Converter station personnel behavior control method based on risk prediction

Abstract

The invention discloses a converter station personnel behavior management and control method based on risk prediction, and relates to the related technical field of data management, and the method comprises the steps: deploying gating security and protection equipment and visual sensing equipment in a converter station in a distributed manner; determining dynamic data of the converter station; interacting risk event records of the converter station, mining risk association rules, and combining with three-dimensional modeling of the converter station to carry out sample-driven training based on a time sequence, and constructing a risk prediction model; and returning the dynamic data of the converter station, carrying out personnel behavior characteristic analysis in combination with the risk prediction model, determining a risk control strategy, and responding to risk management equipment deployed at the front end for alarm management. The technical problem that in the prior art, the safety management efficiency and the response speed of the converter station are poor due to the lack of timely response ability to deal with sudden risk events, inaccurate personnel behavior monitoring and insufficient risk prediction ability is solved, and the technical effect of improving the safety management efficiency and the response speed of the converter station is achieved.

Description

Technical Field

[0001] The present application relates to the technical field related to data management, and in particular to a method for controlling the behavior of converter station personnel based on risk prediction. Background Art

[0002] The converter station is a key facility in the power transmission system. Its main function is to convert high-voltage direct current into alternating current, or vice versa. Its operating environment is complex and dangerous, and has extremely high requirements for personnel safety, equipment protection, and operational compliance. However, the traditional converter station personnel behavior control method relies on manual management and on-site monitoring, which is not only inefficient, but also lacks the ability to respond to sudden risk events in a timely manner. It is difficult to predict and judge potential risks based on real-time data and provide early warnings, which increases the probability of accidents and affects the accuracy and real-time monitoring of converter station personnel behavior and the safety control level of converter stations.

[0003] At the current stage, relevant technologies lack the ability to respond promptly to sudden risk events, inaccurate personnel behavior monitoring, and insufficient risk prediction capabilities, which lead to technical problems such as poor safety management efficiency and response speed of converter stations. Summary of the invention

[0004] This application provides a method for controlling the behavior of converter station personnel based on risk prediction, thereby solving the technical problems existing in the prior art, such as the lack of timely response capability to sudden risk events, inaccurate personnel behavior monitoring, and insufficient risk prediction capability, which lead to poor safety management efficiency and response speed of the converter station, thereby achieving the technical effect of improving the safety management efficiency and response speed of the converter station.

[0005] The present application provides a method for controlling the behavior of converter station personnel based on risk prediction, comprising: distributing gate control security equipment and visual sensing equipment inside the converter station, wherein the gate control security equipment is used for personnel authority management; determining the dynamic data of the converter station based on the supervision of the gate control security equipment and the visual sensing equipment and Beidou positioning based on a terminal worn by personnel, wherein the dynamic data of the converter station is marked with a collection timestamp; interacting with the risk event records of the converter station, mining risk association rules, and combining with the three-dimensional modeling of the converter station to perform sample-driven training based on time series to construct a risk prediction model, wherein the model includes twin modeling of event locations and simplified modeling of non-event locations; transmitting back the dynamic data of the converter station, analyzing the characteristics of personnel behavior in combination with the risk prediction model, determining the risk control strategy, and responding to the risk management equipment deployed at the front end for alarm management, wherein the risk prediction model is communicatively connected to the risk management equipment.

[0006] In a possible implementation, the method for controlling the behavior of converter station personnel based on risk prediction also performs the following processing: traverse the risk event records, integrate and determine M event sequences, wherein event position-event type-feature item is used as the integration standard, and the feature item is an event-related feature; cluster the M event sequences based on event position-event type to determine N groups of event sequences, wherein N is a positive integer less than or equal to M; based on the N groups of event sequences, mine N event association trees, wherein feature item screening is performed through mutual information calculation and symmetric uncertainty calculation; based on the N event association trees, determine the risk association rules.

[0007] In a possible implementation, the risk prediction-based converter station personnel behavior control method further performs the following processing: identifying a set of event sequences, wherein the set of event sequences includes multiple sets of feature items corresponding to each event; setting a minimum support, performing support statistics on the feature items of the set of event sequences, and screening strongly associated feature items that meet the minimum support, wherein support statistics are performed based on the frequency of feature items; and integrating the associated feature items through multi-layer refinement recursion of features to construct a first event association tree.

[0008] In a possible implementation, the risk prediction-based converter station personnel behavior control method further performs the following processing: traverse the first event association tree and calculate the mutual information coefficient of the connection nodes, wherein the calculation formula is: , F(X) and F(Y) represent the information entropy of the connection nodes X and Y respectively, and F(X,Y) represents the joint information entropy of X and Y; the symmetric uncertainty calculation is performed on the connection nodes, where the calculation formula is ; Setting a mutual information threshold and a symmetric uncertainty threshold, and pruning the first event association tree.

[0009] In a possible implementation, the risk prediction-based converter station personnel behavior control method further performs the following processing: identifying the event locations of the N event association trees, performing twin modeling on the event locations of the converter station, and determining a digital twin; performing simplified modeling on the non-event locations of the converter station and determining a simplified model; and spatially splicing the digital twin and the simplified model as a converter station model.

[0010] In a possible implementation, the method for controlling the behavior of converter station personnel based on risk prediction also performs the following processing: monitoring the dynamic data of the converter station and analyzing user behavior characteristics, which include operation behavior characteristics and trajectory behavior characteristics; performing threshold management based on the work ticket data scanned by the gate control and the trajectory behavior characteristics, and determining a first risk control strategy; performing twin deduction of risk events for the operation behavior characteristics, and determining a second risk control strategy, wherein the second risk control strategy includes an alarm strategy and an operation guidance strategy; determining the risk control strategy based on the first risk control strategy and the second risk control strategy.

[0011] In a possible implementation, the method for controlling the behavior of converter station personnel based on risk prediction further performs the following processing: constructing a risk rule engine, wherein the risk rule engine is used to match the response device-response method, and the response device is at least one risk management device; identifying the risk control strategy, combining the risk rule engine, and outputting the risk control execution mode; and performing converter station personnel behavior risk management according to the risk control execution mode.

[0012] In a possible implementation, the method for controlling the behavior of personnel in a converter station based on risk prediction also performs the following processing: introducing a temporary database, which is used to store risk management data; setting an incremental learning cycle, calling the periodic risk management data based on the temporary database, and deleting it within the database; updating the risk association rules based on the periodic risk management data, and performing incremental learning on the risk prediction model.

[0013] The proposed method of risk prediction-based personnel behavior control method for converter stations in this application will deploy access control security equipment and visual sensing equipment in a distributed manner inside the converter station; determine the dynamic data of the converter station; record the risk events of the interactive converter station, mine the risk association rules, and combine the three-dimensional modeling of the converter station to conduct sample-driven training based on time series to build a risk prediction model; transmit the dynamic data of the converter station back, analyze the characteristics of personnel behavior in combination with the risk prediction model, determine the risk control strategy, and respond to the risk management equipment deployed at the front end for alarm management. This solves the technical problems existing in the prior art, such as the lack of timely response capabilities to sudden risk events, inaccurate personnel behavior monitoring, and insufficient risk prediction capabilities, which lead to poor safety management efficiency and response speed of the converter station, and achieves the technical effect of improving the safety management efficiency and response speed of the converter station. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0015] Figure 1 A flow chart of a method for controlling the behavior of converter station personnel based on risk prediction provided in an embodiment of the present application; Figure 2 A schematic diagram of a process for mining risk association rules in a converter station personnel behavior control method based on risk prediction provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0017] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0018] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of the objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, product, or server comprising a series of steps is not necessarily limited to those steps clearly listed, but may include other steps that are not clearly listed or inherent to these processes, methods, products, or devices, and unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0019] The present application embodiment provides a method for controlling the behavior of converter station personnel based on risk prediction, such as Figure 1 As shown, the method includes: Step S100: deploying gate control security equipment and visual sensor equipment in a distributed manner inside a converter station, wherein the gate control security equipment is used for personnel authority management.

[0020] Preferably, the converter station is a key component of the high-voltage direct current (HVDC) system, which is mainly used to realize the conversion of electric power from alternating current (AC) to direct current (DC) or from direct current to alternating current. It is usually located at the starting and ending ends of long-distance transmission lines. Its main function is to convert electric energy from one form (AC or DC) to another form in order to achieve efficient power transmission. The access control security equipment and visual sensor equipment are distributedly deployed inside the converter station. Specifically, the access control security equipment refers to the hardware equipment used to control and manage the authority of personnel entering the converter station. It is mainly used for identity authentication, authority control and monitoring management of personnel entering the converter station, and can identify the identity of the entrants, such as through cards, fingerprints, facial recognition and other technical means to verify the personnel. Whether there is permission to enter specific areas, such as the control room, equipment room, etc., and only certified personnel can enter the relevant areas; not only identity verification of personnel, but also the ability to restrict or authorize them to enter certain specific areas of the converter station based on their identities, roles and authority settings. For example, certain high-risk areas may only allow senior operators or authorized personnel to enter, while ordinary personnel cannot enter; door control security equipment controls the opening and closing of doors through electromagnetic locks, mechanical locks, etc., to ensure that only personnel who meet the authority requirements can enter and exit, and is linked to the monitoring system to record each entry and exit event in real time, and is connected to other security equipment or central control systems through the network to transmit information such as personnel entry and exit, equipment status, etc. in real time for centralized monitoring and analysis. Visual sensing equipment conducts real-time monitoring through surveillance cameras, image recognition systems, etc., to detect and identify people's behaviors and actions. For example, when a staff member enters a restricted area through the access control system, the access control device will record the person's identity information, entry time and other information to cooperate with the authority management of the access control security equipment, further enhance the traceability and security management of personnel behavior, and thus realize access control, authority management and behavior monitoring of personnel in the internal area of ​​the converter station.

[0021] Step S200, determining the dynamic data of the converter station based on the supervision of the door control security equipment and the visual sensor equipment and the Beidou positioning based on the terminal worn by the personnel, wherein the dynamic data of the converter station is marked with a collection timestamp.

[0022] Preferably, the terminal device worn by the personnel (such as a Beidou positioning bracelet, a mobile phone or other device with Beidou positioning function) can accurately obtain the real-time position of the wearer through the satellite positioning system. Through this positioning, the position of the personnel can be tracked in real time in a wider range (especially in large-scale converter stations or complex scenarios), and then combined with the supervision information of the access control security equipment and the visual sensing equipment (i.e., the information of personnel entering the area and behavior monitoring), a more comprehensive understanding of the specific activity trajectory of each staff member in the converter station can be obtained. The data collected in real time by the access control security equipment, the visual sensing equipment, and the wearable terminal device (Beidou positioning) (including personnel location, behavior activities, entry and exit records, area access status, etc.) is used as the dynamic data of the converter station, which can dynamically and real-time display the activity status and behavior trajectory of each staff member, as well as the real-time situation of different areas in the converter station, and each piece of data is marked with a collection timestamp to ensure that all monitoring data can be associated with specific time and events, for example, recording the time when a staff member enters a certain area, the time when a specific behavior is performed, etc., to ensure the timeliness and accuracy of the data, thereby ensuring the accuracy of risk prediction and personnel behavior analysis, and can quickly trace and analyze when an abnormality occurs.

[0023] Step S300, the risk event records of the interactive converter station are mined to mine risk association rules, and combined with the three-dimensional modeling of the converter station, sample-driven training based on time series is performed to build a risk prediction model, which includes twin modeling of event locations and simplified modeling of non-event locations.

[0024] Preferably, the risk event records of the converter station are obtained and risk association rules are mined, wherein the risk event records refer to detailed records of various risk events occurring in the converter station, such as equipment failures, electrical accidents, personnel safety incidents, environmental anomalies, etc., which usually include information such as the time and location of the event, related equipment, personnel participation, and severity of the event. Then, risk association rules are mined, that is, association rule learning (such as Apriori algorithm, etc.) is used to identify the intrinsic connection between different risk events. For example, a certain equipment failure may be associated with abnormal personnel behavior in a specific area, changes in environmental factors, and other factors, and then identify which combinations of factors are likely to lead to the occurrence of certain specific risk events, thereby identifying potential dangers in advance. In other words, by analyzing the historical data of risk events, the laws between them are revealed to form risk patterns.

[0025] Preferably, the three-dimensional modeling of the converter station refers to creating a three-dimensional spatial model of the converter station through digital modeling technology, which can accurately display the various areas, equipment layout, personnel passage paths, etc. inside the converter station. Specifically, combined with the three-dimensional modeling of the converter station, potential risk scenarios are simulated and visualized. For example, the three-dimensional modeling is used to analyze the impact range, escape routes, degree of equipment damage, etc. of various areas of the converter station when a specific equipment fails or personnel misoperates. Then, sample-driven training based on time series is performed, including time series analysis of the historical dynamic data of the converter station (such as equipment operation data, personnel behavior data, etc.), that is, analyzing the trends and laws of these data over time to predict the risks of the converter station, including using these historical data (sample data) for sample-driven training, that is, building a prediction model based on machine learning (such as decision trees), using sample data for supervised training, and building a risk prediction model so that the model can predict risks based on known historical events and time series data.

[0026] Preferably, constructing a risk prediction model includes twin modeling of event locations and simplified modeling of non-event locations. Twin modeling is a virtual modeling technology that reflects the state changes of the physical system in real time by establishing a digital model that is the same as reality (i.e., digital twin). That is, in the risk prediction process, a virtual model corresponding to a certain area of ​​the converter station (the area where the risk event occurred) is established, and the dynamic state of the area is reflected through real-time monitoring and data analysis. For example, when a fault occurs in a certain area of ​​the converter station, the twin model will simulate the state of the equipment in the area, the changes in environmental factors, the impact of personnel activities, etc., to help predict similar risk events that may occur in the future; for areas where no risk events have occurred, the complexity of modeling is reduced by simplifying the modeling, and accurate digital twin modeling of the area is not required. Instead, the potential risk of the area is estimated through a relatively simplified mathematical model or probability model, which can reduce the burden of calculation and analysis and improve the efficiency of the overall system. For example, in some low-risk areas of the converter station, only a simple model is needed to describe the changes in personnel behavior or equipment status, without the need to simulate every detail in detail; thereby more accurately and dynamically predicting and preventing potential risks in the converter station and improving the safety management level of the converter station.

[0027] Further, such as Figure 2As shown, step S300 also includes step S310, traversing the risk event records, integrating and determining M event sequences, wherein event position-event type-feature item is used as the integration standard, and the feature item is an event-related feature; step S320, clustering the M event sequences by event position-event type, and determining N groups of event sequences, wherein N is a positive integer less than or equal to M; step S330, mining N event association trees based on the N groups of event sequences, wherein feature item screening is performed by mutual information calculation and symmetric uncertainty calculation; step S340, determining the risk association rules based on the N event association trees.

[0028] Preferably, the risk event records that have occurred in the converter station are traversed, and based on the location, type and feature items of the event, similar event records are aggregated into M event sequences (arranged in chronological order to describe the entire process of the event), where the event location refers to the specific area or equipment where the event occurs, the event type refers to the category of the event, such as equipment failure, power anomaly, personnel safety, etc., and the feature item refers to other information related to the event, such as the environmental conditions of the event, the behavior of personnel, the status of the equipment, etc., and M is a positive integer, representing the total number of risk events. The M event sequences are grouped according to the similarity of the event location and event type to generate N groups of event sequences, and the events in each group of event sequences have some common characteristics, such as the event types at the same location may have similar occurrence patterns. Through clustering, events with similar backgrounds, properties or development processes are classified into one category, so that it is easier to analyze the relationship between them to find the potential pattern of the event. For example, the power equipment failure in a certain area and the power equipment failure in another area may have similar causes and effects, where N is a positive integer less than or equal to M, representing the number of event sequence groups generated after clustering.

[0029] Preferably, N event association trees are mined according to N groups of event sequences, that is, an event association tree is assigned to each event sequence using a tree structure model to reveal the relationship between the features of this type of event. Specifically, feature items are screened by mutual information calculation and symmetric uncertainty calculation, that is, the relevance and importance of feature items are evaluated based on mutual information and symmetric uncertainty, and the feature items in each event sequence are screened to improve the accuracy and efficiency of association rule mining. Mutual information is used to measure the degree of information sharing between two variables. The higher the mutual information of two variables, the closer the relationship between them. In event data, it is used to measure the relationship between event location, event type and feature items. The mutual relationship between them is analyzed, and the characteristic items that have a strong influence on the event association relationship are screened out through calculation; finally, the risk association rules are determined according to N event association trees, that is, the relationship between the characteristic items revealed in the event association tree is analyzed as the risk association rule, which describes the correlation between certain specific event types or event locations, thereby helping to predict and identify potential risks. For example, through the event association tree, it can be found that equipment failures in a specific area are often accompanied by certain specific environmental conditions (such as high temperature, humidity changes, etc.) or certain specific personnel behaviors (such as irregular equipment operation, etc.); thereby realizing pattern recognition and risk warning of events in the converter station, and then improving the safety management efficiency and accuracy of the converter station.

[0030] Furthermore, step S330 also includes step S331, identifying a group of event sequences, wherein the group of event sequences includes multiple groups of feature items corresponding to each event; step S332, setting a minimum support, performing support statistics on the feature items of the group of event sequences, and screening strongly associated feature items that meet the minimum support, wherein support statistics are performed based on the frequency of feature items; step S333, integrating the associated feature items through multi-layer refinement recursion of features, and constructing a first event association tree.

[0031] Preferably, a group is arbitrarily selected from N groups of event sequences as a group of event sequences, wherein a group of event sequences contains multiple groups of feature items corresponding to each event, which are key data for describing the event. For example, for an equipment failure event, the feature items may include equipment type, failure occurrence time, failure type, environmental conditions (such as temperature, humidity), equipment operating status, etc.; a minimum support is set to filter out frequently occurring feature items, wherein support is an indicator in data mining that measures the frequency of occurrence of a feature item in a data set. The frequency of occurrence of feature items is counted to determine which feature items have a higher correlation in the event sequence. Only feature items whose occurrence frequency is greater than or equal to the minimum support threshold are screened as strongly correlated feature items, which play an important role in the event sequence and can effectively help analyze and predict the occurrence of events.

[0032] Preferably, the associated feature items are gradually refined and integrated in a recursive manner, that is, the relationship between the feature items is analyzed in depth layer by layer, and a higher-level feature combination is gradually constructed. In each layer, the recursive algorithm will combine the feature items of the current layer, construct new features, and continue the next recursive analysis until the preset analysis depth is reached, and then gradually discover deeper feature association relationships, and integrate these relationships into a more comprehensive feature set. Finally, through recursion and integration, the strongly associated feature items are organized into tree nodes to form a hierarchical structure, namely the first event association tree, wherein the root node of the event association tree represents an overall event type or risk category, and the branches and leaf nodes of the tree represent specific feature items or feature item combinations. Through this tree structure, the association and hierarchical structure between the event feature items can be clearly seen. For example, in the risk prediction of the converter station, the root node may be a certain type of fault event, and the branch nodes may be feature items such as environmental factors, equipment status, and operator behavior related to the fault. The event association tree can clearly reveal the potential correlation between different events and feature items in the converter station, thus providing a scientific basis for risk prediction, behavior analysis and the formulation of management strategies.

[0033] Further, step S333 also includes step A, traversing the first event association tree, and calculating the mutual information coefficient of the connection nodes, wherein the calculation formula is: , F(X) and F(Y) represent the information entropy of the connection nodes X and Y respectively, and F(X,Y) represents the joint information entropy of X and Y; Step B, symmetric uncertainty calculation is performed on the connection nodes, where the calculation formula is ; Step C, setting the mutual information threshold and the symmetric uncertainty threshold, and pruning the first event association tree.

[0034] Preferably, traverse the first event association tree, using Calculate the mutual information coefficient of the connecting nodes, where each connecting node represents a feature item of an event. F(X) and F(Y) represent the information entropy of the connecting nodes X and Y, respectively, which represents the uncertainty of a random variable. The larger the entropy value, the greater the randomness of the variable and the greater the amount of information. F(X,Y) represents the joint information entropy of X and Y, that is, the joint uncertainty of X and Y appearing at the same time. The mutual information coefficient is used to measure the degree of information sharing between two random variables (two feature nodes in the event association tree). In other words, the mutual information coefficient characterizes the correlation strength between the two feature items X and Y. If the two feature items are highly correlated, their mutual information coefficient will be high; if they are almost unrelated, the mutual information coefficient will be close to 0.

[0035] Preferably, using the formula Calculate the symmetric uncertainty of the connected nodes, where the symmetric uncertainty is a normalization of the mutual information coefficient, which measures the correlation between the two feature items X and Y and takes into account the influence of information entropy. Unlike the mutual information coefficient, the symmetric uncertainty standardizes the mutual information coefficient to make the results more comparable and not affected by the variable information entropy. is the mutual information coefficient between nodes X and Y, It is the sum of the information entropy of nodes X and Y. The higher the symmetric uncertainty value, the stronger the correlation between X and Y. On the contrary, if the symmetric uncertainty value is low, it means that the relationship between them is weaker.

[0036] Preferably, a mutual information threshold and a symmetric uncertainty threshold are set according to historical data, and the first event association tree is pruned according to the two thresholds, that is, if the mutual information value of a pair of nodes is less than the set threshold, or the symmetric uncertainty value is lower than the threshold, the pair of nodes and their connections are removed from the tree, and the node pairs with mutual information or symmetric uncertainty values ​​higher than the set threshold are retained. Pruning refers to deleting those connections with weaker associations with other nodes in the event association tree. Through pruning, the tree structure can be simplified and redundant or unimportant feature associations can be removed, thereby reducing the amount of calculation, improving the efficiency of the model, and removing noise and unnecessary complexity.

[0037] Furthermore, step S300 also includes step S350, identifying the event positions of the N event association trees, performing twin modeling on the event positions of the converter station, and determining a digital twin; step S360, performing simplified modeling on the non-event positions of the converter station and determining a simplified model; and step S370, spatially splicing the digital twin and the simplified model as a converter station model.

[0038] Preferably, the specific location where each event occurs is extracted from N event association trees, which may be associated with the equipment, personnel activity area, and environmental conditions of the converter station. The event location usually indicates which areas, equipment, or operating behaviors in the converter station are most likely to cause safety risks, and is used for dynamic analysis of related data. For example, mistaken entry into a high-voltage area, retention in a dangerous area, or approaching live equipment requires analysis and decision-making under its mechanism, and then twin modeling is performed for the event location of the converter station. Twin modeling refers to creating a virtual replica of a physical entity through digital means to determine a digital twin (an accurate virtual model created through physical world data and sensor data), which can reflect the state and changes of the physical entity in real time. Specifically, detailed twin modeling is performed for the event location, that is, those areas where risks have occurred or potential risks exist. For example, there may be an equipment area or operating area in the converter station that frequently fails due to historical events. These areas need to be modeled through real-time data such as sensors and monitoring equipment to form digital twins.

[0039] Preferably, non-event locations refer to those areas where no obvious risk events have occurred or where the risk is relatively low. The non-event locations of the converter station are simplified and modeled, that is, these areas are modeled through relatively simple mathematical models or logical models to determine the simplified model. For example, the model of the non-event area does not need to consider all equipment states and environmental changes, but only needs to simply simulate factors such as the equipment operating conditions and personnel activity status. By simplifying the model, the amount of calculation can be reduced and the system efficiency can be improved, providing sufficient accuracy, especially in low-risk areas, without overly refined modeling; finally, the digital twin (detailed modeling of the event location) is combined with the simplified model (modeling of the non-event location), that is, the two models of different precision are spatially spliced, and all areas in the converter station (including high-risk areas and low-risk areas) can be integrated into a unified model, which can not only retain the high-precision event location twin, but also maintain the integrity and computational efficiency of the system through the simplified model, thereby forming a converter station model, which can accurately simulate the operation of the entire converter station in a virtual environment to support better risk prediction and safety management.

[0040] Step S400, transmitting back the dynamic data of the converter station, analyzing the personnel behavior characteristics in combination with the risk prediction model, determining the risk control strategy, and responding to the risk management device deployed at the front end for alarm management, wherein the risk prediction model is communicatively connected with the risk management device.

[0041] Preferably, the dynamic data of the converter station is transmitted back to the cloud processing platform for further processing and analysis to ensure that a timely response can be made when potential risks are discovered. Specifically, the risk prediction model is combined to analyze the behavioral characteristics of personnel, that is, the risk prediction model is used to conduct in-depth analysis of the returned dynamic data to identify the behavioral patterns and potential safety risks of the staff. For example, the risk prediction model is used to identify whether the behavior of the staff in certain specific environments is abnormal, whether they have entered dangerous areas, whether there are irregular operations, etc., and then the risk control strategy is determined based on the analysis results (including personnel behavior characteristics, equipment status, etc.), which refers to response measures based on the prediction results. It is used to manage and control potential risks, which may include reminding or warning staff to take necessary safety measures, restricting or preventing specific personnel from entering certain high-risk areas, and initiating certain emergency mechanisms (such as cutting off equipment power, stopping operations, etc.). Finally, the determined risk control strategy will be responded to the front-end risk management equipment (such as warning lights, sirens, control systems, etc.) for alarm management. The front-end equipment will activate the relevant alarm mechanism according to the instructions received, such as issuing sound or light signals, or directly taking certain measures (such as restricting the passage of personnel in the area, automatically cutting off equipment, etc.), so as to quickly discover and handle risk events in the converter station, thereby effectively improving the safety and management efficiency of the converter station.

[0042] Furthermore, step S400 also includes step S410, monitoring the dynamic data of the converter station and analyzing user behavior characteristics, which include operation behavior characteristics and trajectory behavior characteristics; step S420, performing threshold management based on the work ticket data scanned by the gate control and the trajectory behavior characteristics, and determining a first risk control strategy; step S430, performing twin deduction of risk events for the operation behavior characteristics, and determining a second risk control strategy, wherein the second risk control strategy includes an alarm strategy and an operation guidance strategy; step S440, determining the risk control strategy based on the first risk control strategy and the second risk control strategy.

[0043] Preferably, the operation data, personnel behavior data, equipment status and other information about the converter station collected in real time by various sensors, monitoring equipment and positioning equipment, including the location of personnel, the operating status of equipment, operation records, environmental data, etc., and the characteristics describing the behavior of personnel in the converter station, i.e., user behavior characteristics, parsed from the dynamic data, include operation behavior characteristics and trajectory behavior characteristics, wherein the operation behavior characteristics describe the specific operation behavior of personnel in the converter station, such as equipment start-up, stop, adjustment, maintenance, etc., and the trajectory behavior characteristics describe the movement trajectory of personnel in the converter station, involving information such as the activity path and residence time of personnel in different areas; the converter station usually adopts an access control system to restrict personnel from entering specific areas, and the work ticket data refers to the operation behavior characteristics related to the operation. The approval records related to the task can be verified by scanning the work ticket through the access control system to verify whether the staff has obtained authorization to enter the specific area. Combined with the trajectory behavior characteristics and work ticket data, it can further analyze whether the personnel have entered the specific area as required and whether they have performed tasks consistent with the work ticket requirements in the area. For example, if a staff member enters an area that requires a specific work ticket authorization but performs operations inconsistent with the work ticket, it may cause risks. Threshold management is used to judge whether the current personnel behavior is normal based on the preset threshold standards, thereby forming the first risk control strategy. If a person enters a sensitive area without a valid work ticket, or the behavior after entering the area is inconsistent with the work ticket task, an early warning will be issued or the person will be prohibited from continuing to operate.

[0044] Preferably, based on historical data, operating behaviors and equipment status, the equipment of the converter station is virtually simulated through digital twin technology, and potential risk events are deduced in combination with the operating behavior characteristics, that is, the reactions under different operating behaviors are simulated to identify potential risks in advance. For example, when the staff performs certain operations on the equipment, the consequences of the operation are simulated to predict possible failures or safety incidents. Based on the results of the twin deduction, the second risk control strategy is determined, including an alarm strategy and an operation guidance strategy. Among them, the alarm strategy refers to when the twin deduction analyzes that a certain operating behavior may cause a risk, an alarm signal is issued in real time to notify the management personnel or on-site operators to take timely measures to prevent accidents; the operation guidance strategy refers to providing real-time operation guidance to the staff to help them avoid erroneous operations. For example, when it is found that a certain operating behavior may cause equipment failure, an operation suggestion or warning is pushed to guide the operator to perform the task correctly; finally, the first risk control strategy and the second risk control strategy are integrated to determine the risk control strategy, so as to realize comprehensive monitoring and management of personnel and equipment in the converter station, identify and respond to potential risks in advance, and ensure the safe operation of the converter station.

[0045] Furthermore, step S400 also includes step S450, constructing a risk rule engine, wherein the risk rule engine is used to match the response device-response method, and the response device is at least one risk management device; step S460, identifying the risk control strategy, combining with the risk rule engine, and outputting the risk control execution mode; step S470, performing converter station personnel behavior risk management according to the risk control execution mode.

[0046] Preferably, a risk rule engine is constructed to analyze events and behaviors according to preset rules and automatically take corresponding risk response measures, that is, different policy states use different warning methods, and adaptive adaptation management is performed based on the engine. In the risk control management of the converter station, the role of the risk rule engine is to select appropriate response devices and response methods according to different risk situations. The response device refers to the risk management device, including at least one, such as an alarm system, access control equipment, video surveillance equipment, emergency shutdown device, etc., which performs different actions according to the output of the risk rule engine, such as sounding an alarm, restricting people from entering specific areas, shutting down equipment, etc.; the response method refers to the different response methods taken in response to risk events. For example, if it is identified that a person has entered a dangerous area, the response method may be to sound an alarm and lock the access control; if the operating behavior of a certain device is found to be abnormal, the response method may be to suspend the operation of the device and notify the maintenance personnel.

[0047] Preferably, the risk control strategy is identified and combined with the risk rule engine, so that the system can automatically identify and execute appropriate management measures according to specific risk events, and then output the risk control execution mode, that is, the execution plan generated based on the current risk assessment results and the risk control strategy, including which equipment needs to be started, which risk management measures need to be executed, and how to adjust the response strategy according to different risk levels and types. For example, if an operator is identified to enter a sensitive area without a work ticket, the risk rule engine will automatically select the corresponding response device (such as an access control system) based on this risk control strategy to restrict personnel entry, and handle it through a preset response method (such as an alarm, report); then, according to the risk control execution mode, the risk management of personnel behavior in the converter station is carried out, that is, by monitoring, analyzing and responding to the behavior of the staff, it is ensured that their operations in the converter station comply with safety regulations and effectively prevent potential safety risks, thereby realizing real-time monitoring of personnel behavior and automated risk management. When potential personnel behavior risks are detected, it can respond quickly and accurately, and take corresponding countermeasures to ensure work safety and equipment stability in the converter station.

[0048] Furthermore, step S400 also includes step S480, introducing a temporary database, which is used to store risk management data; step S490, setting an incremental learning cycle, calling the periodic risk management data based on the temporary database, and deleting it within the database; step S500, updating the risk association rules based on the periodic risk management data, and performing incremental learning on the risk prediction model.

[0049] Preferably, a temporary database is introduced to temporarily store, process and manage data. The data in the temporary database is usually time-sensitive and the stored data will not be retained for a long time. The data in the temporary database is more used for short-term analysis and calculation, and will be cleared or updated over time. Incremental learning refers to continuous learning from new data acquired, setting an incremental learning cycle, and in each incremental learning cycle, according to the set cycle (such as daily, weekly, etc.), extracting risk management data within a certain time range from the temporary database for analysis and learning. At the end of each cycle, the processed risk management data will be deleted from the temporary database to make room for storing new data and avoid data backlogs, ensuring that the temporary database will not store a large amount of outdated data and maintaining efficient operation of the database. Finally, based on the periodic risk management data at the end of the incremental learning cycle, the existing risk association rules are updated, usually including discovering new associations and adjusting existing rules, and then incremental learning is performed on the risk prediction model based on these data to update the model, thereby enhancing the risk prediction model's ability to predict new risk events and ensuring the accuracy and flexibility of the prediction.

[0050] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for controlling the behavior of converter station personnel based on risk prediction, characterized in that: The method comprises: Distributed deployment of door control security equipment and visual sensing equipment inside the converter station, wherein the door control security equipment is used for personnel authority management; Determine the dynamic data of the converter station based on the supervision of the door control security equipment and the visual sensor equipment and the Beidou positioning based on the terminal worn by the personnel, wherein the dynamic data of the converter station is marked with a collection timestamp; Interact with the risk event records of converter stations, mine risk association rules, and combine with the three-dimensional modeling of converter stations to conduct sample-driven training based on time series to build a risk prediction model, which includes twin modeling of event locations and simplified modeling of non-event locations; The dynamic data of the converter station is transmitted back, and the personnel behavior characteristics are analyzed in combination with the risk prediction model to determine the risk control strategy, and the risk management device deployed at the front end is responded to for alarm management, wherein the risk prediction model is communicatively connected with the risk management device.

2. A method for controlling converter station personnel behavior based on risk prediction according to claim 1, characterized in that: Interactive converter station risk event records and mining risk association rules, including: Traverse the risk event records and integrate to determine M event sequences, wherein the event location-event type-feature item is used as the integration standard, and the feature item is the event-related feature; Clustering the M event sequences based on event position-event type to determine N groups of event sequences, where N is a positive integer less than or equal to M; Based on the N groups of event sequences, mining N event association trees, wherein feature items are screened by mutual information calculation and symmetric uncertainty calculation; The risk association rule is determined based on the N event association trees.

3. A method for controlling converter station personnel behavior based on risk prediction according to claim 2, characterized in that: Based on the N groups of event sequences, N event association trees are mined, including: Identifying a set of event sequences, wherein the set of event sequences includes a plurality of sets of feature items corresponding to each event; Setting a minimum support, performing support statistics on feature items of the set of event sequences, and screening strongly associated feature items that meet the minimum support, wherein support statistics are performed based on feature item frequencies; The associated feature items are integrated by recursive multi-layer refinement of features to construct a first event association tree.

4. A method for controlling converter station personnel behavior based on risk prediction according to claim 3, characterized in that: After the first event association tree is constructed, it includes: Traverse the first event association tree and calculate the mutual information coefficient of the connection nodes, where the calculation formula is: , F(X) and F(Y) represent the information entropy of connecting nodes X and Y respectively, and F(X,Y) represents the joint information entropy of X and Y; Symmetric uncertainty calculation is performed on the connection nodes, where the calculation formula is: ; A mutual information threshold and a symmetric uncertainty threshold are set to prune the first event association tree.

5. The method for controlling converter station personnel behavior based on risk prediction according to claim 2, characterized in that: The three-dimensional modeling of the converter station includes: Identify the event locations of the N event association trees, perform twin modeling for the event locations of the converter station, and determine a digital twin; Simplifying the modeling of the non-event position of the converter station to determine the simplified model; The digital twin and the simplified model are spatially spliced ​​together to form a converter station model.

6. A method for controlling converter station personnel behavior based on risk prediction according to claim 1, characterized in that: Combined with the risk prediction model, analyze the behavioral characteristics of personnel and determine the risk control strategy, including: Monitoring the dynamic data of the converter station and analyzing user behavior characteristics, including operation behavior characteristics and trajectory behavior characteristics; Based on the work ticket data scanned by the gate control and the trajectory behavior characteristics, threshold management is performed to determine the first risk control strategy; Based on the operational behavior characteristics, risk event twin deduction is performed to determine a second risk control strategy, wherein the second risk control strategy includes an alarm strategy and an operational guidance strategy; Determine the risk control strategy according to the first risk control strategy and the second risk control strategy.

7. The method for controlling converter station personnel behavior based on risk prediction according to claim 1, characterized in that: Respond to the risk management equipment deployed at the front end for alarm management, including: Constructing a risk rule engine, wherein the risk rule engine is used to match a response device with a response mode, and the response device is at least one risk management device; Identify the risk control strategy, combine with the risk rule engine, and output the risk control execution mode; According to the risk control execution mode, the risk management of converter station personnel behavior is carried out.

8. The method for controlling converter station personnel behavior based on risk prediction according to claim 1, characterized in that: After responding to the risk management equipment deployed at the front end for alarm management, it includes: Introducing a temporary database, wherein the temporary database is used to store risk management data; Setting an incremental learning cycle, calling periodic risk management data based on the temporary database, and deleting the data in the database; Based on the periodic risk management data, the risk association rules are updated, and incremental learning is performed on the risk prediction model.

Images