System for cross-period next look analysis of regulatory rule error data

Abstract

The application discloses a cross-period backtracking analysis system for supervisory rules error data, belongs to the field of energy supervision data management, and comprises a data acquisition module, a traceability graph construction module, a period error analysis module and a multi-level early warning processing module.The data acquisition module constructs a compliance traceability dataset;the traceability graph construction module fuses the compliance traceability dataset, equipment association logic and personnel operation association rules, constructs a multi-dimensional weighted association relationship table, generates a cross-period association graph and a preliminary error evolution sequence;the period error analysis module is used for combining the cross-period association graph and the preliminary error evolution sequence, analyzing frequency change, period law and response characteristics through dynamic trend fitting, constructing a periodical error trend feature library and a periodical repeated error list; and the multi-level early warning processing module is used for constructing a to-be-early-warned error set, combining a comprehensive early warning grade score, triggering multi-level early warning and processing, and realizing effective monitoring and management of power plant operation errors.

Description

TECHNICAL FIELD

[0001] The present application relates to a cross-period backtracking analysis system for regulatory rule error data, belonging to the technical field of energy supervision data governance. BACKGROUND

[0002] The existing Chinese patent application with publication number CN120452449A discloses a power operation supervision method and system based on AI behavior recognition and semantic analysis, which includes arranging the control architecture and collection equipment of the power system, continuously collecting voiceprint data and video stream data; verifying the operator's identity through voiceprint recognition and performing semantic analysis on the voice command based on the power standard terminology library to judge the consistency of the command content with the system-issued task; identifying the position of the operator through visual positioning and matching the space with the target equipment interval number, and simultaneously detecting the standardization of the operation behavior through video analysis; generating a voice interaction prompt according to the instruction ambiguity, and triggering a hierarchical alarm through multi-modal data fusion, synchronously recording and storing the illegal operation data. The invention improves the efficiency of power system supervision, ensures authorized operation and reduces human errors, visual monitoring ensures safety, multi-modal fusion triggers alarm, timely responds to violations and records evidence, and supports traceable safety management.

[0003] Although the existing power operation supervision method and system based on AI behavior recognition and semantic analysis realizes continuous voiceprint and video stream data collection, operator identity voiceprint verification, voice command semantic analysis, operator position and equipment interval space matching, operation behavior standardization detection, multi-modal data fusion triggering hierarchical alarm, illegal operation data storage, and generation of front and rear audio and video evidence chain, there is a lack of correlation dimension of the subject and error type of cross-cycle error data, which cannot identify periodic repeated error problems. SUMMARY

[0004] In view of the deficiencies of the prior art, the purpose of the present application is to provide a cross-period backtracking analysis system for regulatory rule error data, which realizes full-dimensional feature extraction, trend tracing and hierarchical early warning of periodic error data, accurately locates and maintains associated causes, optimizes maintenance strategies, improves response efficiency and equipment reliability, and enhances the level of active prevention and control of power plant operation risks.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0006] The cross-period backtracking analysis system for regulatory rule error data comprises a data collection module, a traceability map construction module, a periodic error analysis module, and a multi-level early warning processing module.

[0007] The data collection module is used to collect, clean and analyze the original data of power plant operation, label four-dimensional tags, and construct compliance traceability data sets.

[0008] The traceability graph construction module is configured to fuse the compliance traceability dataset, device association logic and personnel operation association rules, construct a multi-dimensional weighted association relationship table, generate a cross-cycle association graph and a preliminary error evolution sequence;

[0009] The periodic error analysis module is configured to combine the cross-cycle association graph and the preliminary error evolution sequence, analyze frequency changes, periodic laws and response characteristics through dynamic trend fitting, construct a periodic error trend feature library and determine a periodic repeated error list;

[0010] The multi-level early warning processing module is configured to construct a to-be-early-warned error set, combine a comprehensive early warning level score, trigger multi-level early warning and perform processing.

[0011] Specifically, the specific steps of constructing the multi-dimensional weighted association relationship table include:

[0012] Obtaining power plant device association logic, identifying device types and extracting association conditions to form a device association triple of device A, device B and association conditions;

[0013] Filtering complete logs in a time period to Grouping the complete logs according to operator IDs and device numbers, and retaining basic data of each group of operators and device associations;

[0014] Based on the basic data, the time sequence co-occurrence frequency of error types appearing in a time window is counted, the association strength is calculated in combination with the total number of operations, and the association confidence is normalized;

[0015] According to device-device, operator-device and operator-error type associations, a structured association rule library containing device association triples and association confidence is constructed.

[0016] Specifically, the specific steps of constructing the multi-dimensional weighted association relationship table further include:

[0017] Traversing each compliance traceability record of the compliance traceability dataset;

[0018] For device-device association types, operation time sequences and state sequences are extracted, condition compliance degrees are calculated by alignment analysis in combination with time sequence synchronicity scores and state response matching scores;

[0019] For operator-device and operator-error type association types, the number of matching fields of each compliance traceability record and the structured association rule library is counted respectively, and association matching degrees and behavior matching degrees are calculated in combination with the total number of compared fields;

[0020] The condition compliance degrees, the association matching degrees and the behavior matching degrees are integrated to calculate a comprehensive matching degree, and an association intermediate table is generated;

[0021] For each intermediate matching record in the association intermediate table, the association weight is calculated by combining the association confidence and the comprehensive matching degree;

[0022] The core dimension, the association type and the association weight of each intermediate matching record are integrated to construct a multi-dimensional weighted association relationship table.

[0023] Specifically, the specific steps of generating a cross-period association graph include:

[0024] Based on the multi-dimensional weighted association relationship table, operator, device and error type node initialization is completed;

[0025] Directed edges are constructed for the three types of association types to form a basic association graph, and the basic association graph is split into period subgraphs;

[0026] The directed edge weight distribution characteristics of each period subgraph are calculated to generate a subgraph set, and the same type of node pairs are screened by traversing and comparing adjacent period subgraphs;

[0027] If the edge weight of the same type of node pair in the adjacent period subgraph is less than a preset weight threshold, it is determined as a non-cross-period association candidate pair; otherwise, it is determined as a cross-period association candidate pair, the former bridge weight and the latter bridge weight are calculated, the sum is obtained to get the cross-period association strength, a bridge edge is generated and an intermediate graph is formed;

[0028] Clusters are divided by label propagation between nodes, strong association edges with weights exceeding a preset maximum threshold are screened in the clusters, and association paths are retained; and weak association edges with weights lower than a preset minimum threshold are screened and accidental associations are removed;

[0029] A time dimension label is added to the intermediate graph to generate a cross-period association graph.

[0030] Specifically, the specific steps of generating a preliminary error evolution sequence include:

[0031] Association edges and auxiliary features are extracted from the cross-period association graph to generate an original time series association dataset;

[0032] The time series data in the operator ID, device and error type operation error triplets are divided into sliding windows in ascending order of period;

[0033] Based on the sliding window, numerical value features and one-hot encoded category features are normalized and spliced into fixed length feature vectors to generate a time series feature sequence set;

[0034] The self-attention mechanism is introduced to calculate attention weights, and the top n windows are retained as key sequence fragments by descending order sorting;

[0035] According to the operation error triplet, the key sequence fragments are traversed, target fragments with the same operator ID, matching device fragments with the same device number, and matching error fragments with the same error type are screened respectively, and a preliminary error evolution sequence is spliced according to the chronological order of the periods.

[0036] Specifically, the specific steps of constructing the periodic error trend feature library include:

[0037] A first grouping is constructed by device type, operator ID, period label, and error type, error frequencies are counted, and device maintenance states are extracted;

[0038] According to the first grouping, sequence fragments in the preliminary error evolution sequence are located, effective error occurrence timestamps and effective repair completion timestamps are extracted, and response durations are calculated;

[0039] In combination with the first grouping, a second grouping is constructed by device type, post, period label, and error type;

[0040] The error frequencies, the device maintenance states, and the response durations are associated to form periodic error basic feature data;

[0041] The historical error occurrence rates and fluctuation ranges of the same type of device and the same post are taken as historical baselines, are associated to the second grouping, and a periodic error basic data set is generated.

[0042] Specifically, the specific steps of constructing the periodic error trend feature library further include:

[0043] For each device type, post, and error type in the periodic error basic data set, an adjacent period error frequency year-on-year growth rate is calculated in ascending order of the period, and a dynamic trend slope is obtained by linear fitting;

[0044] If the error frequency year-on-year growth rate exceeds the fluctuation range, the corresponding period is marked as an abnormal period; otherwise, it is marked as a normal period;

[0045] An error frequency period sequence is obtained, a frequency spectrum diagram is obtained through signal decomposition, a main frequency frequency with the largest amplitude is screened, and a period length is calculated in combination with a sampling period;

[0046] If the amplitude corresponding to the main frequency frequency exceeds a preset frequency threshold, it is determined that there is a periodic rule; otherwise, it is determined that it is non-periodic;

[0047] A maintenance execution period is obtained, a maintenance post-frequency and a maintenance post-frequency are combined, a maintenance post-frequency drop rate is calculated, the number of errors with a response duration less than a preset response duration threshold is counted, a response duration compliance rate is calculated, and a response efficiency score is calculated through weighted calculation;

[0048] The frequency change unit, the periodic law unit and the response feature unit are constructed to store the calculated various indexes respectively, and a periodic error trend feature library is constructed.

[0049] Specifically, the specific steps of determining the periodic repeated error list include:

[0050] The frequency change unit data is acquired, and high-risk candidate errors exceeding m abnormal periods are screened out;

[0051] If the dynamic trend slope is positive and the average error frequency growth rate exceeds the historical baseline, it is determined that the trend deviates from the error; otherwise, it is determined that it is accidental fluctuation;

[0052] The number of continuous abnormal periods is counted, the trend deviation index is calculated by weighting, the deviation level is determined by combining the preset trend deviation index threshold, and a preliminary error list is formed;

[0053] In combination with the periodic law unit, the errors with periodic law in the preliminary error list are retained, the periodic length is used to calculate the regularity matching degree to determine the repeated mode, and the repeated error list is updated;

[0054] The response feature unit data is acquired, if the response efficiency score in the delayed maintenance corresponding period is lower than the preset score value, and the frequency decline rate after maintenance is negative, the maintenance associated candidate inducement is marked; otherwise, the non-maintenance associated type periodic error is marked;

[0055] The power plant operation and maintenance log is called, the error maintenance record in the repeated error list is queried, if the maintenance is missing or the resource is insufficient, the potential inducement is marked; otherwise, it is marked to be checked; and finally, the periodic repeated error list is determined.

[0056] Specifically, the specific steps of processing the multi-level early warning include:

[0057] Based on the periodic error trend feature library and the periodic repeated error list, a to-be-early-warned error set is constructed;

[0058] The deviation level quantitative value, the repeated mode intensity quantitative value and the potential inducement mark quantitative value are calculated respectively, and the comprehensive early warning level score is calculated by weighting ;

[0059] Three threshold values are set 、 、 , wherein , the early warning level mapping is performed in combination with the to-be-early-warned error set;

[0060] When , it is determined that the third-level early warning needs to be responded immediately;

[0061] When When the value of the error trend slope is greater than the preset error trend slope threshold value, it is determined that a secondary early warning is needed, and close attention is needed.

[0062] When the value of the error trend slope is greater than the preset error trend slope threshold value, it is determined that a secondary early warning is needed, and close attention is needed. When the value of the error trend slope is greater than the preset error trend slope threshold value, it is determined that a secondary early warning is needed, and close attention is needed.

[0063] When the value of the error trend slope is greater than the preset error trend slope threshold value, it is determined that a secondary early warning is needed, and close attention is needed.

[0064] Specifically, the specific steps of processing the multi-level early warning further include:

[0065] According to different early warning levels, a differentiated processing strategy is implemented.

[0066] A dynamic feedback mechanism is started, a back-checking mechanism is set, the number of recurrence in the execution of the differentiated processing strategy and the total number of monitoring are counted, and the error recurrence rate is calculated.

[0067] The number of affected devices before and after the execution of the differentiated processing strategy is counted, and the error impact range reduction ratio is calculated.

[0068] The effectiveness index is obtained by weighted calculation in combination with the error recurrence rate and the error impact range reduction ratio. ;

[0069] If the effectiveness index is greater than a preset index threshold value, it is determined that the early warning processing is effective; otherwise, it is determined that the early warning processing is ineffective, the common features of the ineffective early warning processing are counted, and the weight parameter is dynamically adjusted. The beneficial effects of the present application are as follows:

[0070] 1. The present application realizes deep source tracing and time sequence evolution analysis of power plant operation errors by constructing a cross-period correlation graph and a multi-dimensional weighted correlation relationship table; by fusing device logic, personnel behavior and historical data, the limitations of traditional methods of independent analysis by period are broken through, cross-period correlation patterns are effectively identified, and the explainability of error causes is improved.

[0071] 2. The present application realizes intelligent early warning of repetitive and trend errors by establishing a periodic error trend feature library and a multi-level early warning mechanism, and by combining dynamic trend slope, periodic law and response characteristics for comprehensive scoring; the processing effectiveness is evaluated through a closed-loop feedback mechanism, the weight parameter is dynamically optimized, the self-adaptive ability of the system is enhanced, and the grading control and active prevention and control of power plant operation risks are effectively supported.

[0072] BRIEF DESCRIPTION OF DRAWINGS

[0073] Figure 1 It is a cross-period traceability analysis system structure diagram for supervision rule error data.

[0074] Figure 2 It is a flowchart for generating a cross-period correlation graph in the present application. ​​

[0075] Figure 3 Flow chart for generating preliminary error evolution sequence in the present application;

[0076] Figure 4 Flow chart for constructing periodic error trend feature library in the present application. DETAILED DESCRIPTION

[0077] The technical solutions of the present application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solutions of the present application, and are not limitations of the technical solutions of the present application. In the case of no conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other.

[0078] The term "and / or", only describes the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. In addition, the character " / ", generally represents that the front and rear associated objects are in an "or" relationship.

[0079] Reference Figures 1 to 4 As shown in the figure, the embodiment introduces a cross-period backtracking analysis system for regulatory rule error data, which includes a data acquisition module, a traceability map construction module, a periodic error analysis module, and a multi-level early warning processing module.

[0080] The data acquisition module is used to collect power plant operation raw data, clean and semantically analyze the raw data in combination with a preset power plant regulatory rule library containing error type subdivision standards and periodic regulatory requirements, and based on the matching results of the error determination logic and the periodic attribution rules in the preset power plant regulatory rule library, label the data with four dimensions of operator, equipment, error type, and period to construct a compliance traceability dataset. The raw data includes but is not limited to operator ID, error operation log, timestamp, and equipment number.

[0081] The traceability map construction module is used to fuse the compliance traceability dataset, power plant equipment association logic, and personnel operation association rules to construct a multi-dimensional weighted association relationship table. Through graph association analysis, a cross-period association graph of the power plant scene is generated. Combined with sequence aggregation processing, preliminary error evolution sequences grouped by operators, equipment, and error types are generated. The power plant equipment association logic is obtained from the power plant operation procedure text, and the personnel operation association rules are obtained from the historical log.

[0082] Specifically, the specific steps of constructing the multi-dimensional weighted association relationship table include:

[0083] Extracting power plant equipment association logic from power plant operation procedure text, preprocessing the text, identifying equipment based on BERT-based NER model, extracting association conditions from preprocessed power plant operation procedure text using dependency syntax analysis, forming equipment A, equipment B, and equipment association triplets;

[0084] Filtering time period from historical log to Grouping complete logs within the time period according to operator ID and equipment number, retaining operation time, error type, and equipment state for each group, defining as the basis data of operator and equipment association;

[0085] Based on the personnel operation association rule, setting the time window to for the retained basis data of operator and equipment association, counting the time sequence co-occurrence frequency of error types within after the operator operates the equipment, calculating the ratio of time sequence co-occurrence frequency to total operation times for the same operator on the same equipment to obtain the association strength, and setting the normalized association strength as the association confidence;

[0086] According to the combination of equipment and equipment, the combination of operator and equipment, and the combination of operator and error type, three types of association are divided, and a structured association rule library is constructed, which includes rule source and core association elements such as equipment association triplets and association confidence;

[0087] Traverse each compliance trace record in the compliance trace dataset, match the structured association rule library according to the association type, including: for the combination of equipment and equipment association type, extract the operation time sequence and state sequence of equipment A number and equipment B number in the compliance trace dataset, align and analyze with the preset linkage mode in the structured association rule library, based on the time sequence synchronization score and the state response matching score, calculate the condition compliance degree by weighting; for the combination of operator and equipment association type, count the matching field number between each record in the compliance trace dataset and the structured association rule library, combine the total comparison field number, and calculate the association matching degree; for the combination of operator and error type association type, match the matching field number between each record in the compliance trace dataset and the structured association rule library, combine the total comparison field number, and calculate the behavior matching degree;

[0088] Integrate condition compliance degree, association matching degree, and behavior matching degree, calculate the comprehensive matching degree by weighting, and generate an association intermediate table with matching degree;

[0089] For each intermediate matching record in the association intermediate table with matching degree, calculate the product of association confidence and comprehensive matching degree to obtain association weight;

[0090] By integrating the core dimensions, association types, and association weights of each intermediate matching record, a structured multidimensional weighted association table is constructed to ensure that the information of each dimension corresponds one-to-one with the weight; among which, the core dimensions include operator, equipment, error type, and cycle.

[0091] Specifically, the steps for generating cross-period correlation maps include:

[0092] Based on a multidimensional weighted association table, three types of nodes are defined: operator, equipment, and error type. The power plant's basic database is then called to supplement node attributes, such as operator position, equipment operating years, and error type severity level, to complete node initialization.

[0093] For three types of associations—device-device combinations, operator-device combinations, and operator-error type combinations—corresponding directed edges are constructed. The initial attributes of the directed edges include association weights and corresponding period labels. The NetworkX library is used to store the data, forming a basic association graph containing node attributes, directed edge weights, and period information. The basic association graph is then divided into N periodic subgraphs according to the period labels, with each subgraph retaining only the nodes and directed edges within the corresponding period.

[0094] For each subgraph, calculate the weight distribution characteristics of directed edges of the same type, such as standard deviation. Then, use Z-score standardization to convert the edge weights into relative strengths within a period, enhancing the comparability of associations within the same period. Label each subgraph with a period identifier, such as a period number, forming a set of subgraphs with periodic attributes. ,in For the first Subgraphs of each period, Furthermore, the periods of each subgraph are arranged in chronological order.

[0095] For adjacent periodic subgraphs in the subgraph set and By traversing and comparing and The node set and directed edge set are used to filter out pairs of nodes of the same type that exist simultaneously, such as operator a and device B;

[0096] Set a weight threshold; if similar node pairs are in adjacent periodic subgraphs... or If the edge weight is greater than the weight threshold, it indicates that the same type of node pair has a valid association in adjacent periods and is judged as a cross-period association candidate pair; otherwise, it indicates that the same type of node pair does not exist in adjacent periods and is judged as a non-cross-period association candidate pair.

[0097] For cross-period correlation candidate pairs, by calculating the first... Subgraphs of each period The middle edge weight and the bridge weight of the preceding term with the preset first weight coefficient, the first edge weight subgraph of a period bridge weight between the former bridge weight and the latter bridge weight, and the sum of the former bridge weight and the latter bridge weight is a bridge strength;

[0098] The bridge strength is defined as a cross-period correlation strength, and a subgraph connected to adjacent periods is generated and bridge edges, and the bridge edges are saved to a subgraph set to form a cross-period connected intermediate graph;

[0099] By using a label propagation algorithm, adjacent nodes in the intermediate graph gradually share the same label through label transmission between the nodes, and then closely associated clusters are divided, and community core nodes in the clusters are marked;

[0100] Based on the edge weight and the cross-period correlation strength, a maximum threshold is preset by a statistical method, strong correlation edges with a weight higher than the preset maximum threshold are screened, and the associated path containing the community core node is preferentially retained, and the node pair, period sequence and strength change information are completely recorded; a minimum threshold is preset by a statistical method, and weak correlation edges with a weight lower than the preset minimum threshold are simultaneously screened, and one-time accidental correlation is eliminated, and redundancy of the intermediate graph is reduced;

[0101] By using a label propagation algorithm, adjacent nodes in the intermediate graph gradually share the same label through label transmission between the nodes, and then closely associated clusters are divided, and community core nodes in the clusters are marked;

[0102] Based on the edge weight and the cross-period correlation strength, strong correlation edges with a weight higher than the preset maximum threshold in the cluster are screened, and the associated path containing the community core node is preferentially retained, and the node pair, period sequence and strength change information are completely recorded; weak correlation edges with a weight lower than the preset minimum threshold are simultaneously screened, and one-time accidental correlation is eliminated, and redundancy of the intermediate graph is reduced;

[0103] A time dimension label is added to the intermediate graph, and a core correlation range attribute is supplemented for the community core node, and finally a cross-period correlation graph is generated; wherein the cross-period correlation graph includes but is not limited to node attributes, cross-period correlation strength and time sequence characteristics.

[0104] Specifically, the specific steps of generating the preliminary error evolution sequence include:

[0105] By using a graph database query statement such as Cypher, correlation edges containing operator, device and error type nodes are screened from the cross-period correlation graph, the period label and the cross-period correlation strength of each correlation edge are extracted, and the device state parameter, error details and operator operation time auxiliary features are associated, and are stored in a table form, and an original time sequence correlation dataset is formed;

[0106] The time sequence data in the original time sequence correlation data set is arranged in ascending order of period for each unique operator ID, equipment number, and operation error type triplet, to be continuous The time sequence data is divided into a plurality of overlapping sliding windows with a period as a window size and a single period as a step;

[0107] The numerical features in each sliding window, such as correlation strength, are normalized to eliminate dimensional effects, and the categorical features in each sliding window, such as error type, are converted into binary vectors using one-hot encoding; the numerical features and the categorical features in the same sliding window are concatenated into a fixed-length feature vector to form a window feature vector sequence corresponding to the operation error triplet, and the window feature vector sequences corresponding to all operation error triplets are combined to form a set of encoded time sequence feature sequences; in this embodiment, the fixed length is set to L;

[0108] Based on the window feature vector sequence of each operation error triplet in the set of time sequence feature sequences, a self-attention mechanism is introduced to calculate the attention weight of each sliding window in the window feature vector sequence, the windows are sorted in descending order according to the attention weight, the first n windows are kept as key sequence fragments, redundant fragments are filtered, and the attention weight and the period of the key sequence fragments are recorded simultaneously;

[0109] Grouped by operators, target fragments with the same operator ID are selected by traversing the key sequence fragments, and are concatenated in chronological order to generate error evolution sequences centered on operators;

[0110] Grouped by equipment, matching equipment fragments with the same equipment number are selected by traversing the key sequence fragments, and are concatenated in chronological order to generate error evolution sequences centered on equipment;

[0111] Grouped by error type, matching error fragments with the same error type are selected by traversing the key sequence fragments, and are concatenated in chronological order to generate error evolution sequences centered on error type;

[0112] Finally, preliminary error evolution sequences grouped by operators, equipment, and error type are obtained.

[0113] The period error analysis module is used to comprehensively utilize the cross-period correlation graph and the preliminary error evolution sequence, to take the error occurrence rate and the corresponding fluctuation range of the same type of equipment and the same post as a historical baseline, to analyze the frequency change and the period law through dynamic trend fitting, to analyze the response characteristics in combination with the equipment maintenance state, to form a periodical error trend feature library containing the frequency change, the period law, and the response characteristics, and to determine a periodical repeated error list marked with trend deviation, repeated patterns, and potential causes;

[0114] Specifically, the steps for constructing a periodic error trend feature library include:

[0115] By comprehensively utilizing cross-cycle correlation maps and preliminary error evolution sequences, basic groups are constructed with equipment type, operator ID, cycle label, and error type as core dimensions, and these are recorded as first-level groups. This ensures that each group corresponds to a unique combination of equipment type, operator, cycle label, and error type.

[0116] The frequency of edges associated with each error type node within each group is counted and defined as the error frequency. This ensures that only a unique error frequency record is generated for each group, avoiding duplicate counting. At the same time, the equipment maintenance status within the corresponding period, such as normal maintenance and delayed maintenance, is extracted from the equipment node attributes of the cross-period association graph.

[0117] Based on the first-level grouping, in the initial error evolution sequence, the sequence segments with consistent device type, same operator ID, matching cycle label, and corresponding error type are located, and the valid error occurrence timestamp and valid repair completion timestamp for each error type in the corresponding cycle are extracted from the time series field of the sequence segments;

[0118] The response time is calculated based on the valid error occurrence timestamp and the valid repair completion timestamp, including: if the error type is repaired within the same cycle, the response time is recorded as follows. One cycle; if the error type requires cross-cycle repair, then calculate the number of cycles required to cross the cycle. Response time is recorded as One cycle; if there is no corresponding timestamp record in the initial error evolution sequence, the error handling log of the associated power plant operation and maintenance management system will be supplemented and marked as to be supplemented;

[0119] The system calls upon the employee job configuration database to match the corresponding job information. Combining the equipment type and operator ID of the primary group, it uses equipment type, job position, periodic label, and error type as core dimensions to form a secondary group. Error frequency, equipment maintenance status, and response time are then associated with the secondary group to form structured periodic error basic feature data.

[0120] Using similar equipment and the same job as the core dimensions of the historical baseline, the error occurrence rate and fluctuation range of similar equipment and the same operation are extracted from the preset historical database as the historical baseline. Using equipment type, job position and error type as dimensions, the historical baseline is associated with secondary groups and integrated into a structured periodic error basic dataset. The periodic error basic dataset includes, but is not limited to, error frequency, equipment maintenance status, response time, historical fluctuation range and error frequency periodic sequence.

[0121] Specifically, the steps for constructing a periodic error trend feature library also include:

[0122] For each set of device type, post, error type in the periodic error basic data set, calculate the error frequency year-on-year growth rate of adjacent periods in ascending order of period to obtain the error frequency year-on-year growth rate sequence of each period; use a linear regression model to dynamically trend fit the error frequency year-on-year growth rate sequence of each period, and output the dynamic trend slope; if the dynamic trend slope is positive, it is determined that the frequency is on the rise; otherwise, it is determined that the frequency is on the decline;

[0123] In combination with the historical baseline fluctuation range, if the error frequency year-on-year growth rate exceeds the historical baseline fluctuation range, the corresponding period is marked as an abnormal period, and the number of abnormalities and the deviation amplitude are recorded; otherwise, it is marked as a normal period, and the number of consecutive normal periods is counted;

[0124] The error frequency period sequence is extracted from the periodic error basic data set, Fourier transform is used to decompose the error frequency period sequence, and a frequency spectrum graph in the frequency domain is obtained; the frequency component with the largest amplitude is selected from the frequency spectrum graph, which is defined as the main frequency, and the ratio of the sampling period to the main frequency is calculated to obtain the period length; in this embodiment, the sampling period is set to 1;

[0125] A frequency threshold is preset by expert experience, if the amplitude corresponding to the main frequency exceeds the preset frequency threshold, it is determined that the error frequency period sequence has a periodicity, and the period length is recorded; otherwise, it is determined that the error frequency period sequence does not have a periodicity, and is marked as non-periodic;

[0126] The device maintenance execution period is obtained from the device node attribute of the cross-period correlation graph, and the error frequency before maintenance is selected as the frequency before maintenance and the error frequency after maintenance is selected as the frequency after maintenance, with the device maintenance execution period as the boundary; the difference between the frequency before maintenance and the frequency after maintenance is calculated, and then the difference is divided by the frequency before maintenance to obtain the frequency after maintenance drop rate;

[0127] A response duration threshold is preset by industry standard, the number of errors with a response duration lower than the response duration threshold is counted, the ratio of the number of errors to the total number of errors is calculated to obtain the response duration compliance rate;

[0128] The maintenance after frequency drop rate and the response duration compliance rate are weighted to obtain the response efficiency score of each set of device type, post, and error type;

[0129] The feature library unit is divided according to the device type and the post, each feature library unit is classified according to the error type, including: the frequency change unit for storing the error frequency of each period, the error frequency year-on-year growth rate sequence of each period, the dynamic trend slope, and the abnormal period; the periodicity unit for storing the period length, the main frequency, the amplitude corresponding to the main frequency, and the periodicity judgment result; the response feature unit for storing the device maintenance state, the response duration, and the response efficiency score.

[0130] The frequency change unit, the period rule unit and the response characteristic unit form a periodic error trend characteristic library of full-dimension frequency, period and response.

[0131] Specifically, the specific steps of determining the periodic repeated error list include:

[0132] From the periodic error trend characteristic library, the frequency change unit data is extracted, and for each group of data, the errors that are continuously marked as abnormal periods for more than m periods are filtered as high-risk candidate errors;

[0133] For the high-risk candidate errors, the dynamic trend is verified in combination with the dynamic trend slope, and if the dynamic trend slope is positive and the average error frequency growth rate in the continuous abnormal period exceeds the historical baseline, it is determined as a trend deviation error; otherwise, it is determined as an accidental high-frequency fluctuation and is continuously monitored;

[0134] According to the number of continuous abnormal periods and the average error frequency growth rate, the trend deviation index is calculated by weighting , the trend deviation index threshold is set , , , and ; if , it is determined that the deviation level is serious; if , it is determined that the deviation level is heavy; if , it is determined that the deviation level is general; if , it is determined that the deviation level is attention level; a preliminary error list containing the trend deviation error and the deviation level is formed;

[0135] In combination with the period rule unit, the errors determined as non-periodic are filtered from the preliminary error list, and only the errors with period rules are retained;

[0136] For the retained errors with period rules, based on the period length, the theoretical occurrence point is constructed in the error frequency period sequence, the proportion of actual abnormal periods coinciding with the theoretical period is counted, and is recorded as the rule matching degree; if the rule matching degree exceeds the preset matching degree threshold, it is determined as a repeated mode error, and the matching degree value and the period length are marked; otherwise, it is determined as a non-repeated mode error, and is marked as weak periodic interference;

[0137] Based on the repeated mode error, the matching degree value and the period length, the preliminary error list is updated to obtain a repeated error list;

[0138] The response characteristic unit data is extracted from the periodic error trend characteristic library, including the equipment maintenance state corresponding to the error, the response efficiency score and the frequency decline rate after maintenance;

[0139] For the errors in the repeated error list, the equipment maintenance state in the repeated pattern period is extracted, and maintenance correlation analysis is performed, including: if the response efficiency score in the corresponding period of the delayed maintenance is lower than the preset score value, and the frequency decline rate after maintenance is negative, then mark it as a maintenance correlation candidate inducement; otherwise, mark it as a non-maintenance correlation type period error, and continue to monitor;

[0140] The power plant operation and maintenance log is called to query the maintenance execution record of the error in the repeated error list in the corresponding period of the delayed maintenance. If the record of missing maintenance process or insufficient maintenance resources is found, it is determined that the maintenance correlation candidate inducement is established, and is marked as a potential inducement; otherwise, it is determined that the maintenance correlation candidate inducement is not established, and is marked as to be checked.

[0141] Finally, the three types of information of trend deviation, repeated pattern and potential inducement are comprehensively determined to determine the marked complete periodic repeated error list.

[0142] The multi-level early warning processing module is used to construct a set of errors to be warned according to the periodic error trend feature library and the periodic repeated error list, and to trigger multi-level early warning and multi-level early warning processing by combining the comprehensive early warning level score of deviation level, repeated pattern and potential inducement.

[0143] Specifically, the specific steps of processing multi-level early warning include:

[0144] By analyzing the structured fields of each set of device type, post and error type combination records in the periodic error trend feature library, and combining the key tags of deviation level, cycle length, regularity matching degree and potential inducement extracted from the periodic repeated error list, a set of errors to be warned is constructed;

[0145] The weights are determined by the analytic hierarchy process , , The deviation level quantization value is calculated by the deviation level and value mapping rule, the repeated pattern intensity quantization value is obtained by calculating the ratio of regularity matching degree and cycle length, and the potential inducement flag quantization value is calculated by the inducement state and value mapping rule , , are multiplied by the corresponding weights , , respectively, and then summed to obtain the comprehensive early warning level score , wherein the deviation level and value mapping rule and the inducement state and value mapping rule are determined according to expert experience, and the comprehensive early warning level score expression is as follows:

[0146]

[0147] wherein, is a deviation level weight, is respectively valued as , , , ; is a repetition mode intensity weight; is a potential inducement sign weight, is respectively valued as , ;

[0148] a three-level threshold value , , is set, wherein , in combination with the to-be-alarmed error set, the comprehensive alarm level score is performed alarm level mapping; when , it is determined as a three-level alarm, which needs immediate response and emergency measures; when , it is determined as a two-level alarm, which needs close attention and planned response; when , it is determined as a one-level alarm, which exists potential risk and needs continuous monitoring; when , it is determined as no alarm, which is normal in the current state;

[0149] According to the difference of the alarm level, a differentiated processing strategy is generated, including: for a three-level alarm, immediately pushing to the emergency response system, generating a temporary inspection task and executing; for a two-level alarm, pushing to some responsible persons, generating a special analysis task and executing;

[0150] A dynamic feedback mechanism is started, for a three-level and two-level alarm, setting a periodical review mechanism, counting the recurrence number, the total monitoring number, the number of affected devices before executing the differentiated processing strategy and the number of affected devices after executing the differentiated processing strategy in the process of executing the differentiated processing strategy;

[0151] The recurrence number is defined as the number of times that the same type of error occurs again in a specific monitoring period; the error recurrence rate is obtained by calculating the ratio of the recurrence number to the total number of monitoring times; the affected device is defined as the device that directly appears to have a use failure or a function failure problem due to the error before and after the error occurs or the strategy is executed, the difference between the number of affected devices after the differentiated processing strategy is executed and the number of affected devices before the differentiated processing strategy is executed is calculated based on the number of affected devices before the differentiated processing strategy is executed, combined with the number of affected devices after the differentiated processing strategy is executed, and the difference is divided by the number of affected devices before the differentiated processing strategy is executed to obtain the error impact range reduction ratio; the error recurrence rate and the error impact range reduction ratio are weighted and calculated to obtain the effectiveness index ; the index threshold is set, and if is greater than the index threshold, it is determined that the early warning processing is effective; otherwise, it is determined that the early warning processing is ineffective, and the common features of the ineffective early warning processing, such as the post type, are counted, and the weight parameter is dynamically adjusted 、 、 , such as increasing the weight parameter to improve the early warning sensitivity and optimize the early warning accuracy.

[0152] In summary, in the embodiment, the data acquisition module is used to clean and semantically analyze the original data of the power plant operation, and the pre-set supervision rule library is used to label the four-dimensional tags of operators, devices, error types and periods to construct a compliance traceability data set; the traceability graph construction module is used to fuse the compliance traceability data set, the device association logic and the personnel operation association rule to construct a multi-dimensional weighted association relationship table, to generate a cross-period association graph by using graph analysis, and to generate a preliminary error evolution sequence with the operators, the devices and the error types as the core in combination with sequence aggregation; the periodical error analysis module is used to comprehensively utilize the graph and the preliminary error evolution sequence, to analyze the period-on-period growth rate and the periodical law of the error frequency based on the historical error occurrence rate and the fluctuation range of the same type of device and the same post as the historical baseline, to construct a periodical error trend feature library containing the frequency change, the periodical law and the response characteristics based on the device maintenance state and the response efficiency, and to determine a periodical repeated error list based on the trend deviation, the repeated mode and the potential cause; the multi-level early warning processing module is used to set multi-level early warning threshold values to trigger a first-level, a second-level and a third-level early warning based on the comprehensive early warning level score of the deviation level, the repeated mode intensity and the potential cause, to execute a differentiated processing strategy, and to introduce a dynamic feedback mechanism to verify the effectiveness by using a back-checking mechanism, to evaluate the processing effect based on the effectiveness index, and to dynamically adjust the weight parameter to realize the closed-loop optimization and continuous iteration of the early warning process. The present application realizes the cross-period traceability, the periodical mode recognition and the self-adaptive intelligent early warning of the power plant error data, and improves the active prevention and control capability of the system operation risk.

[0153] The above merely describes the preferred embodiments of the present application, and the protection scope of the present application is not limited to the above-described embodiments. Any technical solution falling within the concept of the present application shall fall within the protection scope of the present application. It should be noted that, for ordinary skilled persons in the art, some improvements and refinements without departing from the principles of the present application shall also be considered as falling within the protection scope of the present application.

Claims

1. A system for cross-period traceability analysis of regulatory rule error data, characterized in that, include: Data acquisition module, source map construction module, periodic error analysis module, multi-level early warning processing module; The data acquisition module is used to collect, clean, and analyze raw power plant operation data, label it with four-dimensional tags, and construct a compliance traceability dataset. The traceability graph construction module is used to integrate the compliance traceability dataset, equipment association logic and personnel operation association rules to construct a multi-dimensional weighted association table and generate a cross-cycle association graph and a preliminary error evolution sequence. The periodic error analysis module is used to combine the cross-period correlation map with the preliminary error evolution sequence, and to analyze frequency changes, periodic patterns and response characteristics through dynamic trend fitting analysis, thereby constructing a periodic error trend feature library and determining a list of periodic recurring errors. The multi-level early warning processing module is used to construct a set of errors to be warned, and combined with the comprehensive early warning level score, to trigger multi-level early warnings and process them.

2. The cross-period traceability analysis system for regulatory rule error data according to claim 1, characterized in that, The specific steps for constructing a multidimensional weighted association table include: Obtain the power plant equipment association logic, identify equipment types and extract association conditions to form a triplet of equipment association: equipment A, equipment B, and association conditions; Filter time period arrive The complete log is grouped by operator ID and device number, retaining the basic data associated with each operator and device. Based on the aforementioned basic data, the temporal co-occurrence frequency of error types within the statistical time window is calculated, and the correlation strength is calculated in conjunction with the total number of operations, and then normalized to the correlation confidence level. Based on the three association types—device-to-device, operator-to-device, and operator-to-error—a structured association rule library containing device association triples and association confidence is constructed.

3. The cross-period traceability analysis system for regulatory rule error data according to claim 2, characterized in that, The specific steps for constructing a multidimensional weighted association table also include: Iterate through each compliance traceability record in the aforementioned compliance traceability dataset; For the device-to-device association type, the operation timing and state sequence are extracted. Through alignment analysis, the condition compliance is calculated by combining the timing synchronization score and the state response matching score. For the operator-equipment and operator-error type association types, the number of matching fields between each compliance traceability record and the structured association rule base is counted. Combined with the total number of matching fields, the association matching degree and behavior matching degree are calculated. The overall matching score is calculated by weighting the condition compliance, the association matching score, and the behavior matching score, and an association intermediate table is generated. For each intermediate matching record in the intermediate association table, the association weight is calculated by combining the association confidence and the overall matching degree. By integrating the core dimensions, association types, and association weights of each intermediate matching record, a multidimensional weighted association table is constructed.

4. The cross-period traceability analysis system for regulatory rule error data according to claim 3, characterized in that, The specific steps for generating cross-period correlation maps include: Based on the multidimensional weighted association table, the operator, equipment, and error type nodes are initialized. Directed edges are constructed for the three types of association to form a basic association graph, which is then split into periodic subgraphs. Calculate the directed edge weight distribution characteristics of each periodic subgraph, generate a set of subgraphs, and filter out similar node pairs by traversing and comparing adjacent periodic subgraphs. If the edge weight of the same type of node pair in the adjacent periodic subgraph is less than the preset weight threshold, it is determined to be a non-cross-period association candidate pair; otherwise, it is determined to be a cross-period association candidate pair. The preceding bridging weight and the following bridging weight are calculated, and the cross-period association strength is obtained by summing them. Bridging edges are generated and an intermediate graph is formed. Clusters are divided by passing labels between nodes. Strongly related edges with weights exceeding a preset maximum threshold are filtered out and related paths are retained. Weakly related edges with weights below a preset minimum threshold are filtered out and accidental associations are eliminated. Add time dimension labels to the intermediate graph to generate a cross-cycle correlation graph.

5. The system for cross-period traceability analysis of regulatory rule error data according to claim 4, characterized in that, The specific steps for generating the initial error evolution sequence include: The association edges and auxiliary features are extracted from the cross-period association graph to generate the original temporal association dataset; For the time-series data in the operator ID, device, and error type triplet of operation error, divide the sliding window in ascending order of period; Based on the sliding window, normalized numerical features and one-hot encoded category features are concatenated into a fixed-length feature vector to generate a set of time-series feature sequences. A self-attention mechanism is introduced to calculate attention weights, and the top n windows are retained as key sequence segments by sorting them in descending order. Based on the aforementioned operation error triplet, the key sequence segments are traversed, and target segments with the same operator ID, matching device segments with the same device number, and matching error segments with the same error type are selected respectively. These segments are then assembled in chronological order to generate a preliminary error evolution sequence.

6. The system for cross-period traceability analysis of regulatory rule error data according to claim 5, characterized in that, The specific steps for constructing a periodic error trend feature library include: The system constructs primary groups based on equipment type, operator ID, cycle label, and error type, counts error frequency, and extracts equipment maintenance status. Based on the first-level grouping, locate the sequence segments in the preliminary error evolution sequence, extract the effective error occurrence timestamp and the effective repair completion timestamp, and calculate the response time; Based on the primary grouping, secondary groupings are constructed using equipment type, job position, cycle label, and error type; By associating the error frequency, the equipment maintenance status, and the response time, basic characteristic data of periodic errors are formed; Using the historical error incidence and fluctuation range of similar equipment and positions as a historical baseline, and associating it with the secondary group, a periodic error basic dataset is generated.

7. The system for cross-period traceability analysis of regulatory rule error data according to claim 6, characterized in that, The specific steps for constructing a periodic error trend feature library also include: For each group of equipment type, job position, and error type in the periodic error basic dataset, calculate the year-on-year growth rate of error frequency in adjacent periods in ascending order of period, and obtain the dynamic trend slope through linear fitting; If the year-on-year growth rate of the error frequency exceeds the fluctuation range, the corresponding period is marked as an abnormal period; otherwise, it is marked as a normal period. Obtain the error frequency periodic sequence, obtain the spectrum through signal decomposition, filter the dominant frequency with the largest amplitude, and calculate the period length by combining the sampling period; If the amplitude corresponding to the main frequency exceeds the preset frequency threshold, it is determined that there is a periodic pattern; otherwise, it is determined to be non-periodic. Obtain the maintenance execution cycle, combine the frequency before and after maintenance, calculate the frequency decrease rate after maintenance, count the number of errors with response time lower than the preset response time threshold, calculate the response time target rate, and obtain the response efficiency score through weighted calculation; Construct frequency variation units, periodicity units, and response characteristic units; The frequency change unit is used to store the error frequency of each period, the error frequency year-on-year growth rate sequence of each period, the dynamic trend slope, and the abnormal period. The periodicity unit is used to store the period length, the main frequency, the amplitude corresponding to the main frequency, and the periodicity judgment result. The response feature unit is used to store the device maintenance status, response time, and response efficiency score. Construct a database of periodic error trend features.

8. The system for cross-period traceability analysis of regulatory rule error data according to claim 7, characterized in that, The specific steps for identifying a list of recurring periodic errors include: Acquire frequency change unit data and filter high-risk candidate errors that have been continuously abnormal for more than m periods; If the slope of the dynamic trend is positive and the year-on-year growth rate of the average error frequency exceeds the historical baseline, then the trend is determined to be a deviation from the error; otherwise, it is determined to be an accidental fluctuation. The number of consecutive abnormal periods is counted, the trend deviation index is calculated by weighting, and the deviation level is determined by combining the preset trend deviation index threshold to form a preliminary error list. Based on the aforementioned periodicity unit, errors exhibiting periodicity in the initial error list are retained. Based on the period length, the pattern matching degree is calculated to determine the repetition pattern, and the list is updated to include repetition errors. If the response efficiency score is lower than the preset score value within the corresponding period of delayed maintenance, and the frequency decrease rate after maintenance is negative, then a maintenance-related candidate cause is marked; otherwise, a non-maintenance-related period error is marked. The power plant operation and maintenance logs are retrieved to query the erroneous maintenance records in the recurring error list. If maintenance is missing or resources are insufficient, the potential cause is marked; otherwise, it is marked as pending verification. Finally, the periodic recurring error list is determined.

9. The system for cross-period traceability analysis of regulatory rule error data according to claim 8, characterized in that, The specific steps for handling multi-level early warnings include: Based on the periodic error trend feature library and the periodic recurring error list, a set of errors to be warned is constructed; The deviation level quantification value, repetition pattern intensity quantification value, and potential trigger indicator quantification value are calculated separately, and a comprehensive early warning level score is calculated by weighting. ; Set three threshold levels , , ,in The warning level is mapped by combining the set of errors to be warned; when When a Level 3 warning is issued, an immediate response is required. when At that time, a level-two warning was issued, requiring close monitoring; when At that time, a Level 1 warning was issued, indicating a potential risk that requires continuous monitoring. when At that time, it was determined that there was no warning.

10. The system for cross-period traceability analysis of regulatory rule error data according to claim 9, characterized in that, The specific steps for handling multi-level early warnings also include: Implement differentiated handling strategies based on different warning levels; A dynamic feedback mechanism is activated, a re-inspection mechanism is set up, and the number of recurrences and total monitoring times during the execution of the differentiated processing strategy are counted to calculate the error recurrence rate. The number of affected devices before and after implementing the differentiated processing strategy is counted, and the reduction ratio of the error impact range is calculated. An effectiveness index is obtained by combining the error recurrence rate and the reduction ratio of the error impact scope through weighted calculation. ; like If the value exceeds the preset threshold, the warning is deemed effective; otherwise, the warning is deemed ineffective. Common characteristics of ineffective warnings are statistically analyzed, and the weight parameters are dynamically adjusted.

Images