Intelligent multi-power-supply automatic switching control system for electric power system

Abstract

The invention discloses an intelligent multi-power-supply automatic switching control system for a power system, and the system comprises the steps: collecting the operation data of a high-voltage-side power supply and a low-voltage-side power supply, mapping the operation data in a digital twin model through a multi-dimensional state, and generating a load coupling coefficient matrix; taking the load coupling coefficient matrix as an input, and carrying out a virtual power flow disturbance test to identify an overload critical state; fusing the real-time operation data of the critical state and digital twinborn model with the power flow response characteristics to generate a load partitioning result and an optimal switching proportion of each region; executing a cross-side bidirectional load flexible transfer control strategy according to a load partitioning result, and performing closed-loop adjustment on a transfer proportion according to operation data and power flow response characteristics in real time; and running data, power flow response characteristics and load partition results in the switching execution process are written back to the digital twinborn model in an original mode, so that the protection precision and the response speed in a future N-2 fault scene are improved. According to the invention, the problem of overload of'one transformer with a whole station 'under repeated configuration of ATSs at a high-voltage side and a low-voltage side and N-2 faults can be solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric power, in particular to an intelligent multi-power automatic switching control system for power system. BACKGROUND

[0002] In modern power systems, in order to ensure the reliability and continuity of power supply, automatic switching devices (ATS) are often used to realize automatic switching of power supply to cope with equipment failure or load fluctuation. However, with the continuous expansion of the scale of the power system and the complication of equipment configuration, the traditional ATS configuration faces some key challenges, especially in the case of repeated configuration of high-voltage side and low-voltage side ATS, which is particularly prominent. Repeated configuration not only causes power supply path redundancy, but also may cause unnecessary power switching, increasing the risk of system instability.

[0003] At the same time, in the N-2 failure mode, that is, the case where any two important power elements in the system fail, often leads to extreme power demand fluctuations, increasing the risk of single transformer overload. When a fault occurs, the power system may have an "all stations on one transformer" overload phenomenon, that is, the load carried by a single transformer suddenly increases far beyond its design load limit, resulting in transformer damage, power interruption and even larger range of system failure. This failure mode is particularly prone to occur in the case of multiple power sources in parallel or strong coupling between power equipment, posing a huge challenge to the safety and reliability of the power grid.

[0004] Therefore, solving the problems of repeated configuration of high-voltage side and low-voltage side ATS and "one transformer carrying all stations" overload in N-2 failure has become a key technical issue to improve the reliability and stability of the power system. Only through intelligent multi-power automatic switching control method, can the negative effects of redundant switching path be more effectively avoided, and more accurate load reconstruction can be realized when a fault occurs, ensuring the continuous and stable operation of the power grid. SUMMARY

[0005] In order to solve the problems of repeated configuration of high-voltage side and low-voltage side ATS and "one transformer carrying all stations" overload in N-2 failure, the present application provides an intelligent multi-power automatic switching control system for power system.

[0006] To achieve the above purpose, the technical solution adopted by the present application is as follows:

[0007] The present application discloses an intelligent multi-power automatic switching control system for power system, comprising the following steps:

[0008] Data acquisition module: acquire the operation data of high-voltage side power supply and low-voltage side power supply, model in the unified operation digital twin model of high-voltage side power supply and low-voltage side power supply by multi-dimensional state mapping, and extract and save the power flow response characteristics from the model;

[0009] The load coupling coefficient matrix generation module: decomposes the operation data and the power flow response characteristics in space and time, calculates and generates the load coupling coefficient matrix, and identifies and labels the invalid switching loop of the repeated configuration automatic transfer switch based on the matrix;

[0010] The N-2 fault scenario deduction module: takes the load coupling coefficient matrix as input, starts the N-2 fault scenario deduction, performs virtual power flow disturbance test to identify the critical state of overload, and outputs the corresponding power flow response characteristics;

[0011] The dynamic partition switching module: fuses the critical state with the real-time operation data and the power flow response characteristics of the unified operation digital twin model, triggers adaptive load partitioning, generates and locks the load partitioning result and the optimal switching ratio of each zone;

[0012] The bidirectional cross-side flexible transfer module: performs the cross-side bidirectional load flexible transfer control strategy according to the load partitioning result, controls the high-voltage side power supply and the low-voltage side power supply to proportionally adjust the bidirectional shunt, and adjusts the transfer ratio in real time with the operation data and the power flow response characteristics in a closed loop, and restricts any transformer from exceeding the upper limit of the safe load;

[0013] The closed-loop self-evolution optimization module: writes the operation data, the power flow response characteristics and the load partitioning result during the switching execution process to the unified operation digital twin model as they are, updates the load coupling coefficient matrix and the cross-side bidirectional load flexible transfer control strategy parameters based on the written data, to improve the protection accuracy and response speed in future N-2 fault scenarios.

[0014] Further, the data acquisition module comprises:

[0015] The operation data of the high-voltage side power supply and the low-voltage side power supply comprises real-time acquisition of voltage, current, frequency and power flow values by deploying synchronous measurement units at each measurement node of the high-voltage side power supply and the low-voltage side power supply;

[0016] The operation data is first subjected to outlier rejection and interpolation completion, and then input into a multi-dimensional state mapping engine, which maps the spatial distribution and time series of the operation data to the corresponding virtual nodes and virtual lines of the unified operation digital twin model; the unified operation digital twin model establishes a virtual computing environment consistent with the actual system using network topology and electrical parameters, and performs power flow calculation and state prediction while continuously receiving operation data; according to the power flow calculation results, the power distribution, phase change and voltage fluctuation of each node and line under different operating conditions are extracted from the model as power flow response characteristics.

[0017] Further, the load coupling coefficient matrix generation module comprises:

[0018] In the unified operation digital twin model, firstly, the operation data and the power flow response characteristics are sequentially processed in the time dimension, and the nodes and lines are correspondingly mapped in the spatial dimension according to the network topological structure, to form a multi-dimensional data set which can reflect the time evolution and spatial distribution at the same time;

[0019] The multi-dimensional data set is decomposed by using a space-time decomposition algorithm, and a dependency index of each load node on the power supply path of the high-voltage side power supply and the low-voltage side power supply under different operation states is extracted;

[0020] Through normalization and correlation calculation on the dependency index, a load coupling strength value between any two power supply paths is obtained, and the load coupling strength value is filled into a load coupling coefficient matrix according to the path combination;

[0021] The load coupling coefficient matrix is subjected to threshold screening and connectivity analysis, to automatically identify the redundant power supply loop between the high-voltage side power supply and the low-voltage side power supply, and to label the invalid switching loop with repeated configuration automatic transfer switches and no actual switching value in combination with the network topology.

[0022] Further, the N-2 fault scenario deduction module comprises:

[0023] Taking the load coupling coefficient matrix as input, firstly, all possible double-element simultaneous failure combinations are automatically generated according to the network topology in the unified operation digital twin model;

[0024] For each combination, a virtual power flow disturbance is applied in the digital twin model, the corresponding elements are forcibly closed, and the full-network power flow distribution and node voltage are recalculated;

[0025] The calculation results are cross-analyzed with the load coupling coefficient matrix to determine the load redistribution of each power supply path and the transformer load change amplitude in the fault state;

[0026] When it is detected that a combination causes a single transformer to bear more than a safe load threshold, it is determined as a critical state that may cause overload;

[0027] For each critical state, the difference mode of the operation data before and after the fault is extracted synchronously, the corresponding power flow response characteristics are formed, and the power flow response characteristics are output in a standardized format.

[0028] Further, the dynamic partition switching module comprises:

[0029] The critical state output by the N-2 fault scenario deduction module is introduced into the unified operation digital twin model, and is synchronously fused with the real-time operation data and the power flow response characteristics in the model;

[0030] The full station load is clustered by using a load coupling coefficient matrix, load nodes with high coupling degree are divided into the same block, and the block boundary is dynamically adjusted in combination with the critical state risk level;

[0031] Based on the power flow response characteristics, the optimal switching ratio between the high-voltage side power supply and the low-voltage side power supply of each block is calculated to minimize the overload risk in the fault scenario.

[0032] The generated and locked load partitioning result and the optimal switching ratio will be used as the direct input of the subsequent cross-side bidirectional load flexible transfer control, ensuring the accuracy and protection effect of the switching action.

[0033] Further: the bidirectional cross-side flexible transfer module comprises:

[0034] When the cross-side bidirectional load flexible transfer control strategy is executed according to the load partitioning result, the load partitioning result and the optimal switching ratio in the unified operation digital twin model are first loaded into the switching execution engine, and the corresponding control channel is established;

[0035] The switching execution engine issues a shunt instruction to the related circuit breaker and automatic transfer switch through a high-speed communication network, and completes the bidirectional load transfer between the high-voltage side power supply and the low-voltage side power supply in milliseconds.

[0036] During the transfer process, real-time operation data is collected and power flow calculation is performed in the digital twin model to obtain updated power flow response characteristics.

[0037] According to the closed-loop feedback of the power flow response characteristics and the operation data, the bidirectional shunt ratio is dynamically fine-tuned to keep the actual load distribution consistent with the optimal switching ratio.

[0038] The path load related to the transformer in the load coupling coefficient matrix is continuously monitored, and when any transformer approaches the upper limit of the safe load, the switching ratio is adjusted or part of the load transfer is delayed.

[0039] Compared with the prior art, the technical progress achieved by the present application is:

[0040] In the traditional system, load distribution often relies on preset rules, and lacks flexibility in adapting to actual operating conditions and load changes, leading to possible supply path redundancy and unnecessary power switching. The present application constructs a load coupling coefficient matrix to evaluate the coupling strength of each power supply path in real time, and accurately divides the load block in combination with the power flow response characteristics. This data-based dynamic optimization enables the system to automatically adjust the power supply path according to real-time load changes, ensuring efficient and balanced load transfer in the event of a fault, thereby avoiding the risk of overload caused by uneven load distribution.

[0041] Traditional power systems rely on static fault-tolerant mechanisms, which cannot accurately predict and respond to risks in complex fault scenarios, especially in the case of multiple transformers or multiple parallel power sources, which can easily cause single device overload and further cause widespread power outages or equipment damage. The present application can identify the critical state that may lead to "one transformer overload" through fault scenario deduction and real-time simulation of power flow distribution and load reconstruction after N-2 fault. This module not only analyzes the load redistribution of each power supply path, but also automatically adjusts the switching strategy according to the real-time power flow response characteristics, ensuring that transformers and key devices do not operate under overload during a fault, avoiding system collapse due to insufficient prediction or slow response in the prior art.

[0042] In traditional automatic switching control systems, devices and switching strategies are usually based on static rules, and cannot be intelligently adjusted in real time according to changes in system state. This static approach often responds slowly or has inaccurate strategies when a fault occurs, making it difficult to meet safety requirements in high-load or complex fault scenarios. By continuously collecting operating data, power flow response characteristics and load partition results, and dynamically learning through online self-evolution algorithms, the present application can continuously optimize control strategies to minimize system risks with each switch. At the same time, the adaptive ability of the system can be optimized in real time according to historical data of different fault scenarios, thereby enhancing the identification and response capabilities for future faults, and maintaining high reliability of the entire power system during long-term operation.

[0043] Notably, the present application has strong device protection capabilities. Traditional ATS systems may ignore the safe load of certain devices when dealing with complex faults, leading to device overload or damage. The present application can adjust the switching ratio in real time to ensure that each transformer operates within its safe load range by monitoring the load conditions of transformers and key devices in real time, combined with adaptive load partitioning and load flexible transfer control strategies. This fine-grained device protection avoids device failure and system downtime caused by overload.

[0044] In summary, the present application not only solves the problem of repeated configuration of high-voltage and low-voltage ATS and "one transformer overload" under N-2 fault, but also significantly improves the overall stability and fault handling capability of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0045] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, to explain the present application, and do not constitute a limitation of the present application.

[0046] In the drawings:

[0047] Figure 1 The figure is a system structure diagram of the present application. DETAILED DESCRIPTION

[0048] The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described again in some embodiments. The embodiments of the present application will be described below with reference to the accompanying drawings.

[0049] As shown in Figure 1 The application discloses an intelligent multi-power automatic switching control system for a power system, comprising:

[0050] The data acquisition module acquires voltage, current, frequency and real-time power flow measurement data of the high-voltage side power supply and the low-voltage side power supply, defines the measurement data as "operation data", maps the operation data in a unified operation digital twin model of the high-voltage side power supply and the low-voltage side power supply by using multi-dimensional state, and extracts and saves "power flow response characteristics" from the model.

[0051] The load coupling coefficient matrix generation module performs space-time decomposition on the operation data and the power flow response characteristics in the unified operation digital twin model of the high-voltage side power supply and the low-voltage side power supply, calculates and generates a "load coupling coefficient matrix", and identifies and labels invalid switching loops of the repeatedly configured automatic transfer switch based on the matrix.

[0052] The N-2 fault scenario deduction module takes the load coupling coefficient matrix as input, starts the N-2 fault scenario deduction module, performs virtual power flow disturbance test to identify a critical state that may cause "one transformer overload of the whole station", and outputs corresponding power flow response characteristics.

[0053] The dynamic partition switching module fuses the critical state output by the N-2 fault scenario deduction module and the real-time operation data and the power flow response characteristics of the unified operation digital twin model of the high-voltage side power supply and the low-voltage side power supply, triggers the adaptive load partition module, generates and locks "load partition results" and an optimal switching ratio of each zone.

[0054] The bidirectional cross-side flexible transfer module performs a cross-side bidirectional load flexible transfer control strategy according to the load partition results, controls the high-voltage side power supply and the low-voltage side power supply to perform millisecond-level proportional adjustable bidirectional shunting, adjusts the transfer ratio in real time with the operation data and the power flow response characteristics in a closed loop, and restricts any transformer to not exceed the upper limit of the safe load.

[0055] The closed-loop self-evolution optimization module writes the operation data, the power flow response characteristics and the load partition results in the process of switching execution back to the unified operation digital twin model of the high-voltage side power supply and the low-voltage side power supply, updates the load coupling coefficient matrix and the cross-side bidirectional load flexible transfer control strategy parameters based on the written data by using an online self-evolution algorithm, so as to improve the protection accuracy and response speed in future N-2 fault scenarios.

[0056] Specifically, the data acquisition module comprises:

[0057] First, the synchronous measurement units (PMU) and high-precision intelligent sensors are deployed at the key measurement nodes of the high-voltage side power supply and the low-voltage side power supply to collect the voltage, current, frequency and power flow data in real time. The synchronous measurement units realize the unified timestamp of all measurement point data through the global positioning system (GPS) or other high-precision time synchronization technology, guarantee the time sequence consistency of cross-node and cross-voltage level data, and thus form a high-precision, time-synchronized operation data stream. The collected raw operation data is preprocessed, in this embodiment, the statistical anomaly detection algorithm is used to eliminate abnormal values caused by sampling errors and sensor noise, and the data completion algorithm based on spatiotemporal interpolation and trend prediction is used to complete the data with missing data, to ensure the integrity and continuity of the data.

[0058] The preprocessed operation data is input into the multi-dimensional state mapping engine, which maps the spatial distribution characteristics and time sequence dynamic change characteristics of the operation data to the corresponding virtual nodes and virtual lines in the unified operation digital twin model of the high-voltage side power supply and the low-voltage side power supply by constructing the spatiotemporal feature vector, combining the geographical location, electrical connection and electrical parameters of the grid nodes. In this embodiment, the mapping process adopts a method based on graph neural network (GNN), which can capture the nonlinear coupling relationship between nodes in a complex grid structure and support incremental updating of real-time data.

[0059] The unified operation digital twin model of the high-voltage side power supply and the low-voltage side power supply is based on the real grid topology and electrical parameters to construct a virtual computing environment, realizing a simulation platform highly consistent with the actual system. The digital twin model adopts a distributed algorithm based on power flow calculation to perform real-time active and reactive power flow calculation, node voltage verification, phase angle synchronization and state prediction on the basis of continuous reception of operation data, ensuring the accurate reflection of the model on the actual system operation state. According to the power flow calculation results, the digital twin model extracts key operation parameters from the virtual nodes and lines, including power distribution, phase angle change, voltage amplitude fluctuation, etc., to form a unified standard power flow response feature set. The power flow response features are stored in a structured data format, supporting fast calling and efficient transmission, serving as the basis input for the subsequent calculation of the load coupling coefficient matrix, N-2 fault scenario deduction and adaptive load partitioning module.

[0060] Specifically, the load coupling coefficient matrix generation module comprises:

[0061] In the high-voltage side power supply and low-voltage side power supply unified operation digital twin model, first, the collected operation data and power flow response characteristics are arranged in time sequence to form continuous and synchronous time axis data sequences. For the spatial dimension, the data is mapped to the corresponding virtual nodes and lines using the topology information of the power grid, and a multi-dimensional data set covering time and space dimensions is constructed, which reflects the dynamic changes of the load of each measuring point with time and the spatial distribution and electrical connection relationship between the load nodes.

[0062] The space-time decomposition algorithm is deployed on this multi-dimensional data set. The algorithm combines high-order tensor decomposition technology and time series feature extraction method. First, the time dimension is identified to capture periodic and mutation characteristics. Second, in the spatial dimension, the electrical coupling relationship between nodes and the load transmission path are extracted by relying on the power grid topology of the digital twin model through graph signal processing. Through this joint space-time decomposition, the dependency index of each load node on each power supply path under different operating conditions is obtained. These index quantifies the degree of electrical energy transmission and interconnection between load nodes and power supply paths. Subsequently, the dependency index is normalized to eliminate the bias caused by dimensional differences and ensure that the index is comparable on a unified scale. Based on the normalized index, the correlation between any two power supply paths is calculated to form a load coupling strength value, which reflects the mutual influence and sharing degree of the two power supply paths in load support. The coupling strength values of all path combinations are filled into the matrix to construct a complete load coupling coefficient matrix. Each element of the matrix represents the coupling strength between the corresponding paths, and the construction and selection of the load coupling coefficient matrix provide quantitative and operable load coupling indicators.

[0063] The load coupling coefficient matrix is screened by a pre-set threshold to eliminate path pairs with coupling strength below the threshold, reducing the calculation complexity and interference information. Through connectivity analysis based on graph theory, redundant power supply loops existing in the high-voltage side power supply and low-voltage side power supply network are identified. Combined with the network topology structure of the digital twin model, the system further locates the repeatedly configured automatic transfer switches (ATS) and determines the actual utility of their switching operations under the current load coupling and network structure. For automatic transfer switches that do not produce load diversion benefits and have redundant switching, the system labels them as invalid switching loops. This invalid switching loop information is fed back to the switching control module to optimize the switching strategy, avoid redundant switching actions caused by repeated configuration of automatic transfer switches, improve system operation stability and equipment life, and effectively alleviate load conflicts and management complexity caused by repeated configuration.

[0064] Through the above method, the system can scientifically and accurately quantify the load coupling relationship between the high-voltage side power supply and the low-voltage side power supply, automatically eliminate the invalid switching loops caused by repeated configuration, and provide a basis for subsequent N-2 fault protection and intelligent switching control.

[0065] Specifically, the N-2 fault scenario deduction module comprises:

[0066] In the N-2 fault scenario deduction module, first, the grid network topology information saved in the unified operation digital twin model of the high-voltage side power supply and the low-voltage side power supply is called, the information contains the connection relationship and electrical parameters of transformers, incoming lines, key tie lines and other important power elements, and all possible double-element failure scenarios are automatically generated through a combination algorithm to ensure that the joint fault of any two elements including two transformers, two incoming lines, a combination of a transformer and an incoming line, and a key tie line is covered. For each double-element failure combination, a virtual fault operation is performed in the digital twin model, that is, the corresponding transformer, incoming line or tie line virtual node is forcibly closed. The digital twin model performs real-time power flow calculation based on the latest operation data and network parameters, redistributes the active and reactive power flow of the entire network, calculates the voltage amplitude and phase angle changes of all nodes, and simulates the actual operation state of the power grid after the fault occurs.

[0067] The simulation results are cross-analyzed with the load coupling coefficient matrix. Specifically, the coupling strength information in the load coupling coefficient matrix is used to analyze the load redistribution of different power supply paths in the fault state, and the increase and decrease amplitudes of the load carried by each transformer are calculated. By comparing the calculated load of each transformer with the designed safety load threshold, the module determines whether there is an overload risk. When at least one transformer load exceeds the safety threshold due to a combination, it is identified as a potential "one transformer carrying the entire station" overload critical state.

[0068] For each identified critical state, the module further extracts the differences in operation data before and after the fault, including voltage, current, frequency and power flow change characteristics. These difference characteristics are standardized to form a unified format of power flow response feature data set as the feature label of the fault scenario.

[0069] All critical states and their corresponding power flow response features are systematically sorted and archived as important inputs for the subsequent adaptive load partitioning module to identify risks and reorganize loads, ensuring that the switching strategy can accurately target high-risk fault scenarios, improving the overall protection capability and operation safety of the system.

[0070] This implementation relies on the high-precision simulation capability of the digital twin model and the load correlation analysis of the load coupling coefficient matrix, combines automatic fault combination generation and virtual power flow disturbance calculation, and builds an N-2 fault risk assessment system that is comprehensive, timely and accurate.

[0071] Specifically, the dynamic partitioning and switching module comprises:

[0072] In this link, the critical state information output by the N-2 fault scenario deduction module is first introduced into the high-voltage side power supply and low-voltage side power supply unified operation digital twin model. In order to ensure the accuracy of data fusion, the system uses a time alignment algorithm to strictly synchronize the time stamp corresponding to the critical state with the real-time operation data and power flow response characteristics in the digital twin model, ensuring complete consistency in the time dimension. Spatial mapping is based on the detailed network topology information in the digital twin model, accurately mapping the power flow distribution, node voltage and load change trend of the critical state to the corresponding virtual nodes and lines, realizing the spatial correspondence between the critical state characteristics and the real-time operation state.

[0073] After time-space fusion is completed, the adaptive load partitioning module calls the load coupling coefficient matrix to perform in-depth clustering analysis on the full-station load. The clustering algorithm, based on the coupling degree threshold, classifies load nodes with high coupling strength into the same block, ensuring the electrical relevance and dynamic coupling of the partition. In response to different risk levels corresponding to different critical states, the module dynamically adjusts the block boundaries, refines or merges the load blocks, and enhances the adaptability and response capability of the partition results to actual fault risks. Subsequently, the module uses the power flow response characteristics corresponding to the critical state, combined with the load coupling within and between blocks, to calculate the optimal switching ratio. This calculation, through a multi-objective optimization method, balances load balancing, transformer safety load limits, and minimizes switching times, generating a switching scheme that meets the safe and efficient allocation of loads under fault conditions. Finally, the generated load partitioning results and corresponding optimal switching ratio are locked in a structured data format and real-time issued to the cross-side bidirectional load flexible transfer control module. This ensures that the execution of the switching control is based on accurate load partitioning strategies and dynamic optimization parameters, effectively reducing the risk of "one transformer affecting the entire station" overload and improving the robustness and safety of the power system under N-2 fault scenarios.

[0074] This implementation relies on high-precision time-space synchronization fusion algorithms, intelligent clustering partitioning techniques based on load coupling coefficient matrices, and multi-objective optimization switching strategies combined with power flow response characteristics, forming a closed-loop intelligent protection system from fault identification to load reconstruction.

[0075] Specifically, the bidirectional cross-side flexible transfer module includes:

[0076] In the execution phase of the cross-side bidirectional load flexible transfer control strategy, the load partitioning results and optimal switching ratio parameters generated in the high-voltage side power supply and low-voltage side power supply unified operation digital twin model are first loaded into the switching execution engine. The switching execution engine establishes a dedicated control channel based on these data, ensuring reliable and real-time high-speed communication connection with the target circuit breaker and automatic transfer switch.

[0077] The shunt command is accurately issued to the specified circuit breaker and automatic transfer switch through high-speed industrial Ethernet or dedicated power communication network, realizing the bidirectional transfer of load between high-voltage side power supply and low-voltage side power supply. The transfer operation is completed in milliseconds, ensuring that the response speed meets the dynamic adjustment requirements of the power system, and avoiding the risk of load fluctuation or overload due to delay. During the transfer process, the system continuously collects real-time operating data (including voltage, current, frequency and power flow) of each measuring point and synchronously inputs the digital twin model for online power flow calculation. The model predicts and updates the power flow response characteristics based on the latest data, reflecting the current system load distribution and electrical state changes. By building a closed-loop feedback mechanism of operating data and power flow response characteristics, the actual load distribution is compared with the preset optimal switching ratio in real time, and the cross-side bidirectional shunt ratio is automatically fine-tuned. The adjustment algorithm is based on proportional control and model predictive control (MPC) method, ensuring smooth and accurate load transfer process, and preventing sudden load impact.

[0078] At the same time, the system monitors the load of the power supply path involving the transformer in the load coupling coefficient matrix in real time. When detecting that the load of any transformer approaches the upper limit of its designed safe load, the safety constraint mechanism is immediately started: automatically adjusting the current load transfer ratio, and delaying part of the load transfer operation if necessary, to prevent transformer overload and ensure safe operation of power equipment. Through the integration of real-time calculation capability of digital twin model, high-speed communication control link and intelligent closed-loop feedback algorithm, this process realizes efficient, safe and accurate cross-side bidirectional load flexible transfer, significantly improving the dynamic regulation and control capability and operation stability of the power system in N-2 fault scenarios.

[0079] Specifically, the closed-loop self-evolution optimization module includes:

[0080] During the switching execution process, the system collects real-time operating data including voltage, current, frequency and power flow through the synchronous measurement units deployed at key measuring points of high-voltage side power supply and low-voltage side power supply, and monitors the power flow response characteristics and corresponding load partition results generated by switching action. All collected data are synchronized by time stamping and then written back to the database of the unified operating digital twin model of high-voltage side power supply and low-voltage side power supply in real time. This database not only stores historical and real-time data, but also serves as a training data warehouse, providing high-quality and time-series complete input data for subsequent online self-evolution algorithms.

[0081] Online self-evolution algorithm is based on incremental learning framework, which can receive and utilize the latest feedback data for model updating in real time without full retraining, significantly improving learning efficiency and response speed. Specifically, the algorithm first uses new data to dynamically correct the load coupling coefficient matrix, by adjusting the coupling strength between each power supply path in the matrix, to correct the coupling deviation caused by changes in operating environment, equipment state fluctuations or load changes, ensuring the timeliness and accuracy of the load coupling coefficient matrix. At the same time, the algorithm combines real-time load partition results with actual switching effect, and uses the method based on reinforcement learning and Bayesian optimization to adaptively optimize the key parameters of the cross-side bidirectional load flexible transfer control strategy, such as transfer proportion adjustment range, response speed threshold and transformer safety load capacity. The optimization process is based on feedback loop, that is, the algorithm constantly compares the deviation between expected control effect and actual operation result, and automatically adjusts the control parameters to reduce the gap, so as to improve the switching accuracy and response efficiency of the system under actual fault scenarios.

[0082] The self-evolution mechanism relies on the high-precision simulation and data synchronization capability of the unified operation digital twin model, combined with efficient incremental learning and optimization algorithms, to realize the continuous intelligent upgrading of the model and control strategy. Through continuous learning and optimization, the power system can more accurately identify N-2 fault risks, improve switching response speed, strengthen system robustness and adaptability, and ensure the safe and stable operation of the power grid.

[0083] Finally, it should be noted that: the above only for the preferred embodiments of the present application, and not for the purpose of limiting the present application, although the foregoing detailed description of the present application is made with reference to the foregoing embodiments, for those skilled in the art, it still can be modified to the technical solutions recorded in the foregoing embodiments, or equivalent replacement of some technical features. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the scope of protection of the claims of the present application.

Claims

1. An intelligent multi-source automatic switching control system for power system, characterized in that, The method comprises the following steps: a data acquisition module: collect the operation data of the high-voltage side power supply and the low-voltage side power supply, map the operation data in a unified operation digital twin model of the high-voltage side power supply and the low-voltage side power supply by using multi-dimensional state mapping, and extract and save the power flow response characteristics from the model; a load coupling coefficient matrix generation module: perform time-space decomposition on the operation data and the power flow response characteristics, calculate and generate a load coupling coefficient matrix, and identify and label invalid switching loops of the repeated configuration automatic transfer switch based on the matrix; an N-2 fault scenario deduction module: input the load coupling coefficient matrix, start N-2 fault scenario deduction, perform virtual power flow disturbance test to identify the critical state of overload, and output the corresponding power flow response characteristics; a dynamic partition switching module: fuse the critical state, real-time operation data and power flow response characteristics of the unified operation digital twin model, trigger adaptive load partitioning, generate and lock the load partitioning result and the optimal switching ratio of each zone; a bidirectional cross-side flexible transfer module: perform cross-side bidirectional load flexible transfer control strategy according to the load partitioning result, control the high-voltage side power supply and the low-voltage side power supply to perform proportional adjustable bidirectional shunting, real-time close-loop adjust the transfer ratio with operation data and power flow response characteristics, and constrain any transformer to not exceed the upper limit of the safe load; a closed-loop self-evolution optimization module: write the operation data, power flow response characteristics and load partitioning result during switching execution to the unified operation digital twin model as they are, update the load coupling coefficient matrix and the cross-side bidirectional load flexible transfer control strategy parameters based on the written data, so as to improve the protection accuracy and response speed in future N-2 fault scenarios.

2. The intelligent multi-source automatic switching control system for power system according to claim 1, characterized in that, The data acquisition module comprises: The operation data of the high-voltage side power supply and the low-voltage side power supply comprises real-time acquisition of voltage, current, frequency and power flow values by deploying synchronous measurement units at each measurement node of the high-voltage side power supply and the low-voltage side power supply; The operation data is first subjected to outlier rejection and interpolation completion, and then input into a multi-dimensional state mapping engine, which maps the spatial distribution and time sequence of the operation data to corresponding virtual nodes and virtual lines of the unified operation digital twin model; the unified operation digital twin model establishes a virtual computing environment consistent with the actual system by using network topology and electrical parameters, and performs power flow calculation and state prediction while continuously receiving operation data; according to the power flow calculation result, the power distribution, phase change and voltage fluctuation of each node and line under different operating states are extracted from the model as power flow response characteristics.

3. The intelligent multi-source automatic switching control system for power system according to claim 2, characterized in that, The load coupling coefficient matrix generation module comprises: In the unified operation digital twin model, the operation data and the power flow response characteristics are first subjected to sequence processing according to the time dimension, and corresponding mapping of nodes and lines according to the network topology structure in the spatial dimension, forming a multi-dimensional data set that can reflect time evolution and spatial distribution at the same time; characteristic decomposition is performed on the multi-dimensional data set by using a time-space decomposition algorithm, and the dependence index of each load node on the power supply path of the high-voltage side power supply and the low-voltage side power supply under different operating states is extracted; By normalizing and correlating the dependency indicators, the load coupling strength values between any two power supply paths are obtained, and they are filled into the load coupling coefficient matrix according to the path combination; The load coupling coefficient matrix is subjected to threshold screening and connectivity analysis to automatically identify the redundant power supply loop between the high-voltage side power supply and the low-voltage side power supply, and to label the invalid switching loop with repeated configuration automatic transfer switches and no actual switching value in combination with the network topology.

4. The intelligent multi-source automatic switching control system for power system according to claim 3, characterized in that, The N-2 fault scenario deduction module comprises: Taking the load coupling coefficient matrix as input, firstly, according to the network topology in the unified operation digital twin model, all possible double-element simultaneous failure combinations are automatically generated; For each combination, a virtual power flow disturbance is applied in the digital twin model, the corresponding elements are forcibly closed, and the full-network power flow distribution and node voltage are recalculated; The calculation results are cross-analyzed with the load coupling coefficient matrix to determine the load redistribution of each power supply path and the transformer load variation amplitude in the fault state; When it is detected that a combination causes a single transformer to carry more than the safe load threshold, it is determined as a critical state that may cause overload; For each critical state, the difference mode of the pre-fault and post-fault operation data is synchronously extracted to form the corresponding power flow response feature, and is output in a standardized format.

5. The intelligent multi-source automatic switching control system for power system according to claim 4, characterized in that, The dynamic partition switching module comprises: The critical state output by the N-2 fault scenario deduction module is imported into the unified operation digital twin model, and is synchronously fused with the real-time operation data and the power flow response feature in the model; Using the load coupling coefficient matrix, the loads of the whole station are clustered and analyzed, the load nodes with high coupling degree are divided into the same block, and the block boundary is dynamically adjusted in combination with the critical state risk level; Based on the power flow response feature, the optimal switching ratio between the high-voltage side power supply and the low-voltage side power supply for each block is calculated to minimize the overload risk in the fault scenario; The generated and locked load partition result and optimal switching ratio will be directly input for subsequent cross-side bidirectional load flexible transfer control, ensuring the accuracy and protection effect of the switching action.

6. The intelligent multi-source automatic switching control system for power system according to claim 4, characterized in that, The bidirectional cross-side flexible transfer module comprises: When executing the cross-side bidirectional load flexible transfer control strategy according to the load partition result, firstly, the load partition result and the optimal switching ratio in the unified operation digital twin model are loaded into the switching execution engine, and the corresponding control channel is established; The switching execution engine issues a shunt instruction to the related circuit breakers and automatic transfer switches through a high-speed communication network to complete the bidirectional load transfer between the high-voltage side power supply and the low-voltage side power supply in milliseconds; During the transfer process, real-time operation data is collected and power flow calculation is performed in the digital twin model to obtain updated power flow response features; According to the closed-loop feedback of the power flow response feature and the operation data, the bidirectional shunt ratio is dynamically fine-tuned to keep the actual load distribution consistent with the optimal switching ratio; The transformer-related path load in the load coupling coefficient matrix is continuously monitored, and when any transformer approaches the upper limit of the safe load, the switching ratio is adjusted or part of the load transfer is delayed.

Images