A power plant electrical full-element sub-real-time simulation method and related equipment

Abstract

The application discloses a power plant electrical full-element sub-real-time simulation method and related equipment, and relates to the technical field of electrical simulation. The method comprises the following steps: constructing a full-element electromagnetic transient model of an electrical system of a power plant, and determining a segmentation position according to the topological characteristics of the full-element electromagnetic transient model; segmenting the full-element electromagnetic transient model into multiple subsystems based on the determined segmentation position, and distributing each subsystem to different processor cores; and after converting time-varying sinusoidal quantities transmitted at each interface into direct current components under steady state, controlling each processor core to perform simulation operation by using the received subsystems. The application guarantees high-fidelity precision of simulation and fully releases parallel computing efficiency.

Description

Technical Field

[0001] This invention relates to the field of electrical simulation technology, and in particular to a real-time simulation method and related equipment for analyzing all electrical elements of a power plant. Background Technology

[0002] The safe and economical operation of nuclear power plants highly depends on the real-time and accurate perception and prediction of the status of their high- and medium-voltage electrical systems. Constructing a full-element digital twin covering generator outlets, plant transformers, medium-voltage distribution busbars, and critical loads to achieve high-fidelity real-time electromagnetic transient simulation is a core technological approach. Due to the dense nodes, complex topology, and short-range electrical connections in the high- and medium-voltage electrical systems of nuclear power plants, the computational burden for full-detail simulation is enormous, necessitating parallel computing techniques using model segmentation to achieve real-time performance. This technological requirement also exists in electrical systems with dense, short-range network characteristics, such as the integrated power grids of large-scale renewable energy plants, integrated power systems for ships, microgrids in high-energy-consuming industrial enterprises, and traction power supply systems for urban rail transit.

[0003] In current parallel simulations of power systems, various model partitioning methods have been developed to address the need for model partitioning. The long-transmission line partitioning method utilizes electromagnetic wave propagation delay to achieve model decoupling. By dividing the network into multiple subsystems along sufficiently long electrical transmission lines, each subsystem is computed independently without data exchange during wave propagation time, thus enabling parallel simulation. The ideal transformer model method allows partitioning at arbitrary network locations. By introducing an ideal transformer and a unit delay element at the partition location, the voltage or current of one subsystem is delayed by one simulation step before being transmitted to the other subsystem, thereby achieving network decoupling and parallel computation.

[0004] However, the existing model segmentation methods mentioned above all have significant drawbacks when dealing with the dense short-range power grids of nuclear power plants. Long-distance transmission line segmentation requires sufficiently long electrical distances between segmentation points, with wave propagation times exceeding the simulation step size. However, the internal power grids of nuclear power plants are tightly connected with short electrical distances, allowing electromagnetic waves to propagate within microseconds, far less than the typical simulation step size, rendering this method completely ineffective. While the ideal transformer model allows for arbitrary network point segmentation, the introduced unit delay creates feedback loops, and simulation stability strictly depends on the Thevenin equivalent impedance ratio on both sides of the interface, converging only when the impedance ratio is less than 1. The electrical system impedance network of nuclear power plants is complex and its operating modes are variable, making it difficult to guarantee that all segmentation points meet this convergence condition. This severely restricts the selection of segmentation points to impedance conditions, resulting in extremely low flexibility and universality, and making it prone to divergence under transient faults, posing engineering risks. Existing improvement methods, such as adding passive compensation circuits and improving interface algorithms, either alter the original network topology or lack universality, failing to fundamentally solve the deep dependence of the ideal transformer model on impedance conditions. Summary of the Invention

[0005] This invention aims to address the shortcomings of existing technologies, specifically providing a real-time simulation method and related equipment for the comprehensive analysis of all electrical elements in power plants, as detailed below: 1) In a first aspect, the present invention provides a real-time simulation method for analyzing all electrical elements of a power plant, the specific technical solution of which is as follows: Construct a full-element electromagnetic transient model of the power plant's electrical system, and determine the segmentation locations based on the topological characteristics of the full-element electromagnetic transient model; Based on the determined segmentation positions, the full-element electromagnetic transient model is divided into multiple subsystems, and each subsystem is assigned to a different processor core. Each segmentation position corresponds to a subsystem on both sides, and each segmentation position corresponds to an interface. Each interface is composed of the boundary nodes of the subsystems on both sides of the corresponding segmentation position. After converting the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component, the processor cores are controlled to perform simulation operation using the subsystems they receive.

[0006] The beneficial effects of the real-time simulation method for full-element electrical analysis of power plants provided by this invention are as follows: By constructing a full-element electromagnetic transient model and determining the segmentation positions based on topological characteristics, the selection of segmentation positions is not constrained by impedance conditions, allowing segmentation at any topological location. This fundamentally eliminates the dependence of segmentation flexibility on the interface impedance ratio. Based on the determined segmentation positions, the full-element electromagnetic transient model is divided into multiple subsystems and assigned to different processor cores. Each segmentation position corresponds to a subsystem on both sides, and each segmentation position corresponds to an interface composed of boundary nodes of the subsystems on both sides. This allows each subsystem to be assigned to different processor cores for parallel computing, fully utilizing the computing power of each processor core. After converting the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component, each processor core is controlled to run the simulation using its respective received subsystem. This eliminates the delay error introduced by the time-varying characteristics of the sinusoidal quantity in discrete data exchange and avoids the accumulation and amplification of steady-state errors on both sides of the interface, making the simulation stability no longer dependent on the impedance conditions on both sides of the interface. The combined effect of these technical solutions enables the full-element segmentation real-time simulation of dense short-range power grids in power plants to run stably at any segmentation position, ensuring both high-fidelity simulation accuracy and fully releasing the efficiency of parallel computing.

[0007] 2) Secondly, the present invention also provides a real-time simulation system for the full analysis of electrical elements in a power plant, the specific technical solution of which is as follows: This includes a construction determination module, a partitioning and allocation module, and a simulation module; The construction determination module is used to: construct a full-element electromagnetic transient model of the power plant's electrical system, and determine the segmentation location based on the topological characteristics of the full-element electromagnetic transient model; The segmentation and allocation module is used to: divide the full-element electromagnetic transient model into multiple subsystems based on the determined segmentation positions, and allocate each subsystem to different processor cores. Each segmentation position corresponds to a subsystem on both sides, and each segmentation position corresponds to an interface. Each interface is composed of the boundary nodes of the subsystems on both sides of the corresponding segmentation position. The simulation module is used to convert the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component, and then control each processor core to perform simulation operation using the subsystems it receives.

[0008] 3) In a third aspect, the present invention also provides an electronic device, the electronic device including a processor coupled to a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor, so as to enable the electronic device to implement any of the above-mentioned real-time simulation methods for the full analysis of electrical elements of a power plant.

[0009] 4) In a fourth aspect, the present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements any of the above-mentioned real-time simulation methods for the analysis of all electrical elements of a power plant.

[0010] It should be noted that the beneficial effects of the technical solutions of the second to fourth aspects of the present invention and their corresponding possible implementations can be found in the above description of the technical effects of the first aspect and its corresponding possible implementations, and will not be repeated here. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below: Figure 1 This is a flowchart illustrating a real-time simulation method for analyzing all electrical elements in a power plant, according to an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the data exchange format under an event-triggered synchronization signal mechanism. Figure 3 This is a schematic diagram comparing the voltage of phase A of the original circuit with the ideal transformer model method and constant calibration method under high impedance ratio. Figure 4 This is a schematic diagram comparing the voltage of phase A of the original circuit with the ideal transformer model method and constant calibration method under low impedance ratio. Figure 5 This is a schematic diagram comparing the voltage of phase A of the original circuit with the ideal transformer model method and constant calibration method under low impedance ratio in fault conditions. Figure 6 This is a schematic diagram comparing the ideal transformer model method, the constant calibration method, and the A-phase current of the original circuit under low impedance ratio in fault conditions. Figure 7 This is a schematic diagram of a real-time simulation system for analyzing all electrical elements of a power plant, according to an embodiment of the present invention. Detailed Implementation

[0012] like Figure 1 As shown in the figure, a real-time simulation method for full-element analysis of electrical systems in a power plant according to an embodiment of the present invention includes the following steps: S1. Construct a full-element electromagnetic transient model of the power plant's electrical system, and determine the segmentation location based on the topological characteristics of the full-element electromagnetic transient model. The full-element breakdown real-time simulation method provided by this invention is not only applicable to the high- and medium-voltage electrical systems of nuclear power plants (in this case, the electrical system of the power plant in step S1 is the high- and medium-voltage electrical system of the nuclear power plant), but can also be extended to scenarios such as the grid of large-scale new energy power plants, integrated power systems of ships, microgrids of high-energy-consuming industrial enterprises, and traction power supply systems of urban rail transit. In the grid of large-scale photovoltaic or wind farms, a large number of inverters and transformer substations are connected by dense short-range cables, with numerous nodes and complex topology. This method can flexibly divide the network without being limited by impedance ratio, realizing real-time simulation of the transient process of the entire field. In integrated power systems of ships, generator sets, power distribution equipment, and propulsion... The electrical distance between loads is extremely short, and electromagnetic transient interactions are intense. This method can solve the problem of traditional segmentation methods failing due to short-range limitations, supporting real-time system status assessment. In the internal microgrids of high-energy-consuming enterprises, frequent switching of impact loads such as electric arc furnaces and large motors causes severe transient disturbances. The event-triggered synchronization mechanism of this method can maintain simulation stability and quickly self-recover accuracy under fault disturbances, providing high-fidelity data for enterprise power system operation. In urban rail transit traction power supply systems, the electrical connection between substations and the overhead contact line is compact, and train movement causes frequent topology changes. This method allows for stable segmentation at any segmentation point and can adapt to the parallel real-time simulation requirements under dynamic topology changes. In this application, the high- and medium-voltage electrical systems of nuclear power plants are used as an example for illustration. The specific implementation process of S1 is as follows: S10. Construct a full-element electromagnetic transient model of the high- and medium-voltage electrical system of the nuclear power plant. This full-element electromagnetic transient model refers to a complete electromagnetic transient simulation model that includes all key electrical equipment and their interconnections within the high- and medium-voltage electrical system of the nuclear power plant. The full-element electromagnetic transient model covers the complete electrical path from the generator outlet through the plant service transformer and medium-voltage distribution busbar to various important loads. The full-element electromagnetic transient model is constructed using graphical modeling and simulation software. Specifically, using the power system module library provided by the graphical modeling and simulation software, the main electrical components of the high- and medium-voltage electrical system of the nuclear power plant, such as generators, transformers, busbars, cables, and dynamic loads, are added one by one to the full-element electromagnetic transient model. For generators, a corresponding mathematical model is established based on their electromagnetic characteristics, including the electrical parameters of the stator and rotor windings. For transformers, a corresponding transformer model is established based on parameters such as rated capacity, turns ratio, and short-circuit impedance. For buses, a bus node model is established based on their voltage level and connection relationships. For cables, a distributed parameter transmission line model is established based on parameters such as length, resistance per unit length, inductance per unit length, and capacitance per unit length. For dynamic loads, a corresponding load model is established based on their power characteristics and dynamic response characteristics. During the construction of the full-element electromagnetic transient model, all electrical components are connected according to the actual physical topology of the high- and medium-voltage electrical systems of a nuclear power plant, ensuring that the electrical connections in the full-element electromagnetic transient model are completely consistent with the actual system. After the full-element electromagnetic transient model is constructed, simulation parameters are set to enable the full-element electromagnetic transient model to perform electromagnetic transient simulations.

[0013] S11. Determine the segmentation location based on the topological characteristics of the full-element electromagnetic transient model. Topological characteristics refer to the electrical connection relationships and distribution characteristics between electrical nodes in the full-element electromagnetic transient model, specifically including the total number of nodes, the electrical distance between nodes, the node connection density, and the overall connectivity structure of the full-element electromagnetic transient model. In the high- and medium-voltage electrical systems of nuclear power plants, nodes are densely packed, the topology is complex, and electrical connections are short-range; the electrical distance between the farthest nodes is usually short. The specific process for determining the segmentation location is as follows: A topology analysis is performed on the full-element electromagnetic transient model to statistically analyze all electrical nodes and their interconnections, generating a complete network topology diagram. The number of subsystems to which the full-element electromagnetic transient model needs to be divided is determined based on simulation computing power requirements. These requirements refer to the total computing resources needed to complete the real-time simulation, determined by the total number of nodes in the full-element electromagnetic transient model, the simulation step size, and the number of processor cores in the real-time simulator. After determining the number of subsystems, candidate partitioning locations are selected from the network topology diagram of the full-element electromagnetic transient model. Each candidate partitioning location corresponds to an electrical connection branch. After cutting the full-element electromagnetic transient model at this branch, it is divided into two parts, each containing several electrical nodes and electrical components. The selection of candidate partitioning locations covers all electrical connection branches in the full-element electromagnetic transient model; no branches are excluded beforehand, and no restrictions are imposed on the impedance characteristics on either side of the branches. After obtaining all candidate segmentation locations, the distribution of the number of nodes in the two subsystems formed after each candidate segmentation location is used to evaluate whether the segmentation location can fully utilize the parallel computing capabilities of the real-time simulator. Specifically, it is assessed whether the number of nodes in the two subsystems matches the computing power allocated to the processor cores. Through comprehensive evaluation, the final segmentation location is determined from the candidate locations, ensuring that after the full-element electromagnetic transient model is segmented at this location, the computational load of each subsystem can be evenly distributed among the different processor cores of the real-time simulator.

[0014] S2. Based on the determined segmentation locations, the full-element electromagnetic transient model is divided into multiple subsystems, and each subsystem is assigned to a different processor core. Each segmentation location corresponds to a subsystem on both sides, and each segmentation location corresponds to an interface. Each interface consists of the boundary nodes of the subsystems on both sides of the corresponding segmentation location. The specific implementation process is as follows: S20. Based on the determined segmentation points, the full-element electromagnetic transient model is divided into multiple subsystems. The segmentation point is an electrical connection branch selected in the full-element electromagnetic transient model. After the full-element electromagnetic transient model is cut off at this branch, it is divided into two independent parts. For the high- and medium-voltage electrical systems of nuclear power plants, the full-element electromagnetic transient model includes all electrical components such as generators, transformers, busbars, cables, and dynamic loads. The electrical connection relationships in the full-element electromagnetic transient model are established based on the actual physical topology. When a segmentation point is determined, a model cutting operation is performed in the simulation software: the electrical connection branch corresponding to the segmentation point is found, and the direct connection relationship between the two ends of the branch is deleted, so that the two nodes that were originally connected at the two ends of the branch are no longer interconnected. After cutting off, the original complete full-element electromagnetic transient model is split into two independent sub-models. Each sub-model retains all its internal electrical components and connection relationships, and each sub-model is a subsystem. If there are multiple determined segmentation points in the full-element electromagnetic transient model, the same cutting operation is performed on each segmentation point in sequence. Each time a cut is performed, the number of blocks in the full-element electromagnetic transient model increases by one. After all cuts are completed, the full-element electromagnetic transient model is divided into multiple subsystems, the number of which equals the number of cuts plus one. All electrical nodes and components within each subsystem maintain their original topology and parameter settings, and the subsystems are completely electrically isolated from each other, with no direct electrical connections.

[0015] S21. Assign each subsystem to a different processor core. A processor core refers to the independent computing unit of the central processing unit in a real-time simulator. Each core can independently execute program instructions and perform floating-point numerical calculations. After completing the full-element electromagnetic transient model segmentation and obtaining multiple subsystems, each subsystem needs to be assigned to a processor core, which will be responsible for all electromagnetic transient calculation tasks of that subsystem during the simulation process. The specific allocation operation is as follows: First, count the number of electrical nodes and electrical components contained in each subsystem. These two indicators directly reflect the computational workload of the subsystem in each simulation step. Then, obtain the total number of available processor cores in the real-time simulator and evaluate its computing power based on performance parameters such as the clock speed and cache capacity of each core. Treat each subsystem as an independent task and allocate each subsystem to a different processor core according to the principle of matching the computational workload with the computing power of the processor core. During allocation, prioritize assigning subsystems with larger computational workloads to processor cores with higher performance, while ensuring that the total computational load of all occupied processor cores is as balanced as possible to avoid a situation where one core is overloaded while other cores are idle. After allocation, each processor core is only responsible for the simulation calculation of the subsystem to which it is assigned. Cores do not share the internal data of the subsystem, and only exchange necessary data at the interface specified in subsequent steps.

[0016] S22. Determine the interface corresponding to each segmentation location. During the segmentation of the full-element electromagnetic transient model, each segmentation location was originally an electrical connection point or branch in the model. After segmentation, the two sides of that location belong to two different subsystems. To enable electrical quantity interaction between these two subsystems in subsequent simulations, each segmentation location needs to be defined as an interface. An interface is a logical data exchange point; its function is to establish a communication channel between the segmented subsystems, allowing the electrical quantities that were originally continuous at that location to be transmitted between the subsystems. For the simulation of the high- and medium-voltage electrical systems of nuclear power plants, the number of interfaces is exactly equal to the number of segmentation locations, with each segmentation location uniquely corresponding to one interface. The establishment of interfaces does not depend on any impedance conditions or electrical distance limitations; any segmentation location can be defined as an interface.

[0017] S23. Each interface consists of boundary nodes of the subsystems on both sides of the corresponding split position. A boundary node is an endpoint node in the subsystem located at the split position. Specifically, at a split position, there is a common electrical node before the split. The split operation disconnects this node, forming two independent nodes, each belonging to one of the two subsystems on either side of the split position. For the subsystem on the left side of the split position, its boundary node is the endpoint of the subsystem connected to the split position after the split; for the subsystem on the right side of the split position, its boundary node is also the endpoint of the subsystem connected to the split position after the split. These two boundary nodes together constitute all the components of the interface corresponding to the split position. During subsequent simulation, the two subsystems on both sides of the interface independently calculate the voltage and current values ​​at their boundary nodes and exchange the electrical quantities at these boundary nodes through the interface, thereby achieving electrical continuity at the position before the split.

[0018] Each subsystem is an independent simulation unit formed by cutting the full-element electromagnetic transient model at the segmentation point. Each subsystem contains a portion of the electrical nodes and components of the original full-element electromagnetic transient model, maintaining complete electrical connections and parameter settings internally. After being assigned to the processor core, each subsystem independently performs electromagnetic transient calculations. The updates of its internal state variables do not depend on the internal calculations of other subsystems; it only exchanges boundary information with adjacent subsystems at the interface.

[0019] The processor core is a physical computing unit in a real-time simulator that possesses complete instruction execution and data processing capabilities. It can independently run simulation programs and perform computational tasks such as floating-point matrix operations and solving differential equations. Each processor core has independent registers, arithmetic logic units, and caches, and can work in parallel with other cores, executing different computational instructions at the same time.

[0020] An interface is a logical-level abstraction of a segmentation location, used to identify the point where electrical quantities need to be exchanged between two subsystems. Each interface corresponds to a segmentation location, and its existence does not change the internal topology of each subsystem; it only serves as a channel for data communication. During simulation, the interface records the boundary node numbers of the two subsystems and stores the voltage and current values ​​exchanged in this instance.

[0021] In this model, boundary nodes are the endpoint nodes of a subsystem at the segmentation location. Before the segmentation operation, they belong to a common node in the original full-element electromagnetic transient model. After the segmentation, they become the endpoints of two independent subsystems. Each boundary node retains electrical parameters such as node voltage and injected current before the segmentation. During simulation calculations within the subsystem, this node participates in solving the node voltage equations of its own subsystem. During data exchange, the voltage and current values ​​on the boundary nodes are transmitted to the adjacent subsystems through the interface.

[0022] S3. After converting the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component, control each processor core to perform simulation operation using the subsystems it receives.

[0023] The process of converting the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component includes: introducing a Clarke-Park coordinate transformation at each interface to convert the time-varying sinusoidal quantity transmitted at the interface into a steady-state DC component. The specific implementation process is as follows: 1) Obtain the three-phase time-varying sinusoidal quantities transmitted at each interface. In the full-element electromagnetic transient model of the high- and medium-voltage electrical system of a nuclear power plant, each interface consists of the boundary nodes of the subsystems on both sides of the segmented location. During the simulation, each processor core independently calculates the voltage and current values ​​of the subsystem it is responsible for. The electrical quantities transmitted at the interface are AC quantities that change sinusoidally with time, i.e., three-phase time-varying sinusoidal quantities. Specifically, for any interface, within each simulation step, the subsystems on both sides of the interface calculate the instantaneous values ​​of the three-phase voltage and three-phase current at their respective boundary nodes. These instantaneous values ​​oscillate sinusoidally with time at the power frequency (50Hz). Before coordinate transformation, these sinusoidal AC quantities are directly transmitted at the interface. To achieve the conversion from time-varying sinusoidal quantities to steady-state DC components, it is first necessary to extract these three-phase instantaneous values ​​from the interface as the input signals for coordinate transformation. The extraction method involves setting signal acquisition points at the interface locations in the simulation model, with each acquisition point acquiring the instantaneous voltage or current values ​​of phases A, B, and C at that location.

[0024] 2) Introduce Clarke coordinate transformation at each interface. Clarke coordinate transformation is a mathematical transformation that converts electrical quantities in a three-phase stationary coordinate system into electrical quantities in a two-phase stationary coordinate system. In the simulation software, a Clarke transformation module is embedded in the signal transmission path of each interface. This module receives the three-phase time-varying sinusoidal quantities collected at the interface as input, and outputs two-phase orthogonal components after transformation. For each interface of the high- and medium-voltage electrical system of a nuclear power plant, an independent Clarke transformation module needs to be configured. The parameters of this module are matched with the electrical frequency at the location of the interface. The input of the Clarke transformation module is the instantaneous values ​​of phases A, B, and C at the interface, and the output is the α-axis component and the β-axis component. These two components still vary sinusoidally with time in the stationary coordinate system, but the three-phase variables have been simplified to two-phase variables.

[0025] 3) The Park coordinate transformation is introduced after the Clarke coordinate transformation. The Park coordinate transformation is a mathematical transformation that converts electrical quantities in a two-phase stationary coordinate system into electrical quantities in a two-phase rotating coordinate system. In the simulation software, a Park transformation module is cascaded at the output of the Clarke transformation module. This module receives the α-axis and β-axis components of the Clarke transformation output as inputs, and simultaneously receives a rotation angle signal synchronized with the system frequency as a reference. This rotation angle signal is obtained by time integration over the system's rated frequency (50Hz), ensuring that the rotating coordinate system of the Park transformation is synchronized with the fundamental positive-sequence component of the power grid. The output of the Park transformation module is the d-axis and q-axis components. During stable operation of the high- and medium-voltage electrical system in a nuclear power plant, the three-phase electrical quantities are symmetrical sinusoidal quantities. The two-phase stationary components obtained after the Clarke transformation are 90 degrees out of phase in space. After the Park transformation, which rotates synchronously with the power grid frequency, the d-axis and q-axis components become constant DC quantities. The Clarke transformation module and the Park transformation module together constitute the Clarke-Park coordinate transformation. When used in series, they completely convert the three-phase time-varying sinusoidal quantity transmitted at the interface into the steady-state DC component.

[0026] 4) The DC component after Clarke-Park coordinate transformation is used as the actual transmitted signal at the interface. After the above coordinate transformation, the original three-phase instantaneous values ​​at each interface, which fluctuate sinusoidally with time, are converted into constant d-axis and q-axis DC components under steady-state conditions. During the simulation, the data exchange between the subsystems on both sides of the interface no longer directly exchanges three-phase sinusoidal AC quantities, but instead exchanges the DC components after Clarke-Park coordinate transformation. The output of each interface provides this DC component value to the other subsystem. After receiving the DC component, the other subsystem can perform inverse coordinate transformation (inverse Park transformation and inverse Clarke transformation) as needed to restore it to the three-phase AC instantaneous value for use in the electrical calculations within its own subsystem. In this way, the time-varying sinusoidal quantity transmitted at each interface is converted into a steady-state DC component, making the signal characteristics at the interface change from time-varying to time-invariant, thereby eliminating the cumulative error caused by the periodic changes in the signal.

[0027] Time-varying sinusoidal quantities refer to physical quantities that change periodically with time according to a sinusoidal function. In the high- and medium-voltage electrical systems of nuclear power plants, the voltage and current transmitted at the interfaces are standard sinusoidal alternating current quantities. Their amplitude and frequency (50Hz) remain constant in steady state, but their instantaneous values ​​change continuously with time, exhibiting a continuous periodic oscillation pattern. This time-varying characteristic means that when directly exchanging discrete data, the values ​​at each moment are different, easily introducing errors caused by time delays.

[0028] The steady-state DC component refers to the DC quantity with a constant amplitude obtained after synchronous rotating coordinate transformation when the system is operating under balanced and symmetrical steady-state conditions. This DC quantity does not change with time and maintains a constant value on the time axis. Under the three-phase balanced steady-state conditions of the high- and medium-voltage electrical systems in nuclear power plants, the d-axis and q-axis components obtained after Clarke-Park coordinate transformation are both constant values, corresponding to the DC representation of the active and reactive components, respectively. This DC characteristic eliminates the interference of sinusoidal quantities fluctuating periodically with time.

[0029] The specific implementation process of S3 is as follows: S30. Load the assigned subsystems onto each processor core. The full-element electromagnetic transient model has previously been divided into multiple subsystems, with each subsystem assigned to a processor core. After completing the coordinate transformation settings at the interface, before starting the simulation, load all model data of the assigned subsystem into the simulation environment of each processor core. The model data includes parameters of electrical components within the subsystem, such as generators, transformers, buses, cables, and dynamic loads, as well as the connection relationships between these components. After loading, the simulation program for each processor core has all the information to independently perform electromagnetic transient calculations for that subsystem.

[0030] S31. Initialize the simulation calculation parameters for each subsystem. At the start of the simulation, set initial state values ​​for each subsystem, including initial voltages at each node, initial currents in each branch, initial angular velocity of the generator rotor, and initial flux linkage of the transformer. The initial state values ​​for each subsystem are obtained from the steady-state power flow calculation results of the full-element electromagnetic transient model. For the two subsystems on both sides of each interface, the voltage and current values ​​at their boundary nodes are kept consistent at the initial moment to ensure the continuity of the initial state before and after the segmentation. After initialization, all processor cores await a unified simulation start command.

[0031] S32. The real-time simulator simultaneously sends a start signal to all occupied processor cores. Each processor core begins its electromagnetic transient calculation of its subsystem at the same time, following a preset simulation step size (e.g., 50 μs). Each processor core independently runs the simulation program corresponding to its subsystem, completing the numerical solution for all node voltages and branch currents within that subsystem within each simulation step. The calculation results of each processor core are only used to update the state variables within its own subsystem and are not directly sent to other processor cores.

[0032] S33. Each processor core calculates the pre-output values ​​of the interface boundary nodes of its subsystem within each simulation step. For a subsystem assigned to a processor core, if the subsystem contains one or more boundary nodes (i.e., the subsystem participates in the construction of one or more interfaces), the processor core extracts the instantaneous three-phase voltage and three-phase current values ​​at these boundary nodes from the calculation results after completing the internal calculation within this step. These instantaneous values ​​are the raw sinusoidal AC quantities without coordinate transformation. After extraction, the processor core sends these raw instantaneous values ​​to the Clarke-Park coordinate transformation module deployed at the interface. After transformation, the steady-state DC components, i.e., the d-axis and q-axis components, are obtained. This transformed DC component is the pre-output value for that side of the interface. For subsystems that do not contain any boundary nodes (i.e., subsystems that do not involve any segmentation positions), it is not necessary to perform the calculation of pre-output values; only internal state updates are required.

[0033] S34. Each processor core exchanges pre-output values ​​through interfaces. Within each simulation step, for each interface, there are two different subsystems and two different processor cores on either side. After each processor core completes its own Clarke-Park coordinate transformation and obtains the DC component of the pre-output value, it transmits its pre-output value to the other through the high-speed data bus or shared memory of the real-time simulator. The exchanged content includes the DC components of voltage and current at the boundary nodes on each side. Specifically, the left processor core sends the DC components of the boundary node voltage and current of its left subsystem to the right processor core, while the right processor core sends the DC components of the boundary node voltage and current of its right subsystem to the left processor core. After the data exchange is completed, each processor core obtains the latest DC component value at its opposite interface.

[0034] S35. Each processor core uses the received DC component from the other side for the calculation of the next simulation step of its subsystem. After receiving the DC component from the other side interface, the processor core can choose to directly use the DC component as an external input, or it can first restore the DC component to a three-phase sinusoidal AC instantaneous value through the Park inverse transform and Clarke inverse transform, and then use it to solve the node voltage equations for the next simulation step of its subsystem, depending on the needs of the simulation program. Regardless of the method used, the signal transmitted at the interface has been stabilized into a DC component and is no longer affected by the time-varying nature of sinusoidal quantities. Each processor core continues to advance the simulation of its subsystem according to a fixed simulation step, repeating the above calculation, pre-output, exchange, and reception process at each step. During the entire simulation operation, the processor cores work in parallel without blocking each other, only engaging in brief synchronous communication during interface data exchange. The simulation continues until the preset simulation end time is reached or a stop command is received.

[0035] The pre-output values ​​refer to the electrical quantities extracted from the calculation results of each processor core after completing the electromagnetic transient calculation of its own subsystem within a simulation step, and which are prepared to be sent to the boundary node electrical quantities of the opposite subsystem via the interface. These values ​​have undergone Clarke-Park coordinate transformation and exist in the form of DC components before being sent. The pre-output values ​​are only valid for the current simulation step and are calculated once per step.

[0036] Synchronous communication refers to a communication method in which multiple processor cores exchange data simultaneously at a predetermined time. Within each simulation step, all processor cores involved in the interface arrive at the data exchange point synchronously after completing their calculations within that step, simultaneously sending and receiving data from the other side, ensuring the timing consistency of data exchange. Synchronous communication avoids data corruption caused by inconsistent sending and receiving timings and is an important means of ensuring the correctness of simulations.

[0037] Optionally, the above technical solution also includes: S4. Set initial constant calibration values ​​and change thresholds for each interface. During islanded simulation, for each interface, before triggering the first synchronization event, set the corresponding initial constant calibration value as the current output. Detect the difference between the pre-output value and the current output in real time. When the difference exceeds the corresponding change threshold, trigger a synchronization event. This synchronization event triggers data synchronization between the two subsystems corresponding to that interface. After data synchronization, update the current output of the interface to the pre-output value after data synchronization. Specifically, when the synchronization event triggers data synchronization between the two subsystems corresponding to the interface, the two subsystems exchange voltage and current values ​​at their respective boundary nodes through the interface. After exchanging the voltage and current values, the two subsystems resume independent operation, waiting for the next synchronization event. The specific implementation process is as follows: S40. Before starting the simulation, an initial constant calibration value needs to be determined for each interface in the full-element electromagnetic transient model. The constant calibration value is a DC value that remains unchanged on the time axis and is used as the representative value for the interface's output when no synchronization event is triggered. The specific method for setting the initial constant calibration value is as follows: First, perform steady-state power flow calculations on the high- and medium-voltage electrical systems of the nuclear power plant to obtain the amplitude and phase of the three-phase voltage and three-phase current at each interface location under steady-state operating conditions. Then, transform the three-phase steady-state sinusoidal quantities at the interface using Clarke-Park coordinates to obtain the corresponding steady-state DC d-axis and q-axis components. Use these two DC components as the initial constant calibration value for that interface. Each interface has its own initial constant calibration value, and the initial constant calibration values ​​for different interfaces may differ, depending on the specific electrical location of the interface in the full-element electromagnetic transient model.

[0038] S41. Set a change threshold for each interface. The change threshold is a pre-set positive value used to determine whether the interface output needs to be updated. For each interface, a change threshold needs to be set along with the initial constant calibration value. The dimension of the change threshold is consistent with the electrical quantity transmitted by the interface. If the interface transmits a voltage signal, the unit of the change threshold is volts; if it transmits a current signal, the unit of the change threshold is amperes. The size of the change threshold is determined based on the simulation accuracy requirements of the high and medium voltage electrical systems of the nuclear power plant and the normal fluctuation amplitude of the signal at the interface. Generally, the change threshold is set between one and five percent of the rated amplitude of the signal at the interface. For example, if the rated voltage amplitude at the interface is 10,000V, the change threshold can be set between 100V and 500V. For different interfaces, different change thresholds can be set according to their voltage level and measurement noise level. The smaller the change threshold, the higher the frequency of synchronous event triggering, and the higher the simulation accuracy, but the greater the communication overhead; the larger the change threshold, the lower the frequency of synchronous event triggering, and the higher the simulation efficiency, but the lower the accuracy. A balance must be struck between accuracy and efficiency when setting the threshold.

[0039] S42. During islanded simulation, before triggering the first synchronization event, the interface outputs its corresponding initial constant calibration value as the current output. Islanded simulation refers to a scenario where each subsystem, after being assigned to different processor cores, independently calculates its internal state variables. Subsystems do not exchange data continuously; each subsystem operates independently, like an island. In the initial stage after simulation startup, for each interface, regardless of the pre-output values ​​calculated by the processor cores on either side of the interface, the output value presented externally by the interface is fixed to the initial constant calibration value. This initial constant calibration value is the current output of that interface. The current output of each interface remains unchanged before the first synchronization event is triggered and is not updated with changes in the internal state of the subsystem.

[0040] S43. Real-time detection of the difference between the pre-output value and the current output of the interface. Within each simulation step, after each processor core completes the electromagnetic transient calculation of its subsystem, it calculates a pre-output value for each interface involved in that subsystem. This pre-output value is the DC component after Clarke-Park coordinate transformation. Meanwhile, the current output of the interface retains its previously set constant calibration value (the initial constant calibration value before the first synchronization event). Within each simulation step, the processor core calculates the absolute difference between the pre-output value and the current output. This difference reflects the deviation between the actual calculated electrical quantity at the interface and the currently maintained output value. The difference calculation is repeated within each simulation step. For each interface, the processor cores on both sides independently calculate the difference between their respective pre-output value and current output.

[0041] S44. For each interface, after calculating the difference within each simulation step, the calculated difference is compared with a pre-set change threshold. If the difference is less than or equal to the change threshold, the deviation between the current output and the pre-output value at the interface is considered to be within an acceptable range, and no interface output update is required. The simulation continues normally without triggering any events. If the difference is greater than the change threshold, the actual calculated electrical quantity at the interface is considered to have deviated too much from the current output, and the interface output needs to be updated. At this time, a synchronization event is triggered. The synchronization event is initiated by the processor core that detects the difference exceeding the limit. After initiation, the core sends a trigger signal to the synchronization management program of the real-time simulator. The synchronization management program then coordinates the two processor cores on both sides of the interface to enter the synchronization process.

[0042] S45. When the synchronization event is triggered, the two subsystems on either side of the interface no longer operate independently but enter the data synchronization phase. Data synchronization refers to the two subsystems exchanging voltage and current values ​​at their respective boundary nodes through the interface. Specifically, the processor core of the subsystem on the left side of the interface sends the instantaneous three-phase voltage and current values ​​(or corresponding DC component values) calculated at its boundary nodes for the current simulation step size to the subsystem on the right side of the interface. Simultaneously, the processor core of the subsystem on the right side of the interface sends the instantaneous three-phase voltage and current values ​​calculated at its boundary nodes for the current simulation step size to the subsystem on the left side of the interface. After the data exchange is completed, each processor core obtains the electrical quantity information at the boundary nodes adjacent to the interface in the other subsystem. This electrical quantity information will be used for updating the boundary conditions of subsequent simulation steps of this subsystem.

[0043] S46. After data exchange is completed, the processor cores on both sides obtain the electrical quantities of the boundary nodes on the other side. At this time, the pre-output value of the interface that triggered the synchronization event has changed (because the boundary conditions on both sides have been updated after synchronization, and subsequent calculations will obtain new pre-output values). The current output of the interface is updated from the original constant calibration value to the pre-output value after this synchronization is completed, i.e., the new DC component value. This new pre-output value will be used as the new current output of the interface in subsequent simulation steps until the next synchronization event is triggered and updated again. After the update is completed, the current output of the interface jumps from the old constant calibration value to the new value, thereby reflecting the actual change of electrical quantities at the interface.

[0044] S47. After exchanging voltage and current values, the two subsystems resume independent operation, awaiting the next synchronization event. Once data synchronization and current output updates are complete, the two subsystems on either side of the interface immediately release synchronization, and each processor core resumes its independent simulation calculations, no longer waiting for each other. Each subsystem continues to update its internal state according to a fixed simulation step size, recalculating the pre-output value at its interface within each step, while simultaneously checking whether the difference between this value and the current output exceeds a change threshold. When the difference exceeds the change threshold again, a synchronization event is triggered again, repeating the data synchronization and output update process. Between two synchronization events, each processor core operates completely independently, and the interface always outputs the current output value after the last update, which remains constant until the next synchronization event occurs. Throughout the simulation, the triggering of synchronization events is discrete and occurs on demand, rather than data exchange at every simulation step.

[0045] The initial constant calibration value is a fixed DC value pre-set for each interface before the simulation starts. This value originates from the DC component of the three-phase sinusoidal quantity at that interface after Clarke-Park coordinate transformation during steady-state operation of the high- and medium-voltage electrical system of the nuclear power plant. Before the first synchronization event is triggered, this value is used as the current output of the interface, and is subsequently updated to a new pre-output value after each synchronization event. Once set, the initial constant calibration value only changes during the simulation when a synchronization event occurs.

[0046] The variation threshold is a positive limit value set individually for each interface. It is used to determine whether the deviation between the pre-output value and the current output of the interface reaches a level that requires triggering a synchronization event. The setting of the variation threshold needs to comprehensively consider the simulation accuracy requirements, signal noise level, and system dynamic characteristics. Its value directly affects the frequency of synchronization event triggering, as well as the efficiency and accuracy of the simulation.

[0047] Islanded simulation operation refers to a mode in which the full-element electromagnetic transient model is divided into multiple subsystems, and each subsystem independently performs electromagnetic transient calculations on its own processor core without continuous, real-time data exchange between them. During islanded operation, each subsystem relies only on its own internal state variables and boundary conditions to advance, and is not directly affected by the states of other subsystems, only briefly exchanging data when synchronization events occur.

[0048] A synchronization event is a discrete data exchange action triggered when the difference between the pre-output value and the current output of an interface exceeds a change threshold. The synchronization event is initiated by the processor core that detects the exceedance and coordinated by the synchronization management program of the real-time simulator. The event execution includes bidirectional exchange of voltage and current values ​​between the subsystems on both sides of the interface, as well as updating the current output of the interface. After the synchronization event ends, each subsystem resumes islanded operation.

[0049] Data synchronization, specifically, is the operation performed after a synchronization event is triggered. It involves the two subsystems corresponding to an interface exchanging voltage and current values ​​at their respective boundary nodes via the interface, enabling both subsystems to obtain the latest boundary electrical quantity information from the other side. After data synchronization is complete, each subsystem updates its own boundary conditions based on the information obtained from the other, thereby ensuring the consistency of electrical quantities on both sides of the segmentation point. Data synchronization occurs discretely, performing only once when a synchronization event is triggered.

[0050] In another possible implementation, the time-varying sinusoidal quantity transmitted at each interface is converted into a steady-state DC component, including: 1) Initialize an adaptive digital filter at each interface. This filter uses the period length corresponding to the system's rated frequency as the integration window to perform sliding period integration on the three-phase time-varying sinusoidal quantities transmitted at the interface. Divide the integral value of each phase over a complete cycle by the cycle length to obtain the initial estimate of the DC component of that phase in steady state. Specifically: An adaptive digital filter is deployed at each interface of the high- and medium-voltage electrical system in the nuclear power plant. The filter is initialized by allocating storage space and setting initial parameters at the start of the simulation. The filter uses the rated frequency of the nuclear power plant system as its reference frequency, which is 50Hz, corresponding to a period length of 20ms. The filter uses a fixed integration window length of 20ms to perform sliding period integration on the three-phase time-varying sinusoidal signals transmitted at the interface. Sliding period integration means that within each simulation step, the filter saves historical sampling data from the past complete cycle length. Each time new sampling data arrives, the oldest data in the window is removed and the latest data is added. Then, all sampled values ​​in the window are accumulated, and the sum is divided by the number of sampling points included in the 20ms period to obtain the initial estimate of the DC component of that phase at the current moment. For the three-phase signals A, B, and C, the filter performs the same three-way sliding period integration operation in parallel to obtain the initial estimates of the DC components of each phase. The parameters set during the initialization of the adaptive digital filter include the integration window length of 20ms, the matching value of the sampling rate and the simulation step size, and the filling method of the data in the integration window at the initial moment. The filling method uses the initial steady-state value of the corresponding phase obtained from the steady-state power flow calculation to fill the entire window.

[0051] An adaptive digital filter is a digital signal processing unit that can automatically adjust its parameters according to changes in the characteristics of the input signal during operation. At the interface of the high and medium voltage electrical system in a nuclear power plant, this filter processes the voltage and current signals transmitted at the interface digitally. Its parameters dynamically change with the real-time fluctuations of the system frequency to adapt to frequency deviations that may occur in the actual operation of the nuclear power plant's electrical system, thereby ensuring accurate extraction of the DC component under different operating conditions.

[0052] Sliding period integration is a time-window integration method for discretizing continuous signals. Within each simulation step, it retains all sampled data from the previous complete system cycle. Whenever new sampled data arrives, the oldest data in the window is removed and the newest data is added, and all sampled values ​​within the window are summed. This integration method slides forward continuously with the simulation progress, outputting an integration result at each step, thus achieving continuous tracking of the DC component of the signal.

[0053] The initial estimate of the DC component is a preliminary value of the steady-state DC component in the three-phase signal obtained after sliding period integration. This value is the average value of the three-phase time-varying sinusoidal quantity at the interface over one complete cycle. When the nuclear power plant system is in steady-state operation, this value is close to the true DC component value. However, due to factors such as the initial transient process of the filter and frequency fluctuations, this value still contains a certain residual error, which requires further processing in subsequent steps.

[0054] 2) The adaptive digital filter dynamically adjusts its filter coefficients based on real-time fluctuations in the system frequency. Using the initial estimate of the DC component calculated from the periodic average as input, it performs secondary attenuation on the residual pulsating components in the input signal, outputting a smoothed three-phase DC component. Specifically: During operation, the adaptive digital filter continuously monitors the actual frequency of the three-phase time-varying sinusoidal quantities at the interface. The actual frequency is monitored by measuring the zero-crossing time interval of one phase voltage signal and taking the reciprocal of the time interval between two adjacent positive zero-crossings as the current frequency value. When the actual frequency deviates from the rated frequency by 50Hz, for example, due to frequency regulation in a nuclear power plant causing a fluctuation of ±0.2Hz, the filter recalculates the filter coefficients based on the detected actual frequency value. The adjustment rule for the filter coefficients is: when the actual frequency deviates from the rated frequency, the filter coefficients increase or decrease proportionally to the offset to ensure consistent attenuation of residual pulsation components at the current frequency. Specifically, the filter employs a first-order low-pass digital filter structure, with its filter coefficients determined by the cutoff frequency, which is set to one-tenth of the current actual frequency. As the actual frequency changes, the cutoff frequency is adjusted synchronously, and the filter coefficients are updated accordingly. After adjusting the filter coefficients, the filter uses the initial estimate of the three-phase DC components as the input signal for this step. This signal undergoes a digital filtering operation with updated filter coefficients to perform secondary attenuation of the residual AC pulsation components in the input signal. After the secondary attenuation is completed, the filter outputs the smoothed DC component values ​​for each of the three phases. The ripple amplitude of these values ​​is significantly reduced compared to the input signal.

[0055] The filter coefficients are a set of parameters in an adaptive digital filter that determine the filter's frequency response characteristics. Their values ​​determine the degree to which the filter attenuates different frequency components. At the interfaces of the high and medium voltage electrical systems in nuclear power plants, the filter coefficients are dynamically adjusted according to the actual system frequency. As the frequency changes, the filter coefficients change accordingly to ensure that the filter maintains the same attenuation effect on residual pulsations after frequency shifts.

[0056] The residual pulsation component refers to the tiny AC fluctuation component that remains in the signal after the three-phase time-varying sinusoidal quantity has been integrated by a sliding period. Under non-ideal steady-state conditions of the electrical system in a nuclear power plant (such as frequency fluctuations, the presence of harmonics, or amplitude imbalance), period integration cannot completely eliminate all AC components. These incompletely eliminated AC fluctuations are the residual pulsation components, which need to be attenuated by subsequent filtering.

[0057] 3) The adaptive digital filter compares the current output three-phase DC component with the historical DC component saved from the previous simulation step. When the difference exceeds a preset range, the filter proportionally corrects and compensates the current output three-phase DC component, using the corrected three-phase DC component as the actual output signal of the interface. Specifically: The adaptive digital filter reads the historical values ​​of the three-phase DC components from the storage unit within each simulation step. The historical DC components of the previous simulation step refer to the final three-phase DC component values ​​output by the filter at the end of the previous simulation step, which are stored in the filter's register. The filter compares the current output three-phase DC components with the read historical DC components phase by phase, calculating the absolute difference between the current value and the historical value for each of phases A, B, and C. During filter initialization, a preset difference range is established. This preset range is determined based on the voltage level of the high- and medium-voltage electrical system in the nuclear power plant, typically using 0.1% of the rated voltage amplitude at the interface as the boundary value of the preset range. The filter compares the calculated three absolute differences with the preset range. If the absolute difference of a phase is less than or equal to the preset range, the current output of that phase is considered to be within a reasonable fluctuation range, and no correction is performed. If the absolute difference of a phase is greater than the preset range, the current output of that phase is considered to have experienced an unexpected jump relative to the previous moment, and the filter initiates a correction compensation operation. The specific method for correction and compensation is as follows: Calculate the deviation between the current output and the historical DC component, multiply this deviation by a preset scaling factor to obtain a compensation value, and subtract this compensation value from the current output value to obtain the corrected DC component. The scaling factor ranges from 0 to 1; in this embodiment, it is set to 0.5, causing the corrected value to move back half the deviation from the historical value. After correction, the corrected three-phase DC component is used as the actual output signal of the interface. This actual output signal simultaneously replaces the historical DC component in the storage unit for comparison in the next simulation step.

[0058] The historical DC component refers to the three-phase DC component values ​​that the adaptive digital filter ultimately outputs and stores in the memory unit at the end of the previous simulation step. This value serves as a reference for the current simulation step, used to determine whether the current output has experienced a sudden change outside the normal range, thereby deciding whether to initiate a correction compensation operation.

[0059] The correction compensation value is a value calculated by the filter to correct the current output when the difference between the current output three-phase DC component and the historical DC component exceeds a preset range. This value is obtained by multiplying the deviation by a preset scaling factor. The corrected DC component is obtained by subtracting this compensation value from the current output value, in order to suppress the error introduced by the sudden change in value.

[0060] 4) The adaptive digital filter simultaneously transmits the corrected three-phase DC components to the subsystems on both sides of the interface. Each subsystem receives these three-phase DC components and uses them as a constant reference quantity for boundary conditions, participating in the electromagnetic transient numerical calculations within its own subsystem. This completes the conversion from time-varying sinusoidal quantities at the interface to steady-state DC components. Specifically: After the adaptive digital filter completes correction and compensation, it obtains the final determined DC component values ​​for each of the three phases: phase A, phase B, and phase C. The filter simultaneously transmits these three DC component values ​​to the subsystems on both sides of the interface via the interface's data transmission channel. The subsystems on both sides of the interface run on different processor cores, each receiving these three DC component values ​​via a high-speed data bus or shared memory. The three-phase DC component values ​​received independently by each subsystem are identical, both originating from the output of the adaptive digital filter at the same interface. Upon receiving the three-phase DC components, each subsystem uses these values ​​as constant reference quantities for its boundary conditions at that interface. Constant reference quantities refer to DC component values ​​that remain unchanged between two synchronization events and do not change with the time step during subsequent electromagnetic transient numerical calculations of the subsystem. In their respective electromagnetic transient numerical calculations, each subsystem substitutes these constant reference quantities into the node voltage equations or branch current equations, along with other internal electrical quantities, to complete the numerical solution for that simulation step. Through the continuous execution of the above four steps, the AC signal at each interface, which originally changed sinusoidally over time, is converted into a constant DC component signal. This DC component signal is transmitted and used in calculations between the subsystems on both sides of the interface in the form of a constant reference quantity, thus completing the full conversion from time-varying sinusoidal quantity to steady-state DC component at the interface.

[0061] The constant reference value refers to the three-phase DC component values ​​received by each subsystem from the interface during islanded simulation, which remain unchanged between two synchronization events. This reference value serves as a fixed input to the subsystem boundary conditions and participates in the internal calculation of each simulation step; its value is only updated after the synchronization event is triggered and data synchronization is completed.

[0062] This invention proposes a constant calibration method, completely eliminating the dependence of simulation stability on the interface impedance ratio, achieving universally stable partitioning at any topological location, and solving the problems of poor stability and low flexibility of traditional methods. It provides a universal, stable, and efficient solution for real-time simulation of full-element partitioning of dense short-range power grids in nuclear power plants. This method changes the data exchange mode from time-axis driven to event-triggered driven, eliminating the dependence of simulation stability on the interface impedance ratio through source suppression, noise filtering, and error control. The technical solution of this invention is further illustrated by the following embodiments, specifically including the following steps: S101. Based on the actual topology of the high- and medium-voltage electrical system of a nuclear power plant, construct a full-element electromagnetic transient model including generators, transformers, busbars, cables, and dynamic loads in graphical modeling and simulation software. The construction process of the full-element electromagnetic transient model, topology feature analysis, and the specific implementation of determining the partitioning positions based on simulation computing power requirements and topology features are consistent with the description in step S1 above, and will not be repeated here. It should be emphasized that when determining the partitioning positions in this step, no restrictions are placed on the impedance ratio of the partitioning positions; any node can be selected as the partitioning point.

[0063] S102. A Clarke-Park coordinate transformation is introduced at each interface of the model segmentation to convert the time-varying sinusoidal quantity transmitted at the interface into a steady-state DC component. The deployment method of the Clarke-Park coordinate transformation, the acquisition of the three-phase time-varying sinusoidal quantity, and the specific process of obtaining the d-axis and q-axis components after transformation are the same as the specific implementation method of converting the time-varying sinusoidal quantity into a DC component in step S3 above. Under ideal steady-state conditions, this DC component remains constant. Within the update interval of the constant calibration value, the initial error generated by the unit delay is zero. Regardless of the value of the impedance ratio G, the cumulative error is always zero, thus theoretically eliminating steady-state error from the source and completely eliminating dependence on the interface impedance ratio. In this embodiment, the simulation step size is 50 microseconds, and G is the Thevenin equivalent impedance ratio on both sides of the interface.

[0064] S103. Set initial constant calibration values ​​and change thresholds for each interface to construct a noise filtering method. The source and setting method of the initial constant calibration values, and the dimensions and value principles of the change thresholds are consistent with the description in step S4 above. High-frequency noise or micro-disturbances with amplitudes lower than the change thresholds at the interface are considered invalid fluctuations, and the interface does not respond or trigger any data synchronization operations, always maintaining the current constant calibration value externally. This method can filter out errors caused by most random noise, avoid noise entering the time-axis driven "generation-amplification" vicious cycle, improve the system's anti-interference capability while maintaining the main characteristics of the signal, and provide a basis for subsequent event-triggered synchronization.

[0065] S104. Establish an event-triggered synchronization method to achieve error control and discrete data synchronization at the interface. The interface outputs a stable constant calibration value. Within each simulation step, the difference between the pre-output value and the current constant calibration value is detected in real time. When the difference exceeds a preset change threshold, a synchronization event is triggered. After the synchronization event is triggered, the two subsystems corresponding to the interface exchange their respective boundary node voltage and current values ​​through the interface to complete a data synchronization. Subsequently, the current output of the interface is updated to the pre-output value after data synchronization. The calculation and exchange method of the pre-output value and the specific content of data synchronization are consistent with the descriptions in steps S3 and S4 above. This method achieves error control on two levels: First, by transforming continuous data exchange into discrete events, the propagation of unit delay error is strictly limited to the initial synchronization range. Without a host computer controlling the synchronization logic, as long as the error is reduced to less than the change threshold, the error can be prevented from continuing to iterate and amplify. Second, during the asynchronous period, the subsystem is in an islanded simulation operation mode. This time provides a dedicated time window for the system to digest and smooth out accumulated errors, further reducing the impact of errors.

[0066] S105. Initiate real-time simulation of the high and medium voltage electrical systems of the nuclear power plant. Each subsystem operates independently, only synchronizing data and updating constant calibration values ​​when a synchronization event is detected. The specific operations of each subsystem on its respective processor core—including loading, initialization, synchronization startup, calculating pre-output values ​​at each step, exchanging DC components via high-speed data bus or shared memory, and using the received DC components from the other side for the next simulation step calculation—are the same as those implemented in step S3 above, controlling the processor core for simulation operation. The simulation supports full-scenario simulations including steady-state operation, load switching, and transient faults. After fault recovery, the system can quickly restore simulation accuracy.

[0067] Example 1: This embodiment takes the high- and medium-voltage electrical system of a typical nuclear power plant as the research object. The electrical distance between the farthest nodes in this system is 1.25 km, which belongs to a typical short-range dense power grid. The simulation step size is set to 50 μs to verify the effectiveness of the constant calibration method proposed in this invention. The specific implementation steps are as follows: Based on the local topology of the high- and medium-voltage electrical system of the nuclear power plant, a full-element electromagnetic transient model including generators, transformers, buses, cables, and dynamic loads was constructed in a graphical modeling and simulation software. Two representative segmentation locations were selected: ① The first segmentation location: close to the power source side, with an impedance ratio G≈1.42, which is greater than 1. After segmentation, the subsystem contains 48 electrical nodes; ② The second segmentation location: close to the load group, with an impedance ratio G≈0.92, which is less than 1. After segmentation, the subsystem contains 8 electrical nodes.

[0068] A constant calibration method is deployed at the interface between the two segmentation locations. Specifically, Clarke-Park coordinate transformation is introduced to achieve source suppression; initial constant calibration values ​​and change thresholds are set to construct a noise filtering method; an event-triggered synchronization method is established, and difference judgment logic and synchronization update rules are set. For example... Figure 2 As shown, the host computer (the executing entity of this application, and there may be multiple host computers) is responsible for sending synchronization instructions to each subsystem (i.e., Figure 2 The "control signal" in the middle) to trigger a synchronization event (i.e. Figure 2 The text in the middle is "activate", t k-1 t k t k+1 The synchronization event trigger time point is defined as the time interval between two adjacent trigger time points during which each subsystem operates in an isolated simulation state. This time interval is called the breathing period. During this period, each subsystem calculates independently, and this breathing period is used to gradually reduce the time delay error introduced by the data synchronization in the previous synchronization event to below a preset change threshold, thereby eliminating the delay and completing the digestion and mitigation of the error. Specifically, at t k-1 At a specific moment, the host computer simultaneously sends synchronization commands to all subsystems. Upon receiving the synchronization command, each subsystem executes a synchronization event, exchanging voltage and current values ​​at its boundary nodes through its corresponding interface to complete one data synchronization. After data synchronization is complete, each subsystem immediately resumes islanded simulation operation and enters phase t. k-1 To t k During the adjustment period between these intervals, each subsystem independently performs its internal electromagnetic transient numerical calculations, while the external interface maintains its current constant calibration value. Simultaneously, the system utilizes this time to gradually reduce the error introduced by unit delay during data synchronization to below the change threshold. When the simulation progresses to t... k At time t, during the islanded simulation, each subsystem detects that the difference between the pre-output value of a certain interface and the current constant calibration value exceeds a preset change threshold, thus triggering the next synchronization event. The synchronization management program issues a synchronization command again, and each subsystem performs data synchronization again. After completion, it enters t again. k To t k+1 The interval between breaths; this cycle repeats, and the synchronous event occurs at t k-1 t k t k+1The synchronization events are triggered on demand at discrete time points, and each subsystem operates in an islanded simulation state between synchronization events. The delay error is eliminated by utilizing the breath adjustment time, transforming the synchronization events from time-axis driven to event-triggered driven, effectively avoiding the communication overhead caused by data synchronization at each simulation step. The deployment method of Clarke-Park coordinate transformation, the setting method of constant calibration values ​​and change thresholds, and the specific logic of difference detection and synchronization triggering in the event-triggered synchronization method are consistent with the descriptions in steps S3 and S4 above, and will not be repeated here.

[0069] Two test scenarios were designed to compare the simulation performance of the constant calibration method with the traditional long-distance transmission line segmentation method and the ideal transformer model method: ① Scenario A is the steady-state and operation scenario: the system starts normally and the load is switched on and off, and the simulation stability and accuracy are tested; ② Scenario B is the transient fault scenario: at the second segmentation position, a two-phase AC short-circuit ground fault is set at 0.1 seconds and recovered at 0.2 seconds, and a three-phase short-circuit ground fault is set at 0.3 seconds and recovered at 0.4 seconds, and the simulation stability and self-recovery capability under transient conditions are tested.

[0070] In scenario A, the long-distance transmission line segmentation method fails completely due to insufficient electrical distance; the transformer model method for handling the first segmentation position diverges rapidly due to an impedance ratio greater than 1, while the constant calibration method remains stable. Figure 3 The comparison of the ideal transformer model method, the constant calibration method, and the phase A voltage of the original circuit under high impedance ratio shows that the waveform of the constant calibration method remains consistent with the original circuit, with a root mean square error of 86.2318V and an average relative error of 0.58%. The second segmentation position treatment shows that both the ideal transformer model method and the constant calibration method are stable. Figure 4 The comparison of the ideal transformer model method, the constant calibration method, and the phase A voltage of the original circuit under low impedance ratio conditions shows that both methods can track the waveform of the original circuit. The root mean square error of the constant calibration method is 72.5792V, and the average relative error is 0.42%. In scenario B, the second segmentation position shows that the signal of the ideal transformer model method diverges during a three-phase short-circuit fault, while the constant calibration method only shows small fluctuations, remaining stable around the true value. Moreover, it quickly recovers its accuracy after the fault is restored, maintaining high-fidelity simulation.

[0071] Figure 5The top subplot shows the A-phase voltage waveform of the original circuit under low impedance ratio fault conditions, serving as a reference to demonstrate the true A-phase voltage change curve of the original circuit without any model segmentation or interface processing throughout the fault occurrence, duration, and recovery process. The bottom left subplot shows the A-phase voltage waveform obtained by the ideal transformer model method under low impedance ratio fault conditions, demonstrating the A-phase voltage change curve obtained after interface processing using the ideal transformer model method at the segmentation location. The bottom right subplot shows the A-phase voltage waveform obtained by the constant calibration method under low impedance ratio fault conditions, demonstrating the A-phase voltage change curve obtained after interface processing using the constant calibration method at the segmentation location. Comparing the three subplots, it can be seen that the ideal transformer model method exhibits significant oscillation and divergence during the three-phase short-circuit fault, deviating from the true value of the original circuit, while the constant calibration method remains stable throughout the entire fault occurrence, duration, and recovery process, with the waveform always fluctuating around the true value of the original circuit without divergence.

[0072] Figure 6 The top subplot shows the A-phase current waveform of the original circuit under low impedance ratio fault conditions, serving as a reference to demonstrate the true A-phase current change curve of the original circuit without any model segmentation or interface processing throughout the fault occurrence, duration, and recovery process. The bottom left subplot shows the A-phase current waveform obtained by the ideal transformer model method under low impedance ratio fault conditions, demonstrating the A-phase current change curve obtained after interface processing using the ideal transformer model method at the segmentation location. The bottom right subplot shows the A-phase current waveform obtained by the constant calibration method under low impedance ratio fault conditions, demonstrating the A-phase current change curve obtained after interface processing using the constant calibration method at the segmentation location. Comparing the three subplots, it can be seen that the current waveform of the ideal transformer model method exhibits severe oscillation and divergence during short-circuit faults, deviating from the true current value of the original circuit. In contrast, the constant calibration method can stably track the current waveform of the original circuit throughout the entire fault occurrence, duration, and recovery process, with only small fluctuations at the moment of fault occurrence and recovery. After the fluctuations, it quickly returns to the true value of the original circuit, demonstrating good transient tracking and self-recovery capabilities.

[0073] Example 2: This embodiment analyzes 168 potential model partitioning locations for the high- and medium-voltage electrical systems of a nuclear power plant, comparing the partitioning flexibility of the constant calibration method and the ideal transformer model method. The ideal transformer model method can only fully utilize computing power when the impedance ratio is less than 1 and the number of electrical nodes in the partitioned subsystem is greater than 30, resulting in only 28 partitioning locations that meet the conditions, demonstrating extremely low flexibility. The constant calibration method, on the other hand, has no limitations on the impedance ratio or the number of electrical nodes in the subsystem, and all 168 potential partitioning locations can be stably partitioned, significantly improving partitioning flexibility. Partitioning locations can be freely selected according to simulation computing power requirements, fully leveraging the parallel computing capabilities of the real-time simulator.

[0074] This invention is applicable to parallel real-time simulation scenarios of high and medium voltage electrical systems in nuclear power plants with dense nodes, complex topologies, and short-range electrical connections. The technical effects are as follows: 1) This method systematically eliminates the constraint of simulation stability on the interface impedance ratio through the three methods of source suppression, noise filtering and error control in the constant calibration method, so that the high and medium voltage electrical system of nuclear power plants can be safely partitioned at any topology location, and solves the problem that the partitioning location of the ideal transformer model method is limited by impedance conditions.

[0075] 2) This method has no cumulative error under steady-state operation, maintains numerical stability when transient faults such as two-phase short-circuit grounding or three-phase short-circuit grounding occur, and can quickly self-recover simulation accuracy after the fault is cleared. It solves the problems of easy divergence of the ideal transformer model method under transient conditions and complete failure of the long-distance transmission line segmentation method in short-distance networks.

[0076] 3) This method breaks through the limitation of impedance ratio convergence condition and can be applied to all potential partition positions of the high and medium voltage electrical system of nuclear power plants, while the ideal transformer model method is only effective for a few partition positions. The former can freely select the partition position according to the simulation computing power requirements and give full play to the parallel computing capabilities of each processor core of the real-time simulator.

[0077] 4) When the method is used for simulation at any segmentation position, the root mean square error is controlled within 86.2V and the average relative error is less than 0.58%. It achieves high-fidelity electromagnetic transient simulation while ensuring the flexibility of segmentation and the stability of simulation, which meets the accuracy requirements of nuclear power plant engineering applications.

[0078] 5) This constant calibration method is an algorithmic innovation of the ideal transformer model method. It does not change the original topological characteristics of the electrical system of the nuclear power plant. It can be deployed and implemented based on existing graphical modeling and simulation software without additional hardware modifications, and its engineering application is easy.

[0079] In the above embodiments, although the steps are numbered S1, S2, etc., they are only specific embodiments given by the present invention. Those skilled in the art can adjust the execution order of S1, S2, etc. according to the actual situation. The scheme after adjusting the order is also within the protection scope of the present invention. It can be understood that in some embodiments, some or all of the above embodiments may be included.

[0080] like Figure 7 As shown, an embodiment of the present invention provides a real-time simulation system for the full analysis of electrical elements in a power plant, comprising a construction determination module, a partitioning and allocation module, and a simulation module; The construction determination module is used to: construct a full-element electromagnetic transient model of the power plant's electrical system, and determine the segmentation location based on the topological characteristics of the full-element electromagnetic transient model; The segmentation and allocation module is used to: divide the full-element electromagnetic transient model into multiple subsystems based on the determined segmentation positions, and allocate each subsystem to different processor cores. Each segmentation position corresponds to a subsystem on both sides, and each segmentation position corresponds to an interface. Each interface is composed of the boundary nodes of the subsystems on both sides of the corresponding segmentation position. The simulation module is used to convert the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component, and then control each processor core to perform simulation operation using the subsystems it receives.

[0081] Optionally, in the above technical solution, the simulation module is also used to: set an initial constant calibration value and a change threshold for each interface; during the island simulation operation, for each interface, before triggering the first synchronization event, make the interface output the corresponding initial constant calibration value as the current output; detect the difference between the pre-output value and the current output of the interface in real time; when the difference is greater than the corresponding change threshold, trigger a synchronization event; the synchronization event is used to trigger the two subsystems corresponding to the interface to perform a data synchronization; after the data synchronization is completed, update the current output of the interface to the pre-output value after data synchronization.

[0082] By setting initial constant calibration values ​​and change thresholds, the data exchange at the interface is changed from a continuous mode to a discrete synchronization mode triggered on demand. The synchronization event is only triggered when the difference between the pre-output value and the current output exceeds the change threshold, avoiding the communication overhead caused by data exchange at every simulation step. After the synchronization event is triggered, the two subsystems exchange the voltage and current values ​​of the boundary nodes through the interface. After synchronization is completed, the current output of the interface is updated to the pre-output value after data synchronization, so that the interface output can follow the actual changes in the system state. While ensuring simulation accuracy, the frequency of data interaction between processor cores is reduced, and the overall efficiency of parallel simulation is improved.

[0083] Optionally, in the above technical solution, when the two subsystems corresponding to the synchronization event trigger interface perform a data synchronization, the two subsystems exchange voltage and current values ​​at their respective boundary nodes through the interface. After completing the exchange of voltage and current values, the two subsystems resume independent operation and wait for the next synchronization event to be triggered. This ensures that the electrical quantity information on both sides of the segmentation position is accurately transmitted, guaranteeing the correct establishment of electrical continuity between the segmented subsystems. After completing the exchange of voltage and current values, the two subsystems immediately resume independent operation and wait for the next synchronization event, so that each processor core does not need to wait for each other between two synchronization events, avoiding long-term data exchange blocking, maintaining the high efficiency of the islanded simulation operation of each subsystem, and improving the overall efficiency of parallel simulation.

[0084] Optionally, in the above technical solution, the simulation module is further specifically used to: introduce Clarke-Park coordinate transformation at each interface, converting the time-varying sinusoidal quantity transmitted at the interface into a steady-state DC component through Clarke-Park coordinate transformation. This transforms the AC signal at the interface, which originally varied periodically with time, into a constant DC quantity, eliminating the delay error introduced by the time-varying characteristics of the sinusoidal quantity in discrete data exchange; under steady-state conditions, the DC component transmitted at the interface remains constant, avoiding the accumulation and amplification of steady-state errors on both sides of the interface, so that the simulation stability no longer depends on the impedance conditions on both sides of the interface.

[0085] It should be noted that the beneficial effects of the real-time simulation system for full-element electrical analysis of power plants provided in the above embodiments are the same as those of the real-time simulation method for full-element electrical analysis of power plants described above, and will not be repeated here. Furthermore, the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, and will not be repeated here.

[0086] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned real-time simulation methods for analyzing all electrical elements of a power plant.

[0087] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the above-mentioned real-time simulation methods for analyzing all electrical elements of a power plant.

[0088] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0089] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A real-time simulation method for analyzing all electrical elements of a power plant, characterized in that, include: Construct a full-element electromagnetic transient model of the power plant's electrical system, and determine the segmentation location based on the topological characteristics of the full-element electromagnetic transient model; Based on the determined segmentation positions, the full-element electromagnetic transient model is divided into multiple subsystems, and each subsystem is assigned to a different processor core. Each segmentation position corresponds to a subsystem on both sides, and each segmentation position corresponds to an interface. Each interface is composed of the boundary nodes of the subsystems on both sides of the corresponding segmentation position. After converting the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component, the processor cores are controlled to perform simulation operation using their respective received subsystems.

2. The real-time simulation method for analyzing all electrical elements of a power plant according to claim 1, characterized in that, Also includes: An initial constant calibration value and a change threshold are set for each interface. During the island simulation, for each interface, before triggering the first synchronization event, the interface outputs the corresponding initial constant calibration value as the current output. The difference between the pre-output value and the current output of the interface is detected in real time. When the difference is greater than the corresponding change threshold, a synchronization event is triggered. The synchronization event is used to trigger the two subsystems corresponding to the interface to perform a data synchronization. After the data synchronization is completed, the current output of the interface is updated to the pre-output value after data synchronization.

3. The real-time simulation method for analyzing all electrical elements of a power plant according to claim 2, characterized in that, When the two subsystems corresponding to the synchronization event triggering interface perform a data synchronization, the two subsystems exchange the voltage and current values ​​at their respective boundary nodes through the interface. After the exchange of the voltage and current values ​​is completed, the two subsystems resume independent operation and wait for the next synchronization event to be triggered.

4. A real-time simulation method for analyzing all electrical elements of a power plant according to any one of claims 1 to 3, characterized in that, Converting the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component includes: introducing a Clarke-Park coordinate transformation at each interface, and converting the time-varying sinusoidal quantity transmitted at the interface into a steady-state DC component through the Clarke-Park coordinate transformation.

5. A real-time simulation system for analyzing all electrical elements of a power plant, characterized in that, This includes a construction determination module, a partitioning and allocation module, and a simulation module; The construction determination module is used to: construct a full-element electromagnetic transient model of the power plant's electrical system, and determine the segmentation position based on the topological characteristics of the full-element electromagnetic transient model; The segmentation and allocation module is used to: segment the full-element electromagnetic transient model into multiple subsystems based on the determined segmentation positions, and allocate each subsystem to a different processor core. Each segmentation position corresponds to a subsystem on both sides, and each segmentation position corresponds to an interface. Each interface is composed of the boundary nodes of the subsystems on both sides of the corresponding segmentation position. The simulation module is used to: convert the time-varying sinusoidal quantity transmitted at each interface into a steady-state DC component, and then control each processor core to perform simulation operation using its respective received subsystem.

6. A real-time simulation system for full-element electrical analysis of a power plant according to claim 5, characterized in that, The simulation module is also used to: set an initial constant calibration value and a change threshold for each interface; during the island simulation operation, for each interface, before triggering the first synchronization event, make the interface output the corresponding initial constant calibration value as the current output; detect the difference between the pre-output value and the current output of the interface in real time; when the difference is greater than the corresponding change threshold, trigger a synchronization event; the synchronization event is used to trigger the two subsystems corresponding to the interface to perform a data synchronization; after the data synchronization is completed, update the current output of the interface to the pre-output value after data synchronization.

7. A real-time simulation system for full-element electrical analysis of a power plant according to claim 6, characterized in that, When the two subsystems corresponding to the synchronization event triggering interface perform a data synchronization, the two subsystems exchange the voltage and current values ​​at their respective boundary nodes through the interface. After the exchange of the voltage and current values ​​is completed, the two subsystems resume independent operation and wait for the next synchronization event to be triggered.

8. A real-time simulation system for full-element electrical analysis of a power plant according to any one of claims 5 to 7, characterized in that, The simulation module is also specifically used to: introduce Clarke-Park coordinate transformation at each interface, and convert the time-varying sinusoidal quantity transmitted at the interface into a steady-state DC component through the Clarke-Park coordinate transformation.

9. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the real-time simulation method for analyzing all electrical elements of a power plant as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements a real-time simulation method for analyzing all electrical elements of a power plant as described in any one of claims 1 to 4.

Images