality formed on both sides of the conflict segment do not share the assigned bolts, and the boundary position is recorded as the boundary of the connecting plate or node plate.

[0071] When the force transmission direction of the abnormal penetration degree is discontinuous and the boundary direction of the abnormal separation degree is cut off from the boundary, the boundary of the connecting plate or node plate is used as the segmentation position to divide the continuous anomaly across the boundary into mutually exclusive anomaly co-occurrence segments; when the force transmission direction of the abnormal penetration degree is discontinuous and the boundary direction of the abnormal separation degree is not cut off from the boundary, the adjacent bolt connection relationship not covered by the continuous anomaly in the target bolt chain segment is read, or the interrupted bolt position in the target bolt chain segment is read, and the uncovered adjacent bolt connection relationship or interrupted bolt position is used as the segment break position, and the parts on both sides of the segment break position that are still composed of abnormal candidate bolts are reselected as anomaly co-occurrence segments respectively.

[0072] After an anomaly co-occurrence segment is formed, the segment number, its associated connection unit, the number of the associated bolt, the loosening state value of the associated bolt, the corresponding connection plate or node plate boundary, and whether the anomaly co-occurrence segment was obtained by reselection of cross-boundary continuous anomalies are recorded. Multiple anomaly co-occurrence segments formed by reselection of cross-boundary continuous anomalies do not share associated bolts. Anomaly co-occurrence segments formed by non-cross-boundary continuous anomalies retain the start and end range of the original continuous anomaly. After the above processing, the anomaly co-occurrence segment can distinguish between grouped anomalies continuing along the force direction within the connection unit and anomalies that need to be separated on both sides of the connection plate or node plate boundary.

[0073] After S3 processing, the abnormal and suspected abnormal states in the loose state set of the connection unit are first extracted as continuous abnormalities. Then, the cross-boundary continuous abnormalities are distinguished by the force transmission direction abnormality penetration degree and the boundary direction abnormality fragmentation degree, and finally an abnormal co-occurrence segment with the belonging bolt, connection unit, boundary position and reselection source is formed.

[0074] Detailed implementation of step S4: S4 follows the anomaly co-occurrence segment formed by S3. The input objects include the anomaly co-occurrence segment, connection unit, assigned bolt, loosening state value, connection plate or node plate boundary, and structural role information in the node diagram. The processing goal is to transform the group loosening from anomaly expression at the segment level to risk expression at the connection unit level. However, the anomaly co-occurrence segment itself only indicates which assigned bolts are commonly abnormal, but it cannot indicate whether the anomaly is concentrated within the connection unit, nor can it indicate which type of connection function in the converter station steel structure the anomaly corresponds to. Therefore, S4 needs to determine the degree of concentration, form the associated risk value of the connection unit, and reflect the structural role information in sequence.

[0075] First, read the abnormal co-occurrence fragments output by S3. The abnormal co-occurrence fragments have recorded the abnormal co-occurrence fragment number, the connection unit to which it belongs, the number of the bolts it belongs to, the loosening status value of the bolts, the corresponding connection plate or node plate boundary, and whether the abnormal co-occurrence fragment was obtained by reselection of cross-boundary continuous abnormalities. After reading the above information, read the bolt connection relationship, hole row affiliation, plate position, and structural role information within the corresponding connection unit from the node diagram.

[0076] The determination of concentration begins with continuous coverage relationships. For an abnormal co-occurrence segment, the bolts included in the segment are checked along the bolt connection relationships recorded in the node map. If there is a bolt connection relationship between any adjacent bolts in the abnormal co-occurrence segment, and there are no bolts in the segment with a normal loose state, the segment is considered continuous coverage. If there are bolts in the segment that are not connected by bolt connections, or if there are bolts in the segment with a normal state, the segment is not considered continuous coverage. In this case, the segments are processed separately according to the mutually exclusive abnormal co-occurrence segments already formed in S3, and the mutually exclusive abnormal co-occurrence segments are not re-merged.

[0077] After the continuous coverage relationship is determined, the position of the abnormal co-occurrence segment relative to the boundary of the connecting plate or node plate is read. In specific processing, the starting bolt and ending bolt of the abnormal co-occurrence segment are traced along the bolt connection relationship to the nearest boundary of the connecting plate or node plate, and the number of adjacent bolt connection relationships passed through is recorded. If there are no other belonging bolts between the starting bolt or ending bolt of the abnormal co-occurrence segment and the boundary of the connecting plate or node plate, or only adjacent belonging bolts in the same hole column are separated, the abnormal co-occurrence segment is identified as close to the boundary. If there are other belonging bolts between both ends of the abnormal co-occurrence segment and the boundary of the connecting plate or node plate, and the abnormal co-occurrence segment is within the hole column range inside the connecting unit, the abnormal co-occurrence segment is identified as located inside the connecting unit.

[0078] The concentration level is determined by both the continuous coverage relationship and the boundary position. When an abnormal co-occurring segment forms a continuous coverage and is close to the boundary of the connecting plate or node plate, the concentration level is determined as boundary concentration. When an abnormal co-occurring segment forms a continuous coverage and is located inside the connecting unit, the concentration level is determined as internal concentration. When there are multiple mutually exclusive abnormal co-occurring segments within the same connecting unit, the concentration level of each mutually exclusive abnormal co-occurring segment is determined separately. When multiple mutually exclusive abnormal co-occurring segments are located in the same or adjacent hole columns, and the segments are separated only by the bolts belonging to the normal state, the concentration level of each mutually exclusive abnormal co-occurring segment is retained, and the same column distribution or adjacent column distribution is additionally recorded. After the above processing, each abnormal co-occurring segment has a clear concentration level, which reflects both the continuity of the abnormal co-occurring segment in the bolt connection relationship and its position relative to the boundary of the connecting plate or node plate.

[0079] Concentration is a discrete state quantity, including at least boundary concentration and internal concentration; concentration does not represent geometric density value, but rather the concentrated location of anomalous co-occurring segments relative to the boundary of the connecting plate or node plate.

[0080] After obtaining the concentration level, the loosening status value of each assigned bolt within the abnormal co-occurrence segment is read, and an ordered status level is formed from low to high: normal state, suspected abnormal state, and abnormal state. The abnormal co-occurrence segment should not contain assigned bolts in the normal state; if an assigned bolt in the normal state is read, the location of the normal state is taken as the segment break point, and the process of determining the concentration level is returned to the above concentration level, and the concentration level is re-determined according to the mutually exclusive abnormal co-occurrence segments after the break.

[0081] The risk level of a segment is determined based on the ordered state hierarchy within the co-occurring anomaly segment. If a co-occurring anomaly segment contains consecutive adjacent anomaly states, and these consecutive adjacent anomaly states are not separated by suspected anomaly states, then the segment risk level is determined to be high-level segment risk. If a co-occurring anomaly segment does not contain consecutive adjacent anomaly states, but anomaly states and suspected anomaly states occur adjacent to each other, then the segment risk level is determined to be mid-level segment risk. If a co-occurring anomaly segment only contains suspected anomaly states, then the segment risk level is determined to be low-level segment risk. The above segment risk levels, together with the concentration level, constitute the segment risk expression. The segment risk expression also records the starting bolt, ending bolt, hole row, and boundary position of the connecting plate or node plate of the co-occurring anomaly segment.

[0082] The risk value associated with a connection unit is a graded score, which includes at least the segment risk level, concentration level, location of co-occurring anomalies, and source of the co-occurring anomalies. The segment risk level indicates the strength of the loosening state; the concentration level indicates whether the anomaly appears in clusters within the connection unit; the location of the co-occurring anomaly indicates whether the risk is located inside the connection unit or near the boundary of the connection plate or node plate; and the source of the co-occurring anomaly indicates whether the risk was obtained by reselection from consecutive anomalies across boundaries. Low-level segment risk corresponds to level 1 of the connection risk score, mid-level segment risk corresponds to level 2, and high-level segment risk corresponds to level 3. If multiple mutually exclusive co-occurring anomalies exist within the same connection unit, the connection risk score uses the segment risk expression with the highest risk level, while retaining the location records of the remaining mutually exclusive co-occurring anomalies. Levels 1, 2, and 3 in the connection risk score only indicate the degree of risk from low to high, and do not represent the specific probability of mechanical failure.

[0083] In one embodiment, the equipment bracket connection unit includes 5 assigned bolts. S3 forms an abnormal co-occurrence segment, which covers 3 consecutive assigned bolts. All 3 assigned bolts are bolted together, and their loosening state values ​​are, in order, abnormal state, abnormal state, and suspected abnormal state. S4 confirms that the abnormal co-occurrence segment forms a continuous coverage, and the starting bolt is close to the boundary of the connection plate or node plate, with the concentration level determined to be boundary concentration. Subsequently, the ordered state level arrangement is read, and it is found that there are consecutive adjacent abnormal states within the abnormal co-occurrence segment. The segment risk level is determined to be a high-level segment risk. Since there are no other mutually exclusive abnormal co-occurrence segments within the same connection unit, the segment risk expression is directly used as the associated risk value of the connection unit.

[0084] When there is only one co-occurring abnormal segment within the same connection unit, the segment risk expression is used as the associated risk value of the connection unit. When there are multiple mutually exclusive co-occurring abnormal segments within the same connection unit, the segment risk expression for each mutually exclusive co-occurring abnormal segment is first formed, and then the associated risk value of the connection unit is determined based on the position of the mutually exclusive co-occurring abnormal segments in the bolt connection relationship. If multiple mutually exclusive co-occurring abnormal segments are located in the same or adjacent hole columns, and at least one segment risk expression is a high-level segment risk, then the associated risk value of the connection unit adopts the segment risk expression corresponding to the high-level segment risk, and the segment numbers of other mutually exclusive co-occurring abnormal segments are recorded. The connection unit's associated risk value records the number and distribution location; if multiple mutually exclusive co-occurring anomalies are all median or low-level segment risks, the connection unit's associated risk value uses the segment risk expression with the higher risk level, and retains the location records of all mutually exclusive co-occurring anomalies; if multiple mutually exclusive co-occurring anomalies are distributed in different well columns and there is no adjacent well column relationship, the connection unit's associated risk value records the main risk expression and the segment records in the columns. The main risk expression comes from the segment risk expression with the highest risk level. After completing the above processing, the connection unit's associated risk value can express the strength, concentration, and spatial distribution of the co-occurring anomalies within the connection unit.

[0085] Read the associated risk value of the connection unit and the structural role information of the corresponding connection unit in the node graph. Since S1 has already divided the connection units according to the foundation boundary and structural role information, the same connection unit corresponds to only one structural role information; the structural role information is used to indicate the connection function of the connection unit in the valve hall steel structure, equipment support, busbar support or maintenance platform.

[0086] The feedback process first verifies whether all abnormal co-occurring segments in the associated risk values ​​of the connection unit belong to the same connection unit. If the connection unit to which the abnormal co-occurring segment belongs is consistent with the connection unit recorded in the node diagram, and the structural role information of the bolt covered by the abnormal co-occurring segment is consistent with the structural role information of the connection unit, then the associated risk value of the connection unit is assigned to the corresponding structural role information to form the connection unit risk value. If the structural role information of the bolt covered by the abnormal co-occurring segment is inconsistent with the structural role information of the connection unit, then the source record of the associated risk value of the connection unit is not changed, but the position of inconsistent structural role is recorded as the structural role verification result and saved together with the connection unit risk value.

[0087] The risk value of a connection unit includes at least the connection unit number, structural role information, associated risk value of the connection unit, number of abnormal co-occurring segments, concentration level, segment risk level, boundary location of the connection plate or node plate, and structural role verification result. The risk value of the connection unit does not re-merge the segment risk expressions under different structural role information, nor does it change the connection unit division result already formed in S1. After the above processing, the risk value of the connection unit can point to the specific connection function in the converter station steel structure and retain the source information of the risk expression formed by abnormal co-occurring segments within the connection unit.

[0088] After S4 processing, the co-occurring abnormal fragments are first arranged according to their continuous coverage relationship and boundary position to form a concentration degree, and then combined with the loosening state value to form the connection unit associated risk value. Finally, the structural role information in the node map is reflected to form the connection unit risk value. The output result not only retains the fragment source of the co-occurring abnormal fragments, but also expresses the connection function of the converter station steel structure corresponding to the group of loosening.

[0089] Detailed implementation of step S5: S5 inherits the connection unit risk value formed by S4. The input objects include connection unit risk value, connection unit associated risk value, abnormal co-occurrence segment, concentration degree, segment risk level, connection plate or node plate boundary position and structural role information. The processing goal is to verify the current connection unit risk value by using the historical treatment result range of similar connection units. However, there may be cross-boundary anomalies, recovery after tightening, abnormal residues after tightening, and deviations in structural role at the connection parts of the converter station steel structure. If the connection risk score is not constrained by the historical treatment result range, the connection risk score may only reflect the current detection status.

[0090] First, read the connection unit risk value output by S4, and extract the connection unit number, structural role information, connection unit associated risk value, abnormal co-occurrence segment number, concentration degree, segment risk level, connection plate or node plate boundary position and structural role verification result from the connection unit risk value; after reading, read the foundation boundary, bolt connection relationship, hole row attribution and plate position of the corresponding connection unit from the node diagram to form the screening basis for similar connection units.

[0091] For a connection unit to be considered a "same type" connection unit, it must simultaneously meet the following criteria: identical structural role information, identical foundation boundary type, and corresponding bolt connection relationships. Identical structural role information means that the current connection unit and the candidate historical connection unit belong to the same connection function within the valve hall steel structure, equipment support, busbar support, or maintenance platform. Identical foundation boundary type means that both the current connection unit and the candidate historical connection unit use a connection plate or a node plate as their foundation boundary. Corresponding bolt connection relationships mean that the current connection unit and the candidate historical connection unit have the same number of hole rows, the same sequence of adjacent bolt connections within the same hole row, and the same position of adjacent bolts across hole rows. When all three conditions are met, the candidate historical connection unit is determined to be a "same type" connection unit; if any condition is not met, the candidate historical connection unit will not be included in this consistency check.

[0092] Once similar connection units are identified, their historical handling results are retrieved. These results are derived from review records, tightening records, and processing records. The retrieved content includes the connection unit's risk value before handling, the pre-handling co-occurring abnormal fragment, the post-handling risk value, changes in the post-handling co-occurring abnormal fragment, and a post-handling confirmation record. The scope of historical handling results consists of closed-loop handling records. A closed-loop handling record refers to a historical handling result for the same connection unit that simultaneously possesses the pre-handling risk value, pre-handling co-occurring abnormal fragment, post-handling risk value, changes in the post-handling co-occurring abnormal fragment, and a post-handling confirmation record. Historical handling results lacking a post-handling confirmation record are not included in the historical handling result scope.

[0093] When the scope of historical disposal results is formed, each closed disposal record is organized into a verifiable record. The verifiable record includes the risk level of the segment before disposal, the concentration level before disposal, the boundary position of the connecting plate or node plate before disposal, whether the abnormal co-occurring segment before disposal was obtained by cross-boundary continuous abnormal reselection, the change of the abnormal co-occurring segment after disposal, the change of the risk value of the connecting unit after disposal, and the confirmation record after disposal. After the above processing is completed, the scope of historical disposal results only includes records that have the same structural relationship as the current connecting unit and whose disposal process is closed.

[0094] Read the current connection unit risk value and the historical handling result range, and decompose the current connection unit risk value into the current verification content; the current verification content includes the segment risk level, concentration degree, the boundary position of the connection plate or node plate corresponding to the abnormal co-occurring segment, whether the abnormal co-occurring segment is obtained by cross-boundary continuous abnormal reselection, and the structural role verification result; the current verification content corresponds to the verifiable records in the historical handling result range one by one to obtain the verification difference.

[0095] The verification differences are recorded according to the differences in risk level, fragment morphology, and structural role. The difference in risk level is used to indicate whether the risk level of the fragment in the current connection unit risk value can be found in the range of historical treatment results for the connection unit risk value with the same fragment risk level before treatment. The difference in fragment morphology is used to indicate whether the concentration of the current co-occurring abnormal fragments, the boundary position of the connection plate or node plate, and the source of cross-boundary continuous abnormal reselection can be found in the range of historical treatment results for the corresponding record of the co-occurring abnormal fragments before treatment. The difference in structural role is used to indicate whether the structural role verification results show that the structural role information of the bolt structure covered by the co-occurring abnormal fragment is consistent with the structural role information of the connection unit.

[0096] When the differences in risk level, fragment morphology, and structural role all correspond to the range of historical treatment results, the verification difference falls within the range of historical treatment results for the same type of connecting unit. When the differences in risk level correspond but the differences in fragment morphology do not, it indicates that the risk level of the current connecting unit's risk value matches the records of the same type of treatment, but the morphology of the abnormal co-occurring fragment near the boundary of the connecting plate or node plate does not conform to the range of historical treatment results, and the verification difference does not fall within the range of historical treatment results for the same type of connecting unit. When the differences in structural role are inconsistent, the verification difference does not fall within the range of historical treatment results for the same type of connecting unit, and the structural role difference is used as one of the criteria for boundary updates. When the range of historical treatment results is empty, the verification difference is recorded as not falling within the range of historical treatment results for the same type of connecting unit, and the source record of the empty range is retained.

[0097] In one embodiment, the risk value of the equipment bracket connection unit includes high-level segment risk, boundary concentration, and cross-boundary continuous anomaly reselection source. The boundary position of the connection plate or node plate corresponds to the connection part of the equipment bracket. Among the historical handling results of similar connection units obtained through screening, there are records of equipment bracket connection units that exhibited high-level segment risk and boundary concentration before handling, and whose anomaly co-occurring segments shortened and transformed into a closed state within a single connection unit after handling. The current segment risk level, concentration degree, connection plate or node plate boundary position, and cross-boundary continuous anomaly reselection source are respectively matched with verifiable records. If the verifiable records contain the same segment risk level, the same concentration degree, and the same cross-boundary continuous anomaly reselection source, and the structural role verification results are consistent, then the verification difference falls within the historical handling results range of similar connection units. If the same segment risk level in the verifiable records all correspond to a state within a single connection unit, but the current anomaly co-occurring segment is still a cross-boundary continuous anomaly reselection source, then the segment morphology difference cannot be matched, and the verification difference does not fall within the historical handling results range of similar connection units.

[0098] After completing the above processing, the verification difference can clearly show the correspondence between the current connection unit risk value and the historical treatment result range of similar connection units, and can indicate the source of the difference from risk level, abnormal co-occurrence fragment morphology or structural role information.

[0099] The system reads the verification difference and, based on this difference, determines the output connection risk score or updates the boundary of the connection board or node board before returning to S2. When the verification difference falls within the historical handling result range of similar connection units, the current connection unit risk value is converted into a connection risk score result. During the conversion, the fragment risk level in the connection unit risk value is read. When the fragment risk level is low, the connection risk score result is output as level 1; when the fragment risk level is mid, the connection risk score result is output as level 2; and when the fragment risk level is high, the connection risk score result is output as level 3. When multiple mutually exclusive anomaly co-occurring fragments exist in the same connection unit, the connection risk score result uses the fragment risk expression with the highest risk level to represent the corresponding level score, while retaining the position records of the remaining mutually exclusive anomaly co-occurring fragments. The connection risk score result retains the connection unit number, structural role information, connection unit risk value, anomaly co-occurring fragment number, verification difference, and corresponding historical handling result range identifier, enabling the connection risk score result to be traced back to the closed handling record of similar connection units.

[0100] When the verification difference does not fall within the historical processing result range of the same type of connection unit, the verification difference type is read first. If the verification difference includes fragment morphology differences, and the abnormal co-occurring fragments are obtained by reselection of continuous anomalies across the boundary, the boundary update position is taken as the boundary plate or node plate boundary position corresponding to the abnormal co-occurring fragments. During the update, the same-side state segments on both sides of the cross-boundary state segment are read. The assigned bolts that maintain the same ordered state level as the same-side state segments on one side of the boundary are assigned to the connection unit on that side, and the assigned bolts that maintain the same ordered state level as the same-side state segments on the other side of the boundary are assigned to the connection unit on the other side. The base boundary and connection unit number of the assigned bolts are updated synchronously in the node diagram. When the verification difference includes structural role differences, the assigned bolts whose structural role verification results are inconsistent are taken as the boundary update objects, and the structural role information of the assigned bolts is reread. If the reread structural role information is inconsistent with the current connection unit, the assigned bolt is removed from the current connection unit and assigned to the connection unit with consistent structural role information and corresponding base boundary. If the verification difference only includes the risk level difference and the abnormal co-occurrence segment has not been converted to the state within a single connection unit, perform boundary attribution verification along the boundary position of the connection plate or node plate corresponding to the abnormal co-occurrence segment. The verification content includes the bolt number attribution on both sides of the boundary, the name of the foundation boundary, the structural role information and the boundary adjacency relationship; return to S2 after the verification is completed.

[0101] Each update to the boundary of a connection board or node board generates a boundary version record. This record includes at least the boundary position before the update, the boundary position after the update, the bolt number to which the new connection unit is assigned, and the type of verification difference that triggered the update. Upon returning to S2, if the same verification difference points to the same boundary position again, the same boundary update is not repeated. Instead, boundary assignment is checked along the boundary of the connection board or node board on the other side of the co-occurring anomaly segment. This boundary version record prevents the process from being repeatedly updated at the same boundary position, which could prevent the process from ending.

[0102] After updating the boundary of the connection plate or node plate, the processing flow returns to S2. S2 regenerates the loose state set of connection units based on the updated node map, S3 regenerates the abnormal co-occurrence fragments, S4 regenerates the risk value of the connection unit, and S5 performs consistency verification again. There are two stopping conditions. The first stopping condition is that the verification difference after re-verification falls within the historical handling result range of the same type of connection unit. The second stopping condition is that the abnormal co-occurrence fragment has been converted to a state within a single connection unit, and updating the boundary of the connection plate or node plate again will not change the ownership bolts contained in the abnormal co-occurrence fragment. When either stopping condition is met, the connection risk score result is output. When the historical handling result range is empty and the abnormal co-occurrence fragment has been converted to a state within a single connection unit, the output connection risk score result retains the verification difference and empty range source record, and does not return to S2.

[0103] After the above processing, the connection risk score result is not the direct output of the current connection unit risk value, but is formed after the range of historical treatment results of similar connection units is verified; when the verification difference points to the deviation of the boundary of the connection plate or node plate, the boundary update causes the connection unit loose state set, abnormal co-occurrence fragments and connection unit risk value to be recalculated.

[0104] After processing by S5, the risk value of the connection unit is first checked for consistency with the historical handling results range of similar connection units. Then, the connection board or node board boundary is updated according to the check difference and returned to S2. This process continues until the check difference falls within the historical handling results range, or the abnormal co-occurrence fragment is converted into a state within a single connection unit before the connection risk score result is output.

[0105] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for risk assessment of steel structure connections based on the characteristics of loose bolts in converter stations, characterized in that, include: S1. Obtain bolt loosening characteristic data at the connection points of the converter station steel structure, divide the connection units based on the connection plate or node plate as the basic boundary and in combination with the structural role information, determine the bolt connection relationship between bolts within the connection unit, and form a node map containing bolt connection relationship and structural role information; the structural role information indicates the connection function of the connection point where the bolt is located in the converter station steel structure. S2. Based on the bolt connection relationships in the node diagram, convert the bolt loosening characteristic data into the corresponding bolt loosening state values, and aggregate the loosening state values ​​by connection unit to form a loosening state set of connection unit; the loosening state values ​​are classified as normal state, suspected abnormal state, and abnormal state. S3. Extract continuous anomalies from the loose state set of the connection unit. When a continuous anomaly crosses the boundary of the connection plate or node plate, calculate the force transmission direction anomaly penetration degree and the boundary direction anomaly separation degree. Based on the direction results of the force transmission direction anomaly penetration degree and the boundary direction anomaly separation degree, reselect the continuous anomaly as an anomaly co-occurrence segment; otherwise, determine the continuous anomaly as an anomaly co-occurrence segment. The force transmission direction anomaly penetration degree is a discrete state quantity. The calculation result includes continuous expansion and discontinuous expansion, indicating whether the cross-boundary continuous anomaly completely covers the adjacent bolt connection relationship that should be passed through in the target bolt chain segment. The boundary direction anomaly separation degree is determined by comparing the state level of the cross-boundary state segment with the state segment on the same side. It is divided into two types of direction: boundary cut-off and boundary uncut-off. It is used to determine whether there is a level jump in the loose state on both sides of the boundary and whether the anomaly segment should be divided according to the boundary. The anomaly co-occurrence segment is obtained from the continuous anomaly after reselection. It is used to distinguish the grouped anomalies that continue along the force direction inside the connection unit and the anomalies that need to be separated on both sides of the boundary of the connection plate or node plate. S4. Calculate the associated risk value of the connection unit based on the concentration of abnormal co-occurring segments in the connection unit, and then map the associated risk value of the connection unit back to the structural role information in the node graph to form the risk value of the connection unit. S5. Perform consistency verification between the risk value of the connection unit and the historical handling results of similar connection units to obtain the verification difference. When the verification difference falls within the range of the historical handling results of similar connection units, output the connection risk score result. Otherwise, update the boundary of the connection board or node board based on the verification difference and return to S2 until the verification difference falls within the historical processing result range of the same type of connection unit, or the abnormal co-occurrence fragment is converted into the state within a single connection unit and the connection risk score result is output.

2. The steel structure connection risk scoring method based on converter station bolt loosening characteristics according to claim 1, characterized in that, Bolt loosening characteristic data are generated separately according to inspection images, vibration records, strain records and maintenance records, and are indexed by bolt number, name of connecting plate or node plate, and collection location or maintenance location. The bolt loosening feature data generated from the inspection images include the relative position of the bolt marking line, the position of the bolt outer contour, the appearance angle of the bolt head, and the exposed state of the bolt. The bolt loosening characteristic data generated by vibration recordings include the acquisition location, acquisition time, vibration amplitude variation, and characteristic frequency band response value variation. The bolt loosening characteristic data generated by strain records include strain acquisition location, strain acquisition time, location of changes in the state of the connection, and direction of strain change. The bolt loosening characteristic data generated in the maintenance record includes bolt number, name of connecting plate or node plate, maintenance location, record of abnormality upon review, description of abnormality before tightening, and confirmation record after tightening.

3. The steel structure connection risk scoring method based on converter station bolt loosening characteristics according to claim 2, characterized in that, When bolt loosening characteristic data falls into multiple candidate bolt positions, first determine whether the data source contains the name of the connecting plate or node plate. If the name of the connecting plate or node plate is included, only candidate bolts with the same name as the connecting plate or node plate will be retained. If the name of the connecting plate or node plate is not included, all candidate bolt entry positions will be compared. The bolts assigned to a particular bolt are determined by comparing their positions in the following order: minimum sum of bidirectional position differences, minimum position difference in the direction of the foundation boundary, and minimum position difference in the direction perpendicular to the foundation boundary. The basic boundary direction is the extension direction of the hole array in the connecting plate or node plate, and the direction perpendicular to the basic boundary is the direction within the plate surface that is perpendicular to the basic boundary direction. The feature position points corresponding to the bolt loosening characteristics and the plate surface positions of the candidate bolts are projected onto the basic boundary direction and the direction perpendicular to the basic boundary, respectively. The corresponding projection position differences are the position difference in the basic boundary direction and the position difference in the direction perpendicular to the basic boundary, respectively. The sum of the two-way position differences is the sum of the values ​​of the position difference in the basic boundary direction and the position difference in the direction perpendicular to the basic boundary. If the candidate bolts cannot be distinguished by location comparison, the structural role information is read, and the candidate bolts whose structural role information matches the data source label are identified as the assigned bolts; if the structural role information still cannot distinguish the candidate bolts, the corresponding bolt loosening feature data is marked as feature data to be reviewed, and it is used to determine the loosening state value after the manual review result is formed.

4. The steel structure connection risk scoring method based on converter station bolt loosening characteristics according to claim 3, characterized in that, In the set of loose connection unit states, continuous anomalies within each connection unit are first extracted on a unit-by-unit basis. Bolts with a loosening status of abnormal or suspected abnormal are designated as abnormal candidate bolts, while bolts with a loosening status of normal are designated as interrupted bolts. When the bolts belonging to the connection plate or node plate on both sides of the boundary are connected through the boundary adjacency relationship, and both bolts belonging to the connection plate on both sides of the boundary are abnormal candidate bolts, the continuous abnormalities inside the corresponding connection unit on both sides of the boundary are merged and marked as cross-boundary continuous abnormalities. When there is no boundary adjacency relationship between the two sides of the boundary, or when the bolt on either side of the boundary is in a normal state, no cross-boundary continuous anomaly is formed, and the continuous anomaly within the connection unit is identified as an anomaly co-occurrence segment.

5. The steel structure connection risk scoring method based on converter station bolt loosening characteristics according to claim 1, characterized in that, Structural role information is read from the converter station steel structure design drawings, as-built drawings, equipment support ledgers, busbar support ledgers, maintenance platform ledgers, or 3D models; When the same connecting plate or node plate corresponds to multiple connection functions in the ledger, the bolts are assigned to the corresponding structural role information according to the bolt number, hole row, and plate position. When the structural role information of a single bolt is not directly recorded in the ledger, the structural role information is determined from the design annotations of the corresponding connection parts by using the name of the connection plate or node plate as the primary index and the hole column and plate position as the secondary index.

6. The steel structure connection risk scoring method based on converter station bolt loosening characteristics according to claim 2, characterized in that, The inspection images and vibration records are compared with the historical non-loosening records to determine whether there are signs of loosening. The contents of the strain records and maintenance records are also used to determine whether there are signs of loosening. The judgment results are then converted into the loosening status value of the corresponding bolt. The historical non-loosening records are obtained from the commissioning and acceptance images, the most recent re-tightening confirmation record, the inspection images that were manually verified as non-loosening, or the confirmation record after the same connection unit was handled. When the first execution is performed and there is no historical record of loosening, the image of the same type of connection unit that is confirmed to be loose during this manual review is used as a temporary record of loosening; When the re-tightening confirmation record is generated, the temporary unloosening record is replaced with the re-tightening confirmation record; If the maintenance record only contains a confirmation record after tightening, or if the confirmation record after tightening is later than the record of abnormality during review and the description of abnormality before tightening, the maintenance record does not indicate any signs of loosening.

7. The steel structure connection risk scoring method based on converter station bolt loosening characteristics according to claim 6, characterized in that, When the historical processing result range is empty, the verification difference is recorded as the historical processing result range that does not fall into the same type of connection unit and the source record of the empty range is retained; If the abnormal co-occurrence fragment is not converted to a state within a single connection unit, the node graph is updated according to the boundary position of the connection plate or node plate corresponding to the abnormal co-occurrence fragment and S2 is returned. If the abnormal co-occurrence fragment has been converted to a state within a single connection unit, the connection risk score result is output, retaining the verification difference and empty range source record.

8. The steel structure connection risk scoring method based on converter station bolt loosening characteristics according to claim 1, characterized in that, The risk value of the connection unit includes the segment risk level, concentration level, abnormal co-occurrence segment number, and the boundary position of the connection plate or node plate. The connection risk scoring results are output according to the segment risk level. When the segment risk level is low segment risk, the connection risk score output is level 1; When the segment risk level is median segment risk, the connectivity risk score output is level 2; When the segment risk level is high segment risk, the connectivity risk score output is level 3; The connection risk scoring results also record the connection unit number, structural role information, connection unit risk value, abnormal co-occurrence segment number, verification difference, and historical handling result range identifier.

9. The steel structure connection risk scoring method based on converter station bolt loosening characteristics according to claim 8, characterized in that, A boundary version record is generated each time the boundary of the connection board or node board is updated. The boundary version record includes at least the boundary position before the update, the boundary position after the update, the bolt number of the reassigned connection unit, and the verification difference type that triggered the update. If the same check difference points to the same boundary position again after returning to S2, the same boundary update will not be repeated, and the boundary assignment will be checked along the other side of the abnormal co-occurrence segment connecting the board or node board boundary.

10. The method for risk assessment of steel structure connections based on the characteristics of loose bolts in converter stations according to claim 1, characterized in that, The scope of historical treatment results consists of closed treatment records, which simultaneously include the risk value of the connection unit before treatment, the co-occurring abnormal fragments before treatment, the risk value of the connection unit after treatment, the changes in the co-occurring abnormal fragments after treatment, and the confirmation record after treatment. The closed-loop disposal records are compiled into verifiable records. Verifiable records include at least the fragment risk level before disposal, the concentration level before disposal, the boundary position of the connecting plate or node plate before disposal, whether the abnormal co-occurring fragments before disposal were obtained by cross-boundary continuous abnormal reselection, the changes in the abnormal co-occurring fragments after disposal, and the changes in the risk value of the connecting unit after disposal.

Images