Transformer substation remote optimization control method and system based on digital twinning

Abstract

The invention relates to the technical field of transformer substation remote control, in particular to a transformer substation remote optimization control method and system based on digital twinning, and the method comprises the steps: calling a main transformer association node number to unfold a path structure, extracting a circuit breaker and an isolation number to form a topological sequence, simulating a voltage change trend to generate a statement set, and matching the risk semantics to generate a sorting result, comparing main and auxiliary path load linkage behaviors, and generating a remote control action instruction. The method comprises the following steps: constructing a topological chain of main transformer associated nodes, describing a position relationship among electrical nodes, summarizing a trend set through a multi-period voltage trend, extracting a risk sorting result through an associated semantic index, comparing load changes of main and auxiliary paths to identify linkage interference characteristics, and screening and adjusting path combinations to generate a control instruction. And the state adaptability and the scene cooperation capability of the control strategy are enhanced.

Description

A method and system for remote optimization control of substations based on digital twins Technical Field

[0001] This invention relates to the field of substation remote control technology, and in particular to a substation remote optimization control method and system based on digital twins. Background Technology

[0002] The field of substation remote control technology encompasses core aspects of centralized management, remote monitoring, and remote control of substation equipment within the power system. Its primary task is to achieve real-time remote control and status awareness of key equipment such as switching equipment, voltage regulators, power quality devices, and protection devices within the substation. Its overall development relies on the integration of communication networks, automation systems, and power dispatching systems to ensure efficient, safe, and stable operation of substations even under unattended or minimally staffed conditions. With the advancement of smart grids and the energy internet, this technology is gradually extending towards cyber-physical convergence, model-driven management, and state prediction, emphasizing system-level operational optimization and distributed resource coordination and control capabilities.

[0003] Among them, the digital twin-based substation remote optimization control method and system refers to constructing a digital model synchronously mapped to the actual substation. Based on this model, it realizes virtual perception of the substation's operating status, simulation of operation control schemes, and remote command generation. This involves constructing a virtual substation model using physical quantity acquisition and synchronous communication, relying on power equipment operating data to drive real-time modeling and status updates, and selecting the optimal scheme after multiple rounds of simulation evaluation of different control strategies through operating rule calculation methods. The central control node then remotely issues control commands to the field actuators to complete the actions. Its core lies in employing digital mapping and virtual-real interaction methods to enable the remote control decision-making process to have scenario matching and state adaptability.

[0004] Existing technologies focus on static modeling and state awareness, lacking a serialized path expression method based on node topology, making it difficult to fully represent the dynamic relationships between electrical nodes. The node state judgment does not involve multi-cycle trend evolution analysis, which easily misses the trend characteristics of node state changes. The risk identification process relies on a single state judgment and lacks a trend-level matching mechanism. In terms of load linkage identification, the behavior response mapping between the main and auxiliary paths has not been established, resulting in insufficient granularity of load interference analysis. The control command generation does not combine dynamic verification and path matching, which easily leads to problems of insufficient policy adaptability. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this invention provides a method and system for remote optimization control of substations based on digital twins. The technical solution is as follows: A method for remote optimization control of substations based on digital twins includes the following steps: S1: Calling the high-voltage bus voltage number mapped from the main transformer in the digital twin virtual space, extracting its associated circuit breaker status number and outgoing line isolation number, locating the number position according to the connection order of electrical nodes in the virtual structure, and arranging them sequentially according to the path number to form an operation monitoring topology path description sequence; S2: Calling all voltage numbers in the operation monitoring topology path description sequence, simulating voltage changes in the digital twin space in a continuous periodic progression manner, and judging the continuous rise and fall and repeated oscillations in the trend. S3: Extract the text information of direction changes to form a set of statements on the continuous change trend of node voltage; S4: Call the set of statements on the continuous change trend of node voltage, compare the trend description with the phrases, find the matching expression with the risk semantic vocabulary in the virtual space, group the risk expressions of high-frequency trend segments, and classify them into a list of statements ranked by risk trend level; S5: According to the operation monitoring topology path description sequence, compare the load behavior of the main control and auxiliary control nodes in the virtual path, track whether the change of the main control path in the current cycle triggers the auxiliary control response, and extract the synchronization behavior segments to form statements on load linkage interference between the main and auxiliary nodes.

[0006] As a further embodiment of the present invention, the operation monitoring topology path description sequence includes a node voltage number sequence, a circuit breaker and isolation number chain, and a relative position mapping relationship between nodes; the node voltage continuous change trend statement set includes voltage change direction expression, trend segment type identifier, and periodic sequence trend summary; the risk trend level sorting statement list includes trend semantic classification results, risk level correspondence, and trend type frequency sorting; and the load linkage interference statement between primary and secondary nodes includes primary control load change characteristics, secondary control response change segments, and load linkage behavior classification.

[0007] As a further embodiment of the present invention, the acquisition step of S1 is as follows: S101: Call the high-voltage bus voltage node number mapped by the main transformer in the digital twin virtual space, and construct an electrical connection topology for the node number, retrieve the status number and outgoing line isolation number of the circuit breaker directly connected to it, and aggregate its connection attribute information, arrange its numbering order according to the node connection direction in the topology diagram, and generate the high-voltage side node connection chain sequence number after the node sequence is constructed; S102: According to the high-voltage side node connection chain sequence number, identify the connection relationship between each node in its path topology, and extract the corresponding connection point. The monitoring nodes are numbered and a node index set is established. The sequential positions of each monitoring point in the node index set relative to the main transformer node are numbered and mapped. The position mapping encoding operation is performed on the mapping structure to generate a monitoring node relative position index table. S103: The number index information corresponding to each node in the monitoring node relative position index table is called. The original number structure corresponding to it in the path chain sequence is processed to construct a number chain. The path direction information is generated according to the arrangement order of each node in the chain. The path direction information is matched with the original topology sequence to obtain the running monitoring topology path description sequence.

[0008] As a further aspect of the present invention, the relative position number mapping relationship of each monitoring node in the relative position index table is established according to the position order of the high-voltage bus voltage node number mapped by the main transformer in the path topology, through a preset sequential arrangement rule; during the generation of the path direction information, the original number structure obtained by the number chain construction process is called, and sequential filtering is performed according to the arrangement order of the node connection direction in the topology diagram, combined with the field used to describe the connection direction in the connection attribute information, retaining only the node numbers whose connection direction meets the continuous conduction path condition; the structure matching processing of the running monitoring topology path description sequence is to match the connection relationship between nodes in the original topology structure sequence one by one according to the arrangement order of each node in the path direction information, and remove node numbers with incomplete connection attribute information or incorrect connection direction, and only perform the path description sequence generation operation on the node numbers that have completed all matching.

[0009] As a further embodiment of the present invention, the acquisition step of S2 is as follows: S201: Call the voltage number in the operation monitoring topology path description sequence, construct a voltage response sequence for ten consecutive control cycles in the digital twin space in a time-progressive manner, retrieve the change amplitude of all nodes in the voltage numerical dimension within each cycle, calculate the voltage difference of the same node between adjacent cycles and aggregate it according to the cycle sequence number to generate a node voltage response difference matrix; S202: According to the node voltage response difference matrix, determine the change of the difference sign of each node in the continuous cycle. If the difference sign is the same between adjacent cycles, it is marked as a unidirectional change segment. If the sign changes and there is alternation between positive and negative, it is marked as an oscillation segment. Perform structural clustering on all node change types according to the cycle order to obtain a node voltage change type sequence set; S203: Call the change identification information corresponding to each node in the node voltage change type sequence set, perform sequential splicing processing on adjacent node segments with the same change trend according to their continuous arrangement order in the ten cycles, and convert the spliced ​​sequence into text format to record the voltage change direction of each node in each cycle to obtain a node voltage continuous change trend statement set.

[0010] As a further embodiment of the present invention, the acquisition step of S3 is as follows: S301: Call each descriptive statement in the set of continuous change trend statements of node voltage, perform word segmentation on the word structure in each statement, and perform keyword extraction operation on the continuous change expression word groups in the word segmentation results. Perform matching and comparison operation on the extraction results with the word set of each risk level in the risk semantic index set established in the digital twin virtual space to generate a trend segment semantic matching group; S302: According to the risk level index information of the word groups in the trend segment semantic matching group, count the semantic matching frequency corresponding to each risk level, and aggregate the frequency results according to the trend segment number. Retrieve the risk level frequency proportion structure of each trend segment in all semantic matches to obtain the trend segment semantic frequency distribution table; S303: Call the risk level frequency value in the trend segment semantic frequency distribution table, sort all level frequencies mapped by the same trend segment in descending order, perform field fusion operation on the sorting result and the original trend segment number, and convert the sorting structure into a sequential expression format to obtain a risk trend level sorting statement list.

[0011] As a further embodiment of the present invention, the acquisition step of S4 is as follows: S401: Based on the path number in the operation monitoring topology path description sequence, monitor the load evolution trajectory of the main control and auxiliary control interval nodes in the digital twin space, collect the load value sequence of each node within the current control cycle, perform differential calculation on the load change of the main control path according to the time series, and synchronously arrange the load changes of the corresponding auxiliary control nodes to generate a comparable load evolution sequence, thereby obtaining the main and auxiliary node load comparison trajectory sequence; S402: Based on the main and auxiliary node load comparison trajectory sequence, determine the load change of the main control node. The change direction of the main control load and the auxiliary control node response change segment is consistent within the same period window. The change direction sign is compared periodically to mark consistent and inconsistent segments. The consistent segments are aggregated and encoded according to continuous periods to generate a main and auxiliary load synchronous change segment index set; S403: Call the synchronization segment number of each synchronization segment in the main and auxiliary load synchronous change segment index set, sequentially concatenate the corresponding main control load change description and auxiliary control response change description, and complete the statement field integration processing according to the path number. The concatenation result is converted into a text expression structure to obtain the load linkage interference statement between the main and auxiliary nodes.

[0012] As a further aspect of the present invention, the method further includes: S5: comparing the content expressed in the risk trend level sorting statement list with the action description in the load linkage interference statement between the main and auxiliary nodes, and extracting the verified operation path after verification of the control behavior in the digital twin space to form a substation remote optimization control action instruction; the substation remote optimization control action instruction includes a control path combination sequence, adjustment action expression statement, and a verified effective control scheme.

[0013] As a further embodiment of the present invention, the acquisition step of S5 is as follows: S501: Referencing the risk expression content in the risk trend level sorting statement list, and performing field association operations with the load behavior description extracted from the load linkage interference statement between the main and auxiliary nodes, performing position matching and semantic coupling discrimination on the semantic structure under the same node path, aggregating and archiving all matching associated field pairs according to the path number, and generating a load risk semantic coupling field set; S502: According to the load risk semantic coupling field set, performing behavioral logic assembly on the risk phrases and load change segments under each group of main and auxiliary node paths, simulating the control behavior path mapped by the logical assembly in the digital twin space, monitoring the node response status under each control path, and recording the effective verification status of each behavior path to obtain an effective control path identification table; S503: Calling the path number and adjustment behavior field identified as effective in the effective control path identification table, combining them in order according to their respective topology path numbers, and extracting all adjustment fields to form a structured instruction data frame, performing field concatenation encoding processing on the data frame, and obtaining the substation remote optimization control action instruction.

[0014] A remote optimization control system for substations based on digital twins includes: a path structure identification module, which acquires the voltage node number of the high-voltage busbar mapped by the main transformer in the digital twin virtual space, calls the status number and outgoing line isolation number of the circuit breaker connected to the node, unfolds the path structure according to the connection order recorded in the spatial model, identifies the relative arrangement relationship between all monitoring nodes, constructs a numbered chain according to the topology direction, and generates an operation monitoring topology path description sequence; a trend segment classification module, which calls the voltage number in the operation monitoring topology path description sequence, acquires the voltage response value sequence of each node in a continuous control cycle, determines whether it belongs to the rising segment, falling segment, or oscillation segment according to the changing direction of adjacent values ​​in the cycle sequence, and forms a trend statement expressing the continuous change state according to the arrangement order of nodes in the path, thus obtaining a set of node voltage continuous change trend statement statements; and a risk trend comparison module, which calls the trend statements in the set of node voltage continuous change trend statement statements, performs keyword matching in the risk semantic index set constructed in the digital twin virtual space, and extracts the trend segment text expression. The associated risk levels indicate fluctuation behaviors, and the frequency of each trend type in semantic matching is used as the basis for risk assessment to obtain a list of risk trend level ranking statements. The load response extraction module, based on the path number in the operation monitoring topology path description sequence, calls the load change information of each interval node in the main control path in the current cycle, and obtains the load response sequence of the corresponding auxiliary control path node. It compares whether the load change direction between the two path nodes is consistent in the cycle order, and marks the segment with obvious differences between nodes as the interference behavior segment. The node number and cycle sequence information are extracted from the interference segment to generate the load linkage interference statement between the main and auxiliary nodes. The control instruction generation module references the risk expression statement in the risk trend level ranking statement list, calls the node number and load behavior description in the load linkage interference statement between the main and auxiliary nodes, retrieves the control action path and adjustment statement related to the interference segment in the digital twin space, reads the operation path number and expression content that have been confirmed to be valid in the control verification, and combines them in the order of the number chain to form the substation remote optimization control action instruction.

[0015] The beneficial effects of the technical solution provided by the embodiments of the present invention include at least the following: In the present invention, by constructing a topological chain structure of the main transformer associated nodes, a continuous description of the positional relationship of electrical nodes is formed; by segmenting and summarizing the multi-cycle voltage change trend, a comparable trend set is formed; by the corresponding association between trend semantics and risk index, a sortable risk evolution result is formed; by comparing the load changes of the main and auxiliary paths in parallel, linkage interference characteristics are identified; and by screening adjustment paths through virtual pre-drilling and sequentially combining them into control commands, the adaptability of control decisions to changes in the operating situation is enhanced, and the control stability and coordination in complex scenarios are improved. Attached Figure Description

[0016] Figure 1 is a flowchart of the method of the present invention; Figure 2 is a flowchart of obtaining S1 of the present invention; Figure 3 is a flowchart of obtaining S2 of the present invention; Figure 4 is a flowchart of obtaining S3 of the present invention; Figure 5 is a flowchart of obtaining S4 of the present invention; Figure 6 is a flowchart of obtaining S5 of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] Please refer to Figure 1. This invention provides a technical solution: a remote optimization control method for substations based on digital twins, comprising the following steps: S1: Calling the high-voltage bus voltage node number mapped to the main transformer in the digital twin virtual space, extracting the circuit breaker status number and outgoing line isolation number connected to it, expanding the path structure according to the electrical node connection sequence in the virtual space, identifying the relative positional relationship of all monitoring points and mapping them to the corresponding numbered chains, obtaining the operation monitoring topology path description sequence; S2: Calling the voltage number in the operation monitoring topology path description sequence, simulating the voltage response change within ten consecutive control cycles in the digital twin space in a time-progressive manner, checking the change direction of each node in the continuous cycle, classifying the rise and fall segments and oscillation segments in the virtual evolution path, and forming a set of node voltage continuous change trend statements using sequential text; S3: Calling each description statement in the node voltage continuous change trend statement set, in the digital twin virtual space... S4: Based on the path number in the operation monitoring topology path description sequence, the load evolution trajectory of the main control and auxiliary control interval nodes is compared in the digital twin space. The auxiliary control response change segment caused by the load change in the main control path under the current cycle is extracted. It is determined whether the two show synchronous behavior in the virtual space. The load linkage interference statement between the main and auxiliary nodes is written in text form. S5: The risk expression content in the risk trend level sorting statement list is referenced and associated with the load behavior description extracted from the load linkage interference statement between the main and auxiliary nodes. After the control behavior is pre-performed in the digital twin space, the operation path and adjustment expression that have been verified as valid in the virtual space are read and combined into the substation remote optimization control action command according to the electrical path sequence.

[0023] The operation monitoring topology path description sequence includes node voltage number sequence, circuit breaker and isolation number chain, and relative position mapping relationship between nodes. The node voltage continuous change trend statement set includes voltage change direction expression, trend segment type identifier, and periodic sequence trend summary. The risk trend level sorting statement list includes trend semantic classification results, risk level correspondence, and trend type frequency sorting. The load linkage interference statement between main and auxiliary nodes includes main control load change characteristics, auxiliary control response change segments, and load linkage behavior classification. The substation remote optimization control action instructions include control path combination sequence, adjustment action expression statement, and verified effective control scheme.

[0024] Please refer to Figure 2. The steps for obtaining S1 are as follows: S101: Call the high-voltage bus voltage node number mapped to the main transformer in the digital twin virtual space, and construct an electrical connection topology for the node number. Retrieve the status number and outgoing line isolation number of the circuit breaker directly connected to it, and aggregate its connection attribute information. Arrange the numbering order according to the node connection direction in the topology diagram. After the node sequence is constructed, generate the high-voltage side node connection chain sequence number. Call the main transformer model with the number T_Main_01 in the digital twin virtual space, and lock the 220kV high-voltage bus voltage node mapped to its high-voltage side winding. The unique identifier number of this node in the system database is Node_HV_Bus_A1. For node number Node_HV_Bus_A1, the system uses a graph database traversal algorithm to retrieve electrical components with direct physical connections to this node. The search scope is limited to primary equipment such as circuit breakers and disconnectors. The circuit breaker status number directly connected to it is obtained as CB_Stat_2201 (current status value 1, indicating closed), and the outgoing line disconnector number is ISO_Line_2201_A (current status value 1, indicating closed). The system extracts the attribute information of these connected components, including equipment type ID, rated voltage level, current on / off status, and power flow direction identifier. Based on the predefined node connection direction in the topology graph, i.e., from the power supply side to the load side, the above nodes are arranged into an ordered sequence. For example, the initial sequence is arranged as high-voltage bus node, outgoing line disconnector, circuit breaker, and outgoing line node. After the node sequence is constructed, the physical ID of each component is replaced with the unified high-voltage side node connection chain sequence number within the system, generating a uniquely identified chain sequence Chain_HV_001.

[0025] S102: Based on the high-voltage side node connection chain sequence number, identify the connection relationships between nodes in the path topology, extract the monitoring node number corresponding to each connection point and establish a node index set. Map the sequence position of each monitoring point in the node index set relative to the main transformer node, and perform a position mapping encoding operation on this mapping structure to generate a monitoring node relative position index table. Based on the generated high-voltage side node connection chain sequence number Chain_HV_001, the system further identifies the connection relationships between adjacent nodes in the path topology, confirming the hierarchical structure of the connection between the bus node and the disconnecting switch, and the connection between the disconnecting switch and the circuit breaker. Extract the monitoring node number of the voltage monitoring device or power monitoring device corresponding to each connection point position and establish a node index set. Calculate the topological depth of each monitoring point in the set relative to the main transformer node; for example, the depth at the bus is 0, and the depth after the circuit breaker is 2. Use this depth value as the sequence position for number mapping. A position mapping encoding operation is performed on the mapping structure. Using binary encoding format, depth 0 is encoded as 0000 and depth 2 is encoded as 0010, generating a relative position index table of monitoring nodes. This table records the correspondence between monitoring IDs and binary position codes, establishing the logical coordinates of each monitoring point in the electrical topology.

[0026] S103: The system retrieves the index information corresponding to each node in the relative position index table of monitoring nodes, performs number chain construction on the original number structure corresponding to each node in the path chain sequence, generates path direction information based on the arrangement order of each node in the chain, and performs structural matching processing on the path direction information with the original topology sequence to obtain the running monitoring topology path description sequence. The system also retrieves the index information corresponding to each node in the relative position index table of monitoring nodes and backtracks to the original number structure corresponding to each node in the path chain sequence Chain_HV_001. The system performs number chain construction on the original number structure in the path chain sequence to form a doubly linked list structure. Based on the arrangement order of each node in the chain, and combined with the fields in the connection attribute information used to describe the connection direction, the system performs sequential filtering. The filtering logic is set as follows: only nodes whose connection direction field is identified as forward conduction or bidirectional conduction, and whose device status field is identified as closed, are retained. For example, if the system reads that the connection direction of a circuit breaker node is forward conduction and its status is closed, then the node is determined to meet the continuous conduction path condition and is retained; if a node's status is open, the chain is interrupted at this point, and subsequent nodes are removed. The filtered path information is matched with the original topology sequence. The node type and connection port ID are compared one by one. Nodes with missing attributes or connection directions that do not match the power flow calculation results are removed. Only the node numbers that have completed all matching verifications are used to generate path description sequences, and finally the operation monitoring topology path description sequence is obtained.

[0027] Please refer to Figure 3. The steps for obtaining S2 are as follows: S201: Call the voltage number in the running monitoring topology path description sequence, construct a voltage response sequence for ten consecutive control cycles in the digital twin space in a time-progressive manner, retrieve the change amplitude of all nodes in the voltage value dimension within each cycle, calculate the voltage difference between adjacent cycles for the same node and aggregate them according to the cycle sequence number to generate a node voltage response difference matrix; call the voltage number in the running monitoring topology path description sequence, for example, select the voltage data of the key monitoring node M_Node_02. In the digital twin space, set the time step to 0.1 seconds, advance 1 second from the current time as the reference, and construct a continuous voltage response sequence containing 10 control cycles. The collected voltage value sequence shows that the voltage value gradually changes from 220.5kV in the first cycle to 219.3kV in the tenth cycle. The system retrieves the change amplitude of nodes in the voltage value dimension within each cycle, and calculates the voltage difference between adjacent cycles by subtracting the voltage value of the previous cycle from the current cycle voltage value. For example, the difference in the first cycle is -0.1kV, the difference in the second cycle is -0.2kV, and so on. These differences are aggregated according to the cycle sequence number to generate a node voltage response difference matrix. This matrix expresses the differential change characteristics of each node in the time domain in a row list.

[0028] S202: Based on the node voltage response difference matrix, determine the sign change of the difference for each node in a continuous cycle. If the sign of the difference is the same between adjacent cycles, it is marked as a unidirectional change segment. If the sign changes and there is alternation between positive and negative, it is marked as an oscillation segment. All node change types are then structurally clustered according to the cycle order to obtain a sequence set of node voltage change types. Based on the node voltage response difference matrix, the system determines the sign change of the difference for each node in a continuous cycle. In this process, a voltage fluctuation dead zone threshold of 0.02kV is set. This threshold is set with reference to the lower limit of allowable deviation and measurement error in the power quality supply voltage deviation standard. Specifically, the standard deviation of voltage noise under normal operation is taken as 0.006kV, and three times this value (approximately 0.018kV) is rounded up to 0.02kV to effectively filter background noise. If the absolute value of the difference is less than this threshold, it is considered as no change. The system scans the difference sequence. If the difference sign remains consistent between adjacent cycles (e.g., consecutively negative), it is marked as a unidirectional change segment. If the sign changes and there is alternation between positive and negative (e.g., from negative to positive and then back to negative), it is marked as an oscillation segment. The system performs structural clustering on all node change types according to the cycle order. For example, four consecutive decreasing change cycles are clustered into a unidirectional decreasing segment, and the subsequent two increasing cycles are clustered into a rising segment, resulting in a set of node voltage change type sequences.

[0029] S203: The system retrieves the change identifier information corresponding to each node in the node voltage change type sequence set. Based on their continuous arrangement in ten cycles, it performs sequential concatenation processing on adjacent node segments with the same change trend, and converts this concatenated sequence into text format to record the voltage change direction of each node in each cycle, obtaining a set of continuous voltage change trend statements for the nodes. It retrieves the change identifier information corresponding to each node in the node voltage change type sequence set, such as "decline," "decline," "decline," "decline," "rebound," and "rebound." Based on their continuous arrangement in ten cycles, it performs sequential concatenation processing on adjacent node segments with the same change trend. The first four "decline" identifiers are concatenated into a description segment of "continuous decline for four cycles," and the following two "rebound" identifiers are concatenated into a description segment of "restorative rebound for two cycles." The system converts this concatenated sequence into text format, recording the voltage change direction and duration of each node in each cycle in natural language. For example, the generated description text clearly indicates that the node exhibits a continuous voltage drop within 0.4 seconds, followed by a slight rebound within 0.2 seconds. In this way, the system ultimately obtains a set of statements indicating the continuous change trend of node voltage.

[0030] Please refer to Figure 4. The steps for obtaining S3 are as follows: S301: Call each descriptive statement in the set of continuous change trend statements for node voltage, perform word segmentation on the phrase structure in each statement, and perform keyword extraction on the continuous change expression phrases in the word segmentation results. Perform matching and comparison operations with the phrase sets belonging to each risk level in the risk semantic index set established in the digital twin virtual space to generate trend segment semantic matching groups; call each descriptive statement in the set of continuous change trend statements for node voltage, for example, "Node 02 exhibits a voltage drop lasting 0.4 seconds". Perform word segmentation on the phrase structure in each statement, using the word segmentation algorithm in the natural language processing toolkit to segment the statement into independent word units such as "Node 02", "lasting", "voltage", "drop", and "0.4 seconds". The system performs keyword extraction on the word segmentation results, identifying "lasting" and "drop" as core change expression phrases. Subsequently, perform matching and comparison operations with the phrase sets belonging to each risk level in the risk semantic index set established in the digital twin virtual space. The risk semantic index set categorizes risks into different levels. For example, Level 1 risk includes "sharp drop," Level 2 risk includes "continuous drop," and Level 3 risk includes "slight fluctuation." During the matching process, the phrase "continuous drop" was successfully mapped to the Level 2 risk set, generating a trend segment semantic matching group containing trend segment number, keywords, and risk level identifier.

[0031] S302: Based on the risk level index information of the phrases in the trend segment semantic matching group, the system calculates the semantic matching frequency corresponding to each risk level, aggregates the frequency results according to the trend segment number, and retrieves the risk level frequency proportion structure of each trend segment in all semantic matches, obtaining a trend segment semantic frequency distribution table. For example, in ten historical matching records for a certain trend segment, level 2 risk was matched 5 times, level 1 risk was matched 2 times, and level 3 risk was matched 3 times. The system aggregates the frequency results according to the trend segment number and retrieves the risk level frequency proportion structure of each trend segment in all semantic matches. Calculations show that the frequency proportion of level 2 risk is 0.5 (i.e., 50%), level 1 risk is 0.2%, and level 3 risk is 0.3%. Based on these statistical data, the system constructs a detailed trend segment semantic frequency distribution table, intuitively displaying the distribution probability of the current voltage change trend at each risk level.

[0032] S303: The system retrieves the risk level frequency values ​​from the trend segment semantic frequency distribution table, sorts all levels mapped to the same trend segment in descending order, merges the sorting results with the original trend segment number, and converts the sorting structure into a sequential format to obtain a list of risk trend level sorting statements. The system retrieves the risk level frequency values ​​from the trend segment semantic frequency distribution table, sorts all levels mapped to the same trend segment in descending order. The sorting results show that level 2 risk has the highest proportion, followed by level 3, and level 1 has the lowest. The system merges this sorting result with the original trend segment number and converts the sorting structure into a sequential format. The generated descriptive text clearly indicates that this trend segment is mainly dominated by level 2 risk, followed by level 3 and level 1 risk. This sorting method effectively filters out occasional noise interference and identifies the main risk characteristic of the current voltage change as a moderate and persistent drop. Through the above processing, the system finally obtains a list of risk trend level sorting statements.

[0033] Please refer to Figure 5. The steps for obtaining S4 are as follows: S401: Based on the path number in the operation monitoring topology path description sequence, monitor the load evolution trajectory of the main control and auxiliary control interval nodes in the digital twin space, collect the load value sequence of each node within the current control cycle, perform differential calculation on the load change of the main control path according to the time series, and synchronously arrange the load changes of the corresponding auxiliary control nodes to generate a comparable load evolution sequence, thus obtaining the load comparison trajectory sequence of the main and auxiliary nodes; Based on the path number in the operation monitoring topology path description sequence, synchronously monitor the load evolution trajectory of the main control interval nodes (such as the low-voltage side bus of the main transformer) and auxiliary control interval nodes (such as the reactive power compensation branch) in the digital twin space. The system collects the load value sequence of each node within the current control cycle, such as the active power load sequence of the main control node and the reactive power load sequence of the auxiliary control node. Perform differential calculation on the load change of the main control path according to the time series to obtain the active power increment at each moment relative to the previous moment; at the same time, synchronously arrange the load changes of the corresponding auxiliary control nodes. By aligning the active power change sequence of the master control node with the reactive power change sequence of the auxiliary control node on the time axis, a comparable load evolution sequence containing the load changes of the master and auxiliary nodes at the same moment is generated, thus obtaining the load comparison trajectory sequence of the master and auxiliary nodes.

[0034] S402: Based on the load comparison trajectory sequence of primary and secondary nodes, the system determines the consistency of the change direction of the load change segment of the primary control node and the response change segment of the secondary control node within the same period window. The system compares the change direction signs period by period, marking consistent and inconsistent segments, and aggregating and encoding consistent segments according to continuous periods to generate an index set of synchronous change segments of primary and secondary loads. The system compares the change direction signs period by period: if at a certain moment, the change in primary control load is positive (increasing) and the change in secondary control load is also positive (increasing), then the two changes are considered to be in the same direction and marked as a consistent segment; if the directions are opposite, then they are marked as inconsistent segments. The system aggregates and encodes the time slices marked as consistent segments according to continuous periods, for example, aggregating six consecutive periods of consistent changes into one synchronous change segment. In this way, the system identifies strongly correlated periods between the growth of primary transformer load and the growth of reactive power demand, generating an index set of synchronous change segments of primary and secondary loads.

[0035] S403: The system retrieves the synchronization segment numbers from the main and auxiliary load synchronization change segment index set, sequentially concatenates the corresponding main control load change description and auxiliary control response change description, and performs statement field integration processing based on the path number. The concatenated result is then converted into a textual expression structure to obtain the load linkage interference statement between the main and auxiliary nodes. The system retrieves the corresponding main control load change description (e.g., "main transformer load continues to climb") and auxiliary control response change description (e.g., "reactive power demand increases synchronously") from the main and auxiliary load synchronization change segment index set. The system sequentially concatenates these two descriptions and performs statement field integration processing based on the path number. The concatenated result is converted into a standardized textual expression structure, for example, "Under path 001, the continuous increase in main transformer load and the synchronous increase in reactive power demand are positively correlated." This statement clearly reveals that the driving source of voltage changes is the load linkage effect, providing a causal basis for the subsequent control strategy formulation. Through this step, the system finally obtains the load linkage interference statement between the main and auxiliary nodes.

[0036] Please refer to Figure 6. The steps for obtaining S5 are as follows: S501: Referencing the risk expression content in the risk trend level sorting statement list, and performing a field association operation with the load behavior description extracted from the load linkage interference statement between main and auxiliary nodes, the semantic structure under the same node path is matched for position and semantic coupling is judged. All matching associated field pairs are aggregated and archived by path number to generate a load risk semantic coupling field set; Referencing the risk expression content (e.g., "Level 2 Risk: Continuous Drop") in the risk trend level sorting statement list, and performing a field association operation with the load behavior description extracted from the load linkage interference statement between main and auxiliary nodes (e.g., "Main transformer load continues to climb"), the semantic structure under the same node path is matched for position and semantic coupling is judged. The judgment logic is set as follows: If the risk type is a voltage drop, and the load behavior at the same time period is a load increase, then it is determined that there is a positive correlation between the two, that is, the causal chain of "load growth leads to voltage drop" is established. The system aggregates and archives all matching associated field pairs by path number to generate a load risk semantic coupling field set, clarifying the specific causes and sensitivity of the risk.

[0037] S502: Based on the load risk semantic coupling field set, the system assembles the risk terms and load change segments under each set of main and auxiliary node paths using behavioral logic. In the digital twin space, it simulates the control behavior paths mapped by this logical assembly, monitors the node response status under each control path, and records the effective verification status of each behavior path to obtain an effective control path identification table. For the "voltage drop" caused by "load increase," the system constructs reverse adjustment logic, namely, "increasing reactive power support to raise the voltage." In the digital twin space, the system simulates the control behavior paths mapped by this logical assembly, for example, simulating the connection of a 5Mvar capacitor. The system monitors the node response status under each control path. Through simulation calculations, it is found that after connecting the capacitor, the voltage of the target node effectively recovers from 219.3kV to 220.1kV without triggering side effects such as harmonic exceedance. The system records the valid verification status of each behavior path, and marks the path that successfully improves voltage without side effects as valid, thus obtaining a valid control path identification table.

[0038] S503: The system retrieves the path numbers and adjustment behavior fields identified as valid in the valid control path identification table, combines them sequentially according to their respective topology path numbers, and extracts all adjustment fields to form a structured instruction data frame. The data frame undergoes field concatenation encoding to obtain the substation remote optimization control action command. The system retrieves the path numbers and adjustment behavior fields identified as valid in the valid control path identification table, and combines them sequentially according to their respective topology path numbers. The system extracts all adjustment fields, including target equipment (capacitor bank), action type (engaged), and adjustment value (5Mvar), to form a structured instruction data frame. The data frame undergoes field concatenation encoding using a standard power telecontrol communication protocol (such as IEC 60870-5-104) to convert the control intent into hexadecimal instruction codes. After verification, this instruction code has the ability to directly drive field equipment. Through this series of rigorous analysis and verification steps, the system finally obtains the substation remote optimization control action command.

[0039] A remote optimization control system for substations based on digital twins includes: a path structure identification module, which acquires the voltage node numbers of the high-voltage busbars mapped to the main transformer in the digital twin virtual space, calls the status numbers and outgoing line isolation numbers of the circuit breakers connected to the nodes, unfolds the path structure according to the connection order recorded in the spatial model, identifies the relative arrangement relationships between all monitoring nodes, constructs a numbered chain according to the topological direction, and generates an operational monitoring topological path description sequence; a trend segment classification module, which calls the voltage numbers in the operational monitoring topological path description sequence, acquires the voltage response value sequence of each node in a continuous control cycle, determines whether it belongs to an rising segment, a falling segment, or an oscillating segment according to the direction of change of adjacent values ​​in the cycle sequence, and forms trend statements expressing the continuous change state according to the arrangement order of nodes in the path, thus obtaining a set of continuous change trend statements for node voltage; and a risk trend comparison module, which calls the trend statements in the set of continuous change trend statements for node voltage, performs keyword matching in the risk semantic index set constructed in the digital twin virtual space, and extracts the textual expressions related to the trend segments. The associated risk level indicates the fluctuation behavior. The frequency of each trend type in semantic matching is summarized as the basis for risk assessment, and a list of risk trend level ranking statements is obtained. The load response extraction module calls the load change information of each interval node in the main control path in the current cycle according to the path number in the operation monitoring topology path description sequence, and obtains the load response sequence of the corresponding auxiliary control path node. The load change direction between the two path nodes is compared in cycle order to see if they are consistent. The segment with obvious differences between the nodes is marked as the interference behavior segment. The node number and cycle sequence information are extracted from the interference segment to generate the load linkage interference statement between the main and auxiliary nodes. The control instruction generation module references the risk expression statement in the risk trend level ranking statement list, calls the node number and load behavior description in the load linkage interference statement between the main and auxiliary nodes, retrieves the control action path and adjustment statement related to the interference segment in the digital twin space, reads the operation path number and expression content that have been confirmed to be valid in the control verification, and combines them in the order of the number chain to form the substation remote optimization control action instruction.

[0040] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A remote optimization control method for substations based on digital twins, characterized in that, Includes the following steps: S1: Call the high-voltage bus voltage number mapped from the main transformer in the digital twin virtual space, extract its associated circuit breaker status number and outgoing line isolation number, locate the number position according to the connection order of electrical nodes in the virtual structure, and arrange them in sequence according to the path number to form an operation monitoring topology path description sequence; S2: Call all voltage numbers in the operation monitoring topology path description sequence, simulate voltage changes in the digital twin space in a continuous periodic manner, determine the continuous rise and fall and repeated oscillation sections in the trend, and extract the direction change text information to form a set of node voltage continuous change trend statements; S3: Call the set of statements on the continuous change trend of node voltage, compare the trend description with the phrases, find the matching expression with the risk semantic vocabulary in the virtual space, group the risk expressions of high-frequency trend segments, and classify them into a list of statements for ranking risk trend levels. S4: Based on the described sequence of operation monitoring topology paths, compare the load behavior of the master control and auxiliary control nodes in the virtual path, track whether the changes in the master control path in the current cycle trigger the auxiliary control response, and extract the synchronization behavior segments to form the load linkage interference statement between the master and auxiliary nodes.

2. The substation remote optimization control method based on digital twin as described in claim 1, characterized in that: The operation monitoring topology path description sequence includes a node voltage number sequence, a circuit breaker and isolation number chain, and a relative position mapping relationship between nodes. The node voltage continuous change trend statement set includes voltage change direction expression, trend segment type identifier, and periodic sequence trend summary. The risk trend level sorting statement list includes trend semantic classification results, risk level correspondence, and trend type frequency sorting. The load linkage interference statements between main and auxiliary nodes include main control load change characteristics, auxiliary control response change segments, and load linkage behavior classification.

3. The substation remote optimization control method based on digital twin as described in claim 1, characterized in that, The steps for obtaining S1 are as follows: S101: Call the high-voltage bus voltage node number mapped by the main transformer in the digital twin virtual space, and construct an electrical connection topology for the node number. Retrieve the status number and outgoing isolation number of the circuit breaker directly connected to it, and aggregate its connection attribute information. Arrange the numbering order according to the node connection direction in the topology diagram. After the node sequence is constructed, generate the high-voltage side node connection chain sequence number; S102: Based on the high-voltage side node connection chain sequence number, identify the connection relationship between each node in its path topology, extract the monitoring node number corresponding to each connection point and establish a node index set. Map the sequence position of each monitoring point in the node index set relative to the main transformer node with numbers, and perform position mapping encoding operation on the mapping structure to generate a monitoring node relative position index table; S103: Call the number index information corresponding to each node in the monitoring node relative position index table, perform number chain construction processing on the original numbering structure corresponding to it in the path chain sequence, and generate path direction information according to the arrangement order of each node in the chain. Perform structural matching processing on the path direction information and the original topology sequence to obtain the running monitoring topology path description sequence.

4. The substation remote optimization control method based on digital twin according to claim 3, characterized in that: The relative position number mapping relationship of each monitoring node in the relative position index table is established according to the position order of the high-voltage bus voltage node number mapped by the main transformer in the path topology, through a preset order arrangement rule. During the generation of the path direction information, the original number structure obtained by constructing the number chain is called, and the node connection direction in the topology diagram is sequentially filtered according to the arrangement order of the node connection direction, combined with the field used to describe the connection direction in the connection attribute information, retaining only the node numbers whose connection direction meets the condition of continuous conduction path. The structure matching process of the running monitoring topology path description sequence is to match the connection relationship between nodes in the original topology structure sequence one by one according to the arrangement order of each node in the path direction information, and remove node numbers with incomplete connection attribute information or incorrect connection direction, and only perform the path description sequence generation operation on the node numbers that have completed all matching.

5. The substation remote optimization control method based on digital twin according to claim 1, characterized in that, The steps for obtaining S2 are as follows: S201: Call the voltage number in the operation monitoring topology path description sequence, construct a voltage response sequence for ten consecutive control cycles in the digital twin space in a time-progressive manner, retrieve the change amplitude of all nodes in the voltage numerical dimension within each cycle, calculate the voltage difference of the same node between adjacent cycles and aggregate it according to the cycle sequence number to generate a node voltage response difference matrix; S202: According to the node voltage response difference matrix, determine the sign change of the difference of each node in the continuous cycle. If the difference sign is the same between adjacent cycles, it is marked as a unidirectional change segment. If the sign changes and there is alternation between positive and negative, it is marked as an oscillation segment. Perform structural clustering on all node change types according to the cycle order to obtain a node voltage change type sequence set; S203: Call the change identification information corresponding to each node in the node voltage change type sequence set, perform sequential splicing processing on adjacent node segments with the same change trend according to their continuous arrangement order in the ten cycles, and convert the spliced ​​sequence into text format to record the voltage change direction of each node in each cycle to obtain a node voltage continuous change trend statement set.

6. The substation remote optimization control method based on digital twin according to claim 1, characterized in that, The steps for obtaining S3 are as follows: S301: Call each descriptive statement in the set of continuous change trend statements of node voltage, perform word segmentation on the word structure in each statement, and perform keyword extraction on the continuous change expression words in the word segmentation results. Perform matching and comparison operations on the extraction results with the word set of each risk level in the risk semantic index set established in the digital twin virtual space to generate a trend segment semantic matching group; S302: According to the risk level index information of the word group in the trend segment semantic matching group, count the semantic matching frequency corresponding to each risk level, and aggregate the frequency results according to the trend segment number. Retrieve the risk level frequency proportion structure of each trend segment in all semantic matches to obtain the trend segment semantic frequency distribution table; S3 03: Call the risk level frequency values ​​in the trend segment semantic frequency distribution table, sort all the level frequencies mapped by the same trend segment in descending order, perform field fusion operation between the sorting result and the original trend segment number, and convert the sorting structure into a sequential expression format to obtain a list of risk trend level sorting statements.

7. The substation remote optimization control method based on digital twin according to claim 1, characterized in that, The acquisition steps of S4 are as follows: S401: Based on the path number in the operation monitoring topology path description sequence, monitor the load evolution trajectory of the main control and auxiliary control interval nodes in the digital twin space, collect the load value sequence of each node within the current control cycle, perform differential calculation on the load change of the main control path according to the time series, and synchronously arrange the load changes of the corresponding auxiliary control nodes to generate a comparable load evolution sequence, thus obtaining the main and auxiliary node load comparison trajectory sequence; S402: Based on the main and auxiliary node load comparison trajectory sequence, determine the consistency of the change direction of the main control node load change segment and the auxiliary control node response change segment within the same period window, compare the change direction signs cycle by cycle, mark the consistent segments and inconsistent segments, and aggregate and encode the consistent segments according to the continuous cycle to generate the main and auxiliary load synchronous change segment index set; S403: Call the synchronous segment numbers of each synchronous segment in the main and auxiliary load synchronous change segment index set, sequentially concatenate the corresponding main control load change description and auxiliary control response change description, and complete the statement field integration processing according to the path number, convert the concatenation result into a text expression structure, and obtain the load linkage interference statement between the main and auxiliary nodes.

8. The substation remote optimization control method based on digital twin according to claim 1, characterized in that, The method further includes: S5: comparing the content expressed in the risk trend level sorting statement list with the action description in the load linkage interference statement between the main and auxiliary nodes, and extracting the verified operation path after verifying the control behavior in the digital twin space to form the substation remote optimization control action instruction; the substation remote optimization control action instruction includes the control path combination sequence, the adjustment action expression statement, and the verified effective control scheme.

9. The substation remote optimization control method based on digital twin according to claim 8, characterized in that, The steps for obtaining S5 are as follows: S501: Referencing the risk expression content in the risk trend level sorting statement list, and performing field association operations with the load behavior description extracted from the load linkage interference statement between the main and auxiliary nodes, performing position matching and semantic coupling discrimination on the semantic structure under the same node path, aggregating and archiving all matching associated field pairs by path number to generate a load risk semantic coupling field set; S502: According to the load risk semantic coupling field set, performing behavioral logic assembly on the risk phrases and load change segments under each group of main and auxiliary node paths, simulating the control behavior path mapped by the logical assembly in the digital twin space, monitoring the node response status under each control path, and recording the effective verification status of each behavior path to obtain an effective control path identification table; S503: Calling the path number and adjustment behavior field identified as effective in the effective control path identification table, combining them in order according to their respective topology path numbers, and extracting all adjustment fields to form a structured instruction data frame, performing field concatenation encoding on the data frame to obtain the substation remote optimization control action instruction.

10. A remote optimization control system for substations based on digital twins, characterized in that, The system is used in the substation remote optimization control method based on digital twins as described in any one of claims 1-9. The system includes: a path structure identification module, which obtains the high-voltage bus voltage node number mapped to the main transformer in the digital twin virtual space, calls the circuit breaker status number and outgoing line isolation number connected to the node, expands the path structure according to the connection order recorded in the spatial model, identifies the relative arrangement relationship between all monitoring nodes, constructs a numbered chain according to the topology direction, and generates an operation monitoring topology path description sequence; a trend segment classification module, which calls the voltage number in the operation monitoring topology path description sequence, obtains the voltage response value sequence of each node in a continuous control cycle, determines whether it belongs to an rising segment, a falling segment, or an oscillating segment based on the direction of change of adjacent values ​​in the cycle sequence, and forms trend statements expressing the continuously changing state according to the node arrangement order in the path, obtaining a set of node voltage continuous change trend statements; and a risk trend comparison module, which calls the trend statements in the set of node voltage continuous change trend statements, performs keyword matching in the risk semantic index set constructed in the digital twin virtual space, and extracts... The risk level associated with the trend segment text expression indicates the fluctuation behavior. The frequency of each trend type in semantic matching is summarized as the basis for risk assessment, and a list of risk trend level ranking statements is obtained. The load response extraction module, based on the path number in the operation monitoring topology path description sequence, calls the load change information of each interval node in the main control path in the current cycle, and obtains the load response sequence of the corresponding auxiliary control path node. The load change direction between the two path nodes is compared in cyclic order to see if they are consistent. The segment with obvious differences between the nodes is marked as the interference behavior segment. The node number and cycle sequence information are extracted from the interference segment to generate the load linkage interference statement between the main and auxiliary nodes. The control instruction generation module references the risk expression statement in the risk trend level ranking statement list, calls the node number and load behavior description in the load linkage interference statement between the main and auxiliary nodes, retrieves the control action path and adjustment statement related to the interference segment in the digital twin space, reads the operation path number and expression content that have been confirmed to be valid in the control verification, and combines them in the order of the number chain to form the substation remote optimization control action instruction.

Images