A real-time simulation test modeling method and system for a new energy station

Abstract

The present application relates to the field of new energy power system simulation technology, in particular to a real-time simulation test modeling method and system for a new energy station, comprising: receiving steady-state parameters and dynamic instructions, constructing a multi-time scale device model library and determining a disturbance scene. The detailed electromagnetic transient model and the simplified electromechanical transient model are mapped through interface variables, a bidirectional data interaction channel is established, and mixed scale closed loop simulation is carried out on a real-time simulation platform. The output data of the two types of models are synchronously collected, the device level transient response characteristics and the station level dynamic response characteristics are respectively extracted, and a dynamic response panoramic atlas is generated by fusion. The atlas is matched with the similarity of the historical typical response mode, and a model confidence evaluation report is output. The method realizes real-time bidirectional interaction between models of different time scales, and provides automatic and panoramic intelligent evaluation capability of simulation results.

Description

Technical Field

[0001] This invention relates to the field of simulation technology for new energy power systems, and in particular to a real-time simulation test modeling method and system for new energy power plants. Background Technology

[0002] In simulation modeling of new energy power plants, the rapid electromagnetic transient processes within equipment and the slow electromechanical dynamic processes of the entire plant typically require analysis using models with different time scales. Electromagnetic transient simulation models are detailed but computationally expensive, making them difficult to scale to the entire power plant; electromechanical transient simulation models are suitable for large-scale analysis but cannot accurately describe the rapid switching dynamics of power electronic equipment. Existing technical solutions often use the two types of models separately, or achieve loosely coupled co-simulation by setting fixed boundary conditions for unidirectional, periodic data interaction. In such methods, the slow dynamic model cannot respond in real time to the instantaneous changes of the fast dynamic model, and the fast dynamic model cannot immediately perceive the state evolution at the system level. This makes it difficult for the simulation process to realistically reproduce the real-time interaction and chain reactions of cross-scale dynamics in the actual system.

[0003] The current evaluation of simulation results relies on professionals analyzing detailed waveforms at the equipment level and trend curves at the site level separately, followed by manual comprehensive judgment. This evaluation method is inefficient and lacks the technical means to systematically correlate and integrate microscopic transient characteristics with macroscopic dynamic performance, making it difficult to objectively and quantitatively evaluate the overall behavior confidence of the simulation model under complex disturbance sequences. Therefore, a modeling method is needed that can achieve tight coupling of multi-timescale models, bidirectional real-time interaction, and automatic panoramic fusion and intelligent evaluation of simulation results. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a real-time simulation test modeling method and system for new energy power plants.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a real-time simulation test modeling method for new energy power stations, comprising:

[0006] Receive steady-state operating parameters from the new energy power plant production monitoring system and dynamic adjustment command streams from the process control system;

[0007] Based on the steady-state operating parameters, a multi-timescale device model library is constructed, which includes detailed electromagnetic transient models and simplified electromechanical transient models. The disturbance scenario sequence for simulation experiments is determined according to the dynamic adjustment command flow.

[0008] Interface variables are mapped between the detailed electromagnetic transient models and the simplified electromechanical transient models in the equipment model library to establish a two-way data interaction channel from the equipment level to the station level.

[0009] Deploy the device model library on the real-time simulation platform, inject the disturbance scene sequence, and start a closed-loop simulation with mixed time scales;

[0010] During the closed-loop simulation, the internal electrical waveforms of the detailed electromagnetic transient model and the global operating state of the simplified electromechanical transient model are collected simultaneously.

[0011] Feature extraction is performed on the internal electrical quantity waveforms to obtain a device-level transient response feature set, and sequence analysis is performed on the global operating state quantities to obtain a station-level dynamic response feature set.

[0012] By integrating the equipment-level transient response feature set and the station-level dynamic response feature set, a dynamic response panoramic map reflecting the overall picture of the new energy power station is generated.

[0013] The dynamic response panorama is matched with the pre-stored historical typical response patterns for similarity, and a model confidence assessment report for the current simulation experiment is output.

[0014] As a further aspect of the present invention, the construction of a multi-timescale device model library based on the steady-state operating parameters, including detailed electromagnetic transient models and simplified electromechanical transient models, specifically includes:

[0015] The steady-state operating parameters include the rated power of each power generation unit, the grid connection point voltage, and the current total active power output;

[0016] The model nameplate data of each power generation unit in the steady-state operating parameters is analyzed. The model nameplate data includes the number of generator pole pairs, rated speed, stator and rotor parameters, and converter switching frequency.

[0017] Based on the model nameplate data, the physical prototype equations of the synchronous generator or power electronic converter are constructed using the equivalent circuit modeling method. The physical prototype equations include differential terms describing changes in magnetic field energy storage and logical judgment terms reflecting the switching behavior of semiconductor devices.

[0018] The physical prototype equations are discretized into a set of difference equations suitable for calculation at the microsecond level, forming the core computational kernel of the detailed electromagnetic transient model.

[0019] Meanwhile, based on the current total active power output and grid connection point voltage in the steady-state operating parameters, a set of transfer functions reflecting the overall external characteristics of the station is constructed by combining algebraic equations with first-order inertial elements.

[0020] The transfer function set is reduced in order, and the low-order approximate expression corresponding to the dominant pole is retained to form the mathematical description of the simplified electromechanical transient model.

[0021] A high-speed sampling output port is set in the detailed electromagnetic transient model to expose the instantaneous voltage and current values ​​of key internal nodes;

[0022] A low-speed aggregation input port is set in the simplified electromechanical transient model to receive power commands and voltage reference values ​​from the station control system;

[0023] An equivalent transformation relationship is established between the high-speed sampling output port and the low-speed aggregation input port. This equivalent transformation relationship is achieved through Fourier decomposition and fundamental component extraction.

[0024] The computational kernel of the detailed electromagnetic transient model, the mathematical description of the simplified electromechanical transient model, and the equivalent transformation relationship are packaged and encapsulated to form an executable code module of the multi-timescale device model library.

[0025] As a further aspect of the present invention, determining the disturbance scenario sequence of the simulation experiment based on the dynamic adjustment command stream specifically includes:

[0026] The dynamic adjustment command stream includes power setpoint adjustment commands, reactive power compensation device switching commands, and frequency response action signals.

[0027] The disturbance scenario sequence includes grid voltage drop events, frequency step change events, and planned power ramp-up events;

[0028] The dynamic adjustment command stream is timestamped and parsed to restore the original control command timing issued by the scheduling master station;

[0029] Identify the periodic adjustment patterns and sudden adjustment events in the timing of the original control commands. The periodic adjustment patterns include power tracking commands for automatic generation control, and the sudden adjustment events include rapid power support requirements for primary frequency regulation.

[0030] According to the technical specifications for grid connection of new energy power plants, a set of standard voltage disturbance waveform templates are defined, including symmetrical voltage drop waveforms, asymmetrical voltage drop waveforms, and phase jump waveforms.

[0031] Based on actual power grid operation statistics, a set of typical frequency change rate curves are defined, including linear decreasing curves, exponential recovery curves, and oscillation decay curves.

[0032] The periodic adjustment mode is combined with the standard voltage disturbance waveform template to generate a planned voltage adaptability test scenario.

[0033] The sudden adjustment event is combined with the typical frequency change rate curve to generate an unplanned frequency stability test scenario.

[0034] Random noise components are inserted into the voltage adaptability test scenario and the frequency stability test scenario. These random noise components are used to simulate measurement errors and background harmonic interference.

[0035] The voltage adaptability test scenarios and frequency stability test scenarios are sorted according to the test objectives to form a sequence of disturbance scenarios with clear test objectives. Each scenario is labeled with the expected disturbance type, intensity parameter and duration.

[0036] As a further aspect of the present invention, the step of mapping interface variables between the detailed electromagnetic transient model and the simplified electromechanical transient model in the equipment model library to establish a bidirectional data interaction channel from the equipment level to the site level specifically involves:

[0037] In the detailed electromagnetic transient model, a set of interface variables that need to be uploaded are defined. The interface variables that need to be uploaded include the fundamental amplitude of the AC side current of the converter, the average value of the DC bus voltage, and the estimated value of the junction temperature of the power device.

[0038] In the simplified electromechanical transient model, a set of interface variables that need to be issued are defined. The interface variables that need to be issued include the voltage phase angle at the grid connection point of the power station, the estimated value of the system equivalent inertia, and the set value of the dispatch active power.

[0039] Design a shared memory data area, which is divided into a high-speed access area and a low-speed access area;

[0040] The interface variables that need to be uploaded from the detailed electromagnetic transient model are low-pass filtered and then written into the low-speed access area of ​​the shared memory data area.

[0041] The interface variables that need to be issued in the simplified electromechanical transient model are directly written into the high-speed access area of ​​the shared memory data area;

[0042] A data reading thread is configured for the detailed electromagnetic transient model. The data reading thread reads the estimated values ​​of the grid connection point voltage phase angle and the system equivalent inertia from the high-speed access area at millisecond intervals, which serve as the basis for updating its boundary conditions.

[0043] A data reading thread is configured for the simplified electromechanical transient model. The data reading thread reads the fundamental amplitude of the AC side current and the average value of the DC bus voltage of the converter from the low-speed access zone at a time interval of seconds, as the basis for its equivalent parameter correction.

[0044] A hardware interrupt mechanism ensures mutual exclusion of two data reading threads, thus avoiding read-write conflicts in the shared memory data area.

[0045] The bidirectional data interaction channel is established through continuous data exchange in the shared memory data area and periodic synchronization of the data reading thread.

[0046] As a further aspect of the present invention, the step of deploying the device model library on a real-time simulation platform and injecting the disturbance scene sequence to initiate a closed-loop simulation with mixed time scales specifically involves:

[0047] The executable code module of the device model library is loaded into the multi-core processor of the real-time simulation platform, the multi-core processor containing a core dedicated to microsecond-level calculation and a core dedicated to millisecond-level calculation;

[0048] The computational kernel of the detailed electromagnetic transient model is assigned to the core dedicated to microsecond-level computation, and a fixed-step interrupt service routine is configured.

[0049] The mathematical description of the simplified electromechanical transient model is assigned to the core dedicated to millisecond-level calculations, and a task scheduler with variable step size is configured.

[0050] Initialize the shared memory data area and assign initial values ​​to all interface variables based on the steady-state operating parameters;

[0051] The description information of the first test scenario is read from the disturbance scenario sequence, and the type and intensity parameters of the disturbance to be applied are parsed out.

[0052] According to the type of disturbance, the corresponding disturbance generation subroutine is invoked, and the disturbance generation subroutine generates a voltage or frequency excitation signal that meets the standard requirements;

[0053] The voltage or frequency excitation signal is superimposed on the grid connection point boundary condition of the simplified electromechanical transient model, and the fault initiation flag of the detailed electromagnetic transient model is triggered simultaneously.

[0054] The computational tasks of the core dedicated to microsecond-level calculations and the core dedicated to millisecond-level calculations are initiated, so that the detailed electromagnetic transient model and the simplified electromechanical transient model begin to synchronously advance the simulation time.

[0055] The two-way data interaction channel enables real-time data exchange between the two models, forming a hybrid time-scale closed-loop simulation that includes the rapid electromagnetic processes of the equipment and the slow electromechanical processes of the station.

[0056] As a further aspect of the present invention, during the closed-loop simulation operation, the internal electrical quantity waveforms of the detailed electromagnetic transient model and the global operating state quantities of the simplified electromechanical transient model are simultaneously acquired, specifically as follows:

[0057] Multiple non-destructive probes are inserted into the calculation kernel of the detailed electromagnetic transient model. These non-destructive probes record the pulse state of the power electronic switching device, the instantaneous value of the filter inductor current, and the ripple component of the DC capacitor voltage.

[0058] The data recorded by the non-destructive probe is saved at a sampling rate no less than the model calculation step size to form a high-fidelity original record of the device-level electrical waveform.

[0059] Key state variables are marked in the mathematical description of the simplified electromechanical transient model. These key state variables include the smoothed value of the active power output of the power station, the cumulative value of the reactive power, and the integral of the frequency deviation.

[0060] The numerical changes of the key state variables are recorded at a rate synchronized with the model calculation step size to form a time series of station-level state variables at equal intervals.

[0061] A synchronous clock source is configured in the real-time simulation platform. The synchronous clock source provides a unified time reference for the acquisition of the high-fidelity original records of equipment-level electrical waveforms and the recording of the time series of station-level state quantities at equal intervals.

[0062] Add a timestamp and a model identifier to the high-fidelity equipment-level electrical waveform raw record. The model identifier is used to distinguish which specific power generation unit model the data comes from.

[0063] Add timestamps and scenario identifiers to the equally spaced station-level state quantity time series, wherein the scenario identifiers are used to associate the current data with the disturbance test scenario to which it belongs;

[0064] The high-fidelity original records of equipment-level electrical waveforms with added identification information and the time series of station-level state variables at equal intervals are stored in the distributed buffer of the real-time simulation platform.

[0065] As a further aspect of the present invention, the step of extracting features from the internal electrical quantity waveforms to obtain a device-level transient response feature set, and simultaneously performing sequence analysis on the global operating state quantities to obtain a station-level dynamic response feature set, specifically involves:

[0066] Extract current and voltage data within a specific time window before and after the disturbance from the high-fidelity original record of the device-level electrical waveform;

[0067] Perform a sliding window fast Fourier transform on the current and voltage data within the specific time window to calculate the fundamental amplitude, phase, and harmonic distortion rate of each window;

[0068] The maximum overshoot, settling time, and steady-state error values ​​in the fundamental amplitude variation curve are extracted as characteristic indicators to characterize the transient process of equipment voltage and current.

[0069] Identify the switching frequency change patterns and latch-up events in the pulse state records of the power electronic switching device, and count the number of pulse loss and the duration of abnormal conduction.

[0070] The characteristic indicators representing the transient process of device voltage and current, the switching frequency change mode, and the latching event are packaged together to form the device-level transient response feature set describing the response characteristics of a single device;

[0071] The active power sequence, reactive power sequence, and grid connection point voltage sequence are separated from the equally spaced station-level state quantity time series.

[0072] The root mean square value of the tracking error of the active power sequence relative to the dispatch command is calculated as a measure of the power control accuracy of the power station.

[0073] Calculate the integral area of ​​the reactive power sequence during voltage disturbances as a measure of the reactive power support capability of the power station;

[0074] The recovery rate of the grid connection point voltage sequence after fault clearance is calculated as a measure of the station voltage stability.

[0075] The measurement of power control accuracy, reactive power support capability, and voltage stability of the power station are combined to form the power station-level dynamic response feature set that describes the overall behavior of the power station.

[0076] As a further aspect of the present invention, the step of fusing the equipment-level transient response feature set and the station-level dynamic response feature set to generate a dynamic response panoramic map reflecting the overall picture of the new energy power station specifically includes:

[0077] A unique spatial location code is assigned to the equipment-level transient response feature set of each power generation unit, and the spatial location code corresponds to the actual position of the power generation unit in the electrical wiring diagram of the new energy power station;

[0078] The site-level dynamic response feature set is treated as a global layer and placed as the background layer of the dynamic response panoramic map.

[0079] According to the spatial location encoding, the device-level transient response feature set of each power generation unit is placed in the corresponding position of the global layer in the form of icons;

[0080] Within each icon, the fill color represents the maximum overshoot magnitude among the characteristic indicators representing the transient process of the device's voltage and current, and the icon border thickness represents the frequency of abnormal events in the switching frequency variation mode.

[0081] On the global layer, the change trajectories of the active power sequence, reactive power sequence, and grid connection point voltage sequence in the field-level dynamic response feature set are plotted in the form of curves.

[0082] Add markers at the key inflection points of the changing trajectory. The markers display the simulation time and specific feature values ​​corresponding to the key inflection points.

[0083] Establish an interactive link between the icon and the change trajectory. When an icon is selected, highlight the impact traces caused by the action time point of the device unit on the field-level sequence curve.

[0084] The global layer, the set of icons with visual encoding, the change trajectory curve, and the interactive links are integrated into a unified graphical view framework, which supports zooming of the timeline and switching of layer display.

[0085] The final rendered output is the dynamic response panoramic map.

[0086] As a further aspect of the present invention, the step of performing similarity matching between the dynamic response panoramic map and pre-stored historical typical response patterns to output a model confidence evaluation report for the current simulation experiment specifically involves:

[0087] Retrieve historical typical response patterns from the simulation case database. The historical typical response patterns include compliant response patterns that have been verified by previous simulations and real fault response patterns recorded in field tests.

[0088] Extract the feature vector of the historical typical response pattern. The feature vector is composed of historical equipment-level feature indicators and historical site-level measurement parameters combined according to the same rules.

[0089] The equipment-level transient response feature set and the station-level dynamic response feature set contained in the dynamic response panoramic map are transformed into the feature vector of the current test according to the same rules.

[0090] Calculate the Euclidean distance between the feature vector of the current experiment and the feature vector of each historical typical response pattern;

[0091] Set a similarity threshold, and filter out historical typical response patterns whose Euclidean distance is less than the similarity threshold as a candidate set of similar patterns;

[0092] Analyze the category to which each pattern belongs in the candidate set of similar patterns, and count the number of times each compliance response pattern and real fault response pattern appear;

[0093] If the frequency of the compliant response pattern is dominant, the response behavior of the current simulation model is determined to be in line with expectations, and a higher confidence score is assigned.

[0094] If the frequency of occurrence of the actual fault response mode is dominant, it is determined that the response behavior of the current simulation model is close to the actual fault waveform, but it is necessary to verify whether it is the expected fault ride-through behavior.

[0095] Based on the combined confidence scores and pattern category analysis results, a structured model confidence assessment report is generated. The report includes a list of similarity matching results, descriptions of major deviation features, and the overall confidence level.

[0096] As a further aspect of the present invention, the present invention also includes a real-time simulation test modeling system for new energy power stations. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the real-time simulation test modeling method for new energy power stations as described above.

[0097] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0098] By establishing an interface variable mapping and bidirectional data interaction channel between the detailed electromagnetic transient model and the simplified electromechanical transient model, nanosecond to millisecond-level electromagnetic transient processes at the device level and second-level electromechanical dynamic processes at the site level can exchange data and influence each other within the real-time simulation step. The rapid response of the device model can instantly affect the power flow and state of the site network, while the state changes of the site model can also be fed back to the controller and circuit interface of the device model in real time. This tight coupling realizes a full-chain, closed-loop simulation from local devices to the entire site, from fast transients to slow dynamics, enabling the simulation system to more realistically simulate the dynamic interaction and energy transfer processes across time scales in the actual system.

[0099] The waveforms of internal electrical quantities of equipment and the global operating status of the site, collected during simulation, are processed in parallel to extract transient response features and dynamic response features, respectively. These two heterogeneous features are then fused into a unified panoramic map. This map integrates microscopic equipment behavior and macroscopic system performance. This panoramic map is then automatically matched with a pre-stored library of historical typical patterns to directly generate a quantitative evaluation report on the credibility of the overall model behavior in the current simulation experiment. This process avoids the subjectivity and inefficiency of manual piecemeal evaluation, enabling intelligent and comprehensive assessment of the consistency and rationality of the "equipment-system" joint response of complex hybrid simulation models when dealing with various disturbances. Attached Figure Description

[0100] Figure 1 This is a flowchart of the real-time simulation test modeling method for new energy power stations described in this invention;

[0101] Figure 2Flowchart for the construction of a detailed electromagnetic transient and simplified electromechanical transient model library;

[0102] Figure 3 A flowchart for generating voltage adaptability and frequency stability disturbance scenario sequences;

[0103] Figure 4 The active power dynamic response tracking characteristic curve;

[0104] Figure 5 This is the dynamic response time-series curve for new energy power plants. Detailed Implementation

[0105] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0106] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0107] See Figure 1The system receives steady-state operating parameters from the production monitoring system of the new energy power plant and dynamic adjustment command streams from the process control system. Based on the received steady-state operating parameters, a multi-timescale equipment model library is constructed, including detailed electromagnetic transient models and simplified electromechanical transient models. Simultaneously, the disturbance scenario sequence for the simulation test is determined according to the dynamic adjustment command stream. Interface variable mapping is performed between the detailed electromagnetic transient models and the simplified electromechanical transient models in the equipment model library, thereby establishing a bidirectional data interaction channel from the equipment level to the power plant level. The equipment model library is deployed on the real-time simulation platform, and the determined disturbance scenario sequence is injected to initiate a mixed-timescale closed-loop simulation. During the closed-loop simulation, the internal electrical waveforms of the detailed electromagnetic transient model and the global operating state quantities of the simplified electromechanical transient model are simultaneously acquired. Feature extraction is performed on the acquired internal electrical waveforms to obtain the equipment-level transient response feature set, and sequence analysis is performed on the acquired global operating state quantities to obtain the power plant-level dynamic response feature set. The equipment-level transient response feature set and the power plant-level dynamic response feature set are fused to generate a dynamic response panoramic map reflecting the overall situation of the new energy power plant. Finally, the dynamic response panorama is matched with the pre-stored historical typical response patterns to output a model confidence assessment report for the current simulation experiment.

[0108] In one embodiment of the present invention, see [reference] Figure 2 The steady-state operating parameters provided by the new energy power plant production monitoring system include the rated power, grid connection voltage, and current total active power output of each power generation unit within the plant. For example, for a power plant containing fifty doubly-fed induction generator (DFIG) wind turbines, the steady-state operating parameters record that the rated power of each unit is 2 MW, the grid connection voltage is 35 kV, and the current total active power output is 80 MW. The system analyzes the model nameplate data of each power generation unit within these steady-state operating parameters. The model nameplate data specifically includes the number of generator pole pairs, rated speed, stator and rotor parameters, and converter switching frequency. For example, the analyzed DFIG generator has 2 pole pairs, a rated speed of 1800 rpm, a stator resistance of 0.01 ohms, a rotor resistance of 0.015 ohms, and a converter switching frequency of 2 kHz.

[0109] In some embodiments, based on model nameplate data, the physical prototype equations of a synchronous generator or power electronic converter are constructed using the equivalent circuit modeling method. These physical prototype equations include differential terms describing changes in magnetic field energy storage and logical decision terms reflecting the switching behavior of semiconductor devices. For the doubly-fed induction generator rotor-side converter in the example, its physical prototype equations include differential equations describing changes in AC-side inductor current and logical decision terms determining the on / off state of the insulated-gate bipolar transistor based on the pulse width modulation signal. The physical prototype equations are discretized into a set of difference equations suitable for microsecond-level step-size calculations, forming the core computational kernel of the detailed electromagnetic transient model. For example, the forward Euler method is used to discretize the inductor current differential equation, resulting in a difference equation for updating the current value at a 10-microsecond simulation step-size.

[0110] Understandably, based on the current total active power output and grid connection voltage in the steady-state operating parameters, a set of transfer functions reflecting the overall external characteristics of the power station is constructed by combining algebraic equations with a first-order inertial element. For example, using the current 80 MW total active power output as the initial condition, a simplified transfer function is constructed with dispatch commands as input and the total active power output of the power station as output. The transfer function set is then reduced in order, retaining the low-order approximate expressions corresponding to the dominant poles, forming a simplified mathematical description of the electromechanical transient model. For example, a transfer function that was originally third-order is simplified to a first-order inertial element with a time constant of 2 seconds after the order reduction.

[0111] Optionally, a high-speed sampling output port is set in the detailed electromagnetic transient model to expose the instantaneous voltage and current values ​​of key internal nodes. The high-speed sampling output port outputs the instantaneous values ​​of the AC phase current and DC bus voltage of the converter at microsecond intervals. A low-speed aggregation input port is set in the simplified electromechanical transient model to receive power commands and voltage reference values ​​from the station control system. The low-speed aggregation input port receives new active power setpoint commands at hundreds of millisecond intervals. An equivalent transformation relationship is established between the high-speed sampling output port and the low-speed aggregation input port. This equivalent transformation relationship is achieved through Fourier decomposition and fundamental component extraction. For example, the instantaneous current value obtained from the high-speed sampling output port is subjected to a Fourier analysis over one power frequency cycle, and its fundamental effective value is extracted as the equivalent current for parameter correction in the simplified electromechanical transient model. In some embodiments, the computational kernel of the detailed electromagnetic transient model, the mathematical description of the simplified electromechanical transient model, and the equivalent transformation relationship are packaged and encapsulated to form an executable code module of a multi-timescale device model library. The computational kernel of the detailed electromagnetic transient model is implemented in the form of C language functions, the mathematical description of the simplified electromechanical transient model is described by a set of parameterized state-space equations, and the equivalent transformation relationship is encapsulated as an independent signal processing function. The three are compiled together into a dynamic link library file that can be called in a real-time simulator.

[0112] In one embodiment of the present invention, see [reference] Figure 3 The dynamic adjustment command stream issued by the process control system includes power setpoint adjustment commands, reactive power compensation device switching commands, and frequency response action signals. In an example scenario, the dynamic adjustment command stream records a command to adjust the power setpoint from 80 MW to 85 MW, a command to switch on a group of capacitor banks, and a rapid power boost support signal triggered when the frequency is below 49.8 Hz. The disturbance scenario sequence to be constructed for the simulation experiment includes grid voltage drop events, frequency step change events, and planned power ramp-up events. For example, a preset disturbance scenario sequence consists of a symmetrical voltage drop event lasting 500 ms, an event where the frequency steps from 50 Hz to 49.5 Hz, and a planned power ramp-up event at a rate of 10 MW per minute.

[0113] In some embodiments, the dynamic adjustment command stream is timestamped and parsed to reconstruct the original control command timing sequence issued by the scheduling master station. The original control command timing sequence records the specific time when the power setpoint adjustment command is issued (T1) and the specific time when the frequency response action signal is triggered (T2) with millisecond-level precision. Periodic adjustment patterns and sudden adjustment events in the original control command timing sequence are identified. Periodic adjustment patterns include power tracking commands for automatic generation control, and sudden adjustment events include rapid power support requirements for primary frequency regulation. In the example timing sequence, a power tracking command pattern that adjusts every 4 seconds starting from T1 is identified, as well as a sudden event that appears at time T2, requiring 5 MW of power support within 2 seconds.

[0114] Understandably, according to the technical specifications for grid connection of new energy power plants, a set of standard voltage disturbance waveform templates are defined. These templates include symmetrical voltage drop waveforms, asymmetrical voltage drop waveforms, and phase jump waveforms. For example, a defined symmetrical voltage drop waveform template describes a voltage amplitude dropping to 35% of the rated value within 1 millisecond, maintaining this low voltage level for 500 milliseconds, and then recovering to the rated value within 60 milliseconds. Based on actual grid operation statistics, a set of typical frequency change rate curves are defined, including linear decline curves, exponential recovery curves, and oscillation decay curves. A defined linear decline curve describes the frequency decreasing from 50 Hz to 49.5 Hz at a constant rate of 0.5 Hz per second.

[0115] Optionally, a planned voltage adaptability test scenario can be generated by combining a periodic regulation mode with a standard voltage disturbance waveform template; a scenario can be generated by combining a periodic power point tracking command from automatic generation control with a 20% voltage symmetrical drop waveform, where a test site responds to voltage drops during power regulation. An unplanned frequency stability test scenario can be generated by combining a sudden regulation event with a typical frequency change rate curve; and a scenario can be generated by combining a sudden power support demand from primary frequency regulation with a linear frequency drop curve, where a test site provides power support during rapid frequency drops.

[0116] In some embodiments, random noise components are inserted into the voltage adaptability test scenario and the frequency stability test scenario. The random noise components are used to simulate measurement errors and background harmonic interference. A Gaussian white noise sequence with a mean of zero and a standard deviation of 0.5% of the rated voltage is superimposed onto the voltage amplitude signal of the voltage drop waveform. The voltage adaptability test scenario and the frequency stability test scenario are sorted according to the test purpose to form a disturbance scenario sequence with a clear test objective. Each scenario is labeled with the expected disturbance type, intensity parameter, and duration. The first scenario in the final disturbance scenario sequence is labeled as "Type: Symmetrical voltage drop, Intensity: 35%, Duration: 500 milliseconds", and the second scenario is labeled as "Type: Frequency step drop, Intensity: 0.5 Hz, Duration: Continuous".

[0117] In practical implementation, a set of interface variables to be uploaded is defined in the detailed electromagnetic transient model. These variables include the fundamental amplitude of the AC current on the converter side, the average DC bus voltage, and the estimated junction temperature of the power devices. For a detailed electromagnetic transient model of a doubly-fed induction generator (DFIG) wind turbine, the specific interface variables to be uploaded are the effective value of the fundamental AC current of the rotor-side converter obtained through real-time calculation, the 1-second moving average of the DC link capacitor voltage, and the estimated junction temperature of the insulated gate bipolar transistor (IGBT). A set of interface variables to be sent is defined in the simplified electromechanical transient model. These variables include the phase angle of the voltage at the grid connection point, the estimated equivalent inertia of the system, and the dispatch active power setpoint. The specific interface variables to be sent from the simplified electromechanical transient model are the 35 kV bus voltage phase angle from the grid equivalent model, the equivalent inertia constant estimated based on the regional grid status, and the 85 MW active power setpoint from the dispatch command.

[0118] Design a shared memory data area, divided into a high-speed access area and a low-speed access area. The total capacity of the shared memory data area is 1 megabyte, with the first 512 kilobytes defined as the high-speed access area allowing microsecond-level read / write operations, and the latter 512 kilobytes defined as the low-speed access area allowing millisecond-level read / write operations. Interface variables from the detailed electromagnetic transient model that need to be uploaded are low-pass filtered before being written to the low-speed access area of ​​the shared memory data area. For example, the original update value of the fundamental amplitude of the AC current of the converter is smoothed using a first-order low-pass filter with a cutoff frequency of 10 Hz, and then the processed result is written to a designated address in the low-speed access area at a frequency of 100 times per second. Interface variables from the simplified electromechanical transient model that need to be sent are directly written to the high-speed access area of ​​the shared memory data area. The latest grid-connected voltage phase angle value calculated by the simplified electromechanical transient model is directly updated to a predetermined position in the high-speed access area at the beginning of each 5-millisecond simulation step.

[0119] A data read thread is configured for the detailed electromagnetic transient model. This thread reads the grid-connected voltage phase angle and the estimated equivalent inertia of the system from the high-speed access zone at millisecond intervals, serving as the basis for updating its boundary conditions. The data read thread is triggered at a fixed 2-millisecond interval to read the latest voltage phase angle data from the high-speed access zone, used to update the grid voltage source phase in the detailed electromagnetic transient model. A data read thread is configured for the simplified electromechanical transient model. This thread reads the fundamental amplitude of the converter AC side current and the average DC bus voltage from the low-speed access zone at second intervals, serving as the basis for correcting its equivalent parameters. The data read thread is triggered at a fixed 1-second interval to read the average fundamental amplitude of the current uploaded by all generating units in the past second from the low-speed access zone, used to calculate and correct the equivalent current source amplitude in the simplified electromechanical transient model. A hardware interrupt mechanism ensures mutual exclusion of access between the two data reading threads, preventing read-write conflicts in the shared memory data area. Different hardware interrupt priorities are set for write and read operations in the high-speed access area. When the simplified electromechanical transient model is writing data to the high-speed access area, a high-level interrupt is triggered to temporarily suspend the data reading thread of the detailed electromagnetic transient model. A bidirectional data interaction channel is established through continuous data exchange in the shared memory data area and periodic synchronization of the data reading threads.

[0120] In one embodiment of the present invention, the executable code module of the device model library is loaded into the multi-core processor of the real-time simulation platform. The multi-core processor includes cores dedicated to microsecond-level calculations and cores dedicated to millisecond-level calculations. For example, a real-time simulation platform is equipped with an eight-core processor, where physical cores 0 and 1 are configured as cores dedicated to microsecond-level calculations, physical cores 2 and 3 are configured as cores dedicated to millisecond-level calculations, and the remaining cores are used for system scheduling. The calculation kernel of the detailed electromagnetic transient model is assigned to the core dedicated to microsecond-level calculations and configured with a fixed-step interrupt service routine. The calculation kernel of the detailed electromagnetic transient model is bound to physical core 0 as an independent real-time task, and the interrupt service routine is set to trigger once every 10 microseconds to ensure that the calculation kernel advances the simulation calculation in a fixed step of 10 microseconds.

[0121] In some embodiments, the mathematical description of the simplified electromechanical transient model is assigned to a core dedicated to millisecond-level calculations, and a variable-step task scheduler is configured. The mathematical description of the simplified electromechanical transient model is compiled into an executable thread and deployed on physical core 2. The task scheduler allows the calculation step size of this thread to be dynamically adjusted between 1 millisecond and 10 milliseconds according to the model's numerical stability requirements. The shared memory data area is initialized, and all interface variables are assigned initial values ​​based on steady-state operating parameters. The grid connection point voltage of 35 kV and the current total active power output of 80 MW are read from the steady-state operating parameters. The "grid connection point voltage phase angle" variable in the shared memory data area is initialized to 0 radians, and the "total active power output of the power station" variable is initialized to 80 MW.

[0122] In some embodiments, real-time data exchange between the two models is achieved through a bidirectional data interaction channel, forming a hybrid time-scale closed-loop simulation that includes the fast electromagnetic processes of equipment and the slow electromechanical processes of the station. During the simulation, the detailed electromagnetic transient model is calculated every 10 microseconds and reads the latest grid-connected point voltage phase angle through the high-speed access area of ​​the shared memory data area as the boundary condition for its network solution. The simplified electromechanical transient model is calculated every 5 milliseconds and reads the current fundamental amplitude value uploaded by the detailed electromagnetic transient model through the low-speed access area of ​​the shared memory data area to update its equivalent admittance parameters. This continuous data exchange enables the detailed electromagnetic transient model to perceive the station-level grid dynamics described by the simplified electromechanical transient model, while the simplified electromechanical transient model can also respond to the equipment-level dynamic characteristics aggregated by the detailed electromagnetic transient model, thus forming a hybrid time-scale closed-loop simulation loop. The formula describing the fixed-step calculation relationship of the detailed electromagnetic transient model is as follows:

[0123]

[0124] in: This indicates the simulation time point calculated in step n of the detailed electromagnetic transient model. This indicates a fixed calculation step size. This indicates the next simulation time point.

[0125] In one embodiment of the present invention, multiple non-destructive probes are inserted into the computational kernel of the detailed electromagnetic transient model. These probes record the pulse states of power electronic switching devices, the instantaneous values ​​of the filter inductor current, and the ripple components of the DC capacitor voltage. For example, in the detailed electromagnetic transient model of a photovoltaic inverter, non-destructive probes are inserted onto the gate drive signal line of an insulated-gate bipolar transistor, the current sampling point of the AC-side filter inductor, and the voltage sampling point of the DC-side supporting capacitor. The data recorded by the non-destructive probes is saved at a sampling rate no less than the model's computational step size, forming a high-fidelity original record of the device-level electrical waveforms. The computational step size of the detailed electromagnetic transient model is 10 microseconds, and the non-destructive probes record data at the same 10-microsecond cycle. Each record includes a timestamp, pulse state, instantaneous inductor current value, and instantaneous capacitor voltage value, forming a high-fidelity original record of the device-level electrical waveforms arranged in a time sequence.

[0126] A synchronous clock source is configured in the real-time simulation platform. This source provides a unified time reference for the acquisition of high-fidelity raw records of equipment-level electrical waveforms and the recording of equally spaced station-level state quantity time series. The synchronous clock source uses a high-precision IEEE 1588 protocol clock, ensuring that the timestamps of all data records are based on the same globally coordinated time source. This allows for precise alignment of microsecond-level data from the detailed electromagnetic transient model with millisecond-level data from the simplified electromechanical transient model on the time axis. Timestamps and model identifiers are added to the high-fidelity raw records of equipment-level electrical waveforms. The model identifier distinguishes which specific generator unit model the data originates from. Each record containing instantaneous current and voltage values ​​is appended with an IEEE 1588 timestamp accurate to microseconds and a string-type model identifier. Timestamps and scenario identifiers are added to the equally spaced station-level state quantity time series. The scenario identifier associates the current data with the disturbance test scenario. Each 5-millisecond recorded state quantity data point is appended with a timestamp accurate to milliseconds and an integer scenario identifier; for example, the identifier for the scenario "35% voltage symmetric drop" is 101. The high-fidelity original records of equipment-level electrical waveforms with added identification information and the time series of equally spaced station-level state variables are stored in the distributed buffer of the real-time simulation platform, respectively. The high-fidelity original records of equipment-level electrical waveforms are written to the non-volatile memory buffer of core 0, while the time series of equally spaced station-level state variables are written to the non-volatile memory buffer of core 2.

[0127] It is understandable that current and voltage data within a specific time window before and after a disturbance are extracted from the high-fidelity original recording of the equipment-level electrical waveforms. A specific time window is set with the voltage drop initiation time as the zero point, extended by 100 milliseconds before and after it. All instantaneous inductor current values ​​within the time range of -100 milliseconds to +100 milliseconds are extracted from the high-fidelity original recording of the equipment-level electrical waveforms in "PV_Inverter_Unit_23". A sliding window Fast Fourier Transform (FSFT) is performed on the current and voltage data within the specific time window to calculate the fundamental amplitude, phase, and harmonic distortion rate for each window. A sliding window with a width of one power frequency cycle and a step size of 10 microseconds is used to perform continuous FFTs on the extracted current data to calculate the effective value of the fundamental current for each window. Fundamental phase angle and total harmonic distortion The maximum overshoot, settling time, and steady-state error values ​​are extracted from the fundamental amplitude variation curve as characteristic indicators of the transient voltage and current processes of the equipment; the effective value of the fundamental current is also extracted. In the time-varying curve, the maximum overshoot during the current drop process was calculated to be 105%, the settling time was 65 ms, and the steady-state error was -2%. The switching frequency change pattern and latch-up event in the pulse state record of the power electronic switching device were identified, and the number of pulse loss and abnormal conduction duration were counted. The pulse state record of the gate drive signal of the insulated gate bipolar transistor was analyzed, and two pulse loss events and one abnormal conduction event with a duration of 150 μs were identified during the voltage drop. The characteristic indicators characterizing the transient process of voltage and current of the device, the switching frequency change pattern and the latch-up event were packaged to form a device-level transient response feature set describing the response characteristics of a single device. The feature data of a device were packaged into a structure containing the fields: maximum overshoot (105%), settling time (65 ms), steady-state error (-2%), number of pulse loss (2), and abnormal conduction duration (150 μs).

[0128] Active power, reactive power, and grid connection point voltage sequences are separated from equally spaced station-level state variable time series. The numerical sequence of variable P_out_smooth is extracted from the station-level state variable time series as the active power sequence, the numerical sequence of variable Q_sum is extracted as the reactive power sequence, and the amplitude sequence of grid connection point voltage is extracted as the grid connection point voltage sequence. The root mean square (RMS) value of the tracking error of the active power sequence relative to the dispatch command is calculated as a measure of the station power control accuracy. The difference between the active power sequence and the dispatch command sequence (e.g., a command stepping from 80 MW to 85 MW) constitutes the error sequence, and the RMS value of this error sequence within a 200-millisecond time window after the disturbance is calculated. The calculation formula is:

[0129]

[0130] Where: N is the total number of sampling points within a 200-millisecond time window. This represents the measured active power value at the k-th sampling point (from the active power sequence). This is the active power reference value for the k-th sampling point (from the scheduling instruction sequence). See Table 1.

[0131] Table 1: Equipment-level Electrical Waveform Original Record Table

[0132]

[0133] The area under the reactive power sequence during voltage disturbances is calculated as a measure of the reactive power support capacity of the power station. Within a 500-millisecond time window of voltage dip duration, the reactive power sequence is numerically integrated, and the area enclosed by its curve and the zero line is calculated. The recovery rate of the grid-connected voltage sequence after fault clearance is calculated as a measure of the station's voltage stability. The change in the grid-connected voltage sequence is calculated over a 50-millisecond time interval following the voltage recovery start point. Change over time ratio ,Right now The measurement of power control accuracy, reactive power support capability, and voltage stability is combined to form a power station-level dynamic response feature set describing the overall behavior of the power station. This feature set includes three numerical features: the root mean square value of active power tracking error. Reactive power support integral area and voltage recovery rate .

[0134] See Figure 4This simulation presents the dynamic response process of a power plant's total active power increasing from 80MW to 85MW in response to a 35% voltage symmetric dip disturbance scenario. Specifically, the blue curve represents the dispatch reference value, signifying the ideal power command target; the orange curve represents the measured active power value at the plant, reflecting the actual dynamic response; and the orange shaded area represents the tracking error, i.e., the instantaneous deviation between the measured value and the dispatch reference value. In the initial simulation phase (0-50ms), affected by the voltage dip disturbance, the measured active power fluctuates drastically, rapidly dropping from an initial overshoot peak of 88MW to around 80MW. This process corresponds to the power decoupling and rapid adjustment characteristics of the power electronic converter under voltage transients. In the 50-200ms range, the measured value gradually converges through closed-loop control, and the tracking error continuously decreases from a maximum of approximately 8MW. After 200ms, the measured active power stabilizes around the 85MW dispatch command value with slight oscillations. At this point, the tracking error is at a low level, indicating that the plant's power control loop possesses good dynamic recovery capability and steady-state tracking accuracy after the disturbance. At the parameter level, the step amplitude of the dispatch command in this simulation was 5MW, the voltage drop depth was 35%, and the simulation time scale covered 0-500ms, fully capturing the power change characteristics of the three stages of disturbance transient, dynamic recovery and steady-state tracking.

[0135] In one embodiment of the present invention, a unique spatial location code is assigned to the device-level transient response feature set of each power generation unit. The spatial location code corresponds to the actual location of the power generation unit in the electrical wiring diagram of the new energy power station. For example, a photovoltaic inverter located in electrical zone A, feeder 7, port 3 of the power station is assigned the spatial location code "A-7-3". The power station-level dynamic response feature set is placed as a global layer in the background layer of the dynamic response panoramic map. The power station-level dynamic response feature set includes three metric parameters: root mean square value of active power tracking error, reactive power support integral area, and voltage recovery rate. These parameters are rendered as semi-transparent numerical label areas in the background layer.

[0136] In some embodiments, according to spatial location coding, the device-level transient response feature sets of each power generation unit are placed in the form of icons at the corresponding positions in the global layer. The icon corresponding to the spatial location code "A-7-3" is placed at the (120, 85) pixel point in the graphical view coordinate system. Inside each icon, the fill color is used to represent the maximum overshoot magnitude among the feature indicators characterizing the device voltage and current transient process, and the icon border thickness is used to represent the frequency of abnormal events in the switching frequency change mode. For example, the icon with a maximum overshoot of 105% is filled with dark red, the icon with a maximum overshoot of 98% is filled with light green, the icon with an abnormal event frequency of 2 times has a border thickness of 3 pixels, and the icon with an abnormal event frequency of 0 times has a border thickness of 1 pixel.

[0137] An interactive link is established between icons and change trajectories. When an icon is selected, the impact trace of the device unit's action time point on the site-level sequence curve is highlighted. For example, when a user clicks the icon corresponding to the spatial location code "A-7-3", the graphical view will highlight the corresponding segment on the active power sequence curve where a slight dip appears in the total active power of the site when the inverter experiences a pulse loss event at 1.101 seconds. The global layer, the set of visually coded icons, the change trajectory curves, and the interactive links are integrated into a unified graphical view framework. The graphical view framework supports zooming of the time axis and switching of layer displays. Users can use the slider to zoom the time axis from a panoramic 2-second range to display details of 0.5 seconds during the voltage drop period, and can also select to hide the reactive power sequence curve layer using checkboxes.

[0138] Historical typical response patterns were retrieved from the simulation case database. These patterns included compliant response patterns verified by previous simulations and actual fault response patterns recorded in field measurements. The database search criteria were set to "Disturbance type: symmetrical voltage drop, intensity range: 30%-40%", returning 15 historical typical response pattern records. Feature vectors of the historical typical response patterns were extracted. These feature vectors were composed of historical equipment-level characteristic indicators and historical site-level measurement parameters combined according to the same rules. The feature vector of a historical compliant response pattern was represented as [maximum overshoot: 102%, settling time: 70ms, steady-state error: -1.5%, pulse loss: 0, abnormal conduction: 0μs, RMSE_P: 0.05MW, A_Q: 15Mvar·s, K_U: 0.5pu / s].

[0139] The equipment-level transient response feature sets and the site-level dynamic response feature sets contained in the dynamic response panorama are transformed into feature vectors for the current test according to the same rules. The feature vectors for the current test are represented as [maximum overshoot: 105%, settling time: 65ms, steady-state error: -2%, pulse loss: 2, abnormal conduction: 150μs, RMSE_P: 0.08MW, A_Q: 12Mvar·s, K_U: 0.6pu / s]. The Euclidean distance between the feature vectors of the current test and the feature vectors of each historical typical response mode is calculated, as is the Euclidean distance between the feature vectors of the current test and the feature vector of the j-th historical mode. Calculated using the following formula:

[0140]

[0141] in: This represents the Euclidean distance between the current experimental feature vector and the j-th historical typical response pattern feature vector, where m represents the total dimension of the feature vector (m=8 in this example). This represents the i-th eigenvalue of the current experimental eigenvector. Let represent the i-th eigenvalue of the eigenvector of the j-th historical typical response pattern.

[0142] See Figure 5 This demonstrates the timing characteristics of the dynamic response at the renewable energy power plant level. Specifically, within the voltage dip range (approximately 0.5s to 1.0s) marked in pink, the total active power (blue curve) fluctuates slightly before rapidly recovering to a steady-state level of approximately 10MW, reflecting the plant's active power's ability to withstand disturbances. The total reactive power (orange curve) rises rapidly to approximately 5Mvar during the disturbance and then gradually declines, demonstrating the plant's reactive power support response during voltage dips. The grid connection point voltage (red curve) simultaneously drops to approximately 0.6pu (corresponding to a 35% drop), and recovers to its rated value of approximately 1.0pu at a relatively fast rate after the disturbance is cleared, verifying the plant's voltage recovery capability. The overall curves indicate that under symmetrical voltage dip disturbances, this renewable energy power plant can achieve rapid recovery of the grid connection point voltage through stable control of active power and dynamic support of reactive power, meeting the requirements of grid connection technical specifications for fault ride-through.

[0143] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A real-time simulation test modeling method for new energy power stations, characterized in that, Includes the following steps: It receives steady-state operating parameters from the new energy power plant production monitoring system and dynamic adjustment command streams from the process control system; Based on the steady-state operating parameters, a multi-timescale device model library is constructed, which includes detailed electromagnetic transient models and simplified electromechanical transient models. The disturbance scenario sequence for simulation experiments is determined according to the dynamic adjustment command flow. Interface variables are mapped between the detailed electromagnetic transient models and the simplified electromechanical transient models in the equipment model library to establish a two-way data interaction channel from the equipment level to the station level. Deploy the device model library on the real-time simulation platform, inject the disturbance scene sequence, and start a closed-loop simulation with mixed time scales; During the closed-loop simulation, the internal electrical waveforms of the detailed electromagnetic transient model and the global operating state of the simplified electromechanical transient model are collected simultaneously. Feature extraction is performed on the internal electrical quantity waveforms to obtain a device-level transient response feature set, and sequence analysis is performed on the global operating state quantities to obtain a station-level dynamic response feature set. By integrating the equipment-level transient response feature set and the station-level dynamic response feature set, a dynamic response panoramic map reflecting the overall picture of the new energy power station is generated. The dynamic response panorama is matched with the pre-stored historical typical response patterns for similarity, and a model confidence assessment report for the current simulation experiment is output. The step of mapping interface variables between the detailed electromagnetic transient models and simplified electromechanical transient models in the equipment model library to establish a two-way data interaction channel from the equipment level to the site level is as follows: In the detailed electromagnetic transient model, a set of interface variables that need to be uploaded are defined. The interface variables that need to be uploaded include the fundamental amplitude of the AC side current of the converter, the average value of the DC bus voltage, and the estimated value of the junction temperature of the power device. In the simplified electromechanical transient model, a set of interface variables that need to be issued are defined. The interface variables that need to be issued include the voltage phase angle at the grid connection point of the power station, the estimated value of the system equivalent inertia, and the set value of the dispatch active power. Design a shared memory data area, which is divided into a high-speed access area and a low-speed access area; The interface variables that need to be uploaded from the detailed electromagnetic transient model are low-pass filtered and then written into the low-speed access area of ​​the shared memory data area. The interface variables that need to be issued in the simplified electromechanical transient model are directly written into the high-speed access area of ​​the shared memory data area; A data reading thread is configured for the detailed electromagnetic transient model. The data reading thread reads the estimated values ​​of the grid connection point voltage phase angle and the system equivalent inertia from the high-speed access area at millisecond intervals, which serve as the basis for updating its boundary conditions. A data reading thread is configured for the simplified electromechanical transient model. The data reading thread reads the fundamental amplitude of the AC side current and the average value of the DC bus voltage of the converter from the low-speed access zone at a rate of seconds, which serves as the basis for its equivalent parameter correction. A hardware interrupt mechanism ensures mutual exclusion of two data reading threads, thus avoiding read-write conflicts in the shared memory data area. The bidirectional data interaction channel is established through continuous data exchange in the shared memory data area and periodic synchronization of the data reading thread.

2. The real-time simulation test modeling method for a new energy power station according to claim 1, characterized in that, The construction of a multi-timescale device model library based on the steady-state operating parameters, including detailed electromagnetic transient models and simplified electromechanical transient models, specifically includes: The steady-state operating parameters include the rated power of each power generation unit, the grid connection point voltage, and the current total active power output; The model nameplate data of each power generation unit in the steady-state operating parameters is analyzed. The model nameplate data includes the number of generator pole pairs, rated speed, stator and rotor parameters, and converter switching frequency. Based on the model nameplate data, the physical prototype equations of the synchronous generator or power electronic converter are constructed using the equivalent circuit modeling method. The physical prototype equations include differential terms describing changes in magnetic field energy storage and logical judgment terms reflecting the switching behavior of semiconductor devices. The physical prototype equations are discretized into a set of difference equations suitable for microsecond-level step-size calculations, forming the core computational kernel of the detailed electromagnetic transient model. Meanwhile, based on the current total active power output and grid connection point voltage in the steady-state operating parameters, a set of transfer functions reflecting the overall external characteristics of the station is constructed by combining algebraic equations with first-order inertial elements. The transfer function set is reduced in order, and the low-order approximate expression corresponding to the dominant pole is retained to form the mathematical description of the simplified electromechanical transient model. A high-speed sampling output port is set in the detailed electromagnetic transient model to expose the instantaneous voltage and current values ​​of key internal nodes; A low-speed aggregation input port is set in the simplified electromechanical transient model to receive power commands and voltage reference values ​​from the station control system; An equivalent transformation relationship is established between the high-speed sampling output port and the low-speed aggregation input port. This equivalent transformation relationship is achieved through Fourier decomposition and fundamental component extraction. The computational kernel of the detailed electromagnetic transient model, the mathematical description of the simplified electromechanical transient model, and the equivalent transformation relationship are packaged and encapsulated to form an executable code module of the multi-timescale device model library.

3. The real-time simulation test modeling method for a new energy power station according to claim 2, characterized in that, The step of determining the disturbance scenario sequence for the simulation experiment based on the dynamic adjustment command stream specifically involves: The dynamic adjustment command stream includes power setpoint adjustment commands, reactive power compensation device switching commands, and frequency response action signals. The disturbance scenario sequence includes grid voltage drop events, frequency step change events, and planned power ramp-up events; The dynamic adjustment command stream is timestamped and parsed to restore the original control command timing issued by the scheduling master station; Identify the periodic adjustment patterns and sudden adjustment events in the timing of the original control commands. The periodic adjustment patterns include power tracking commands for automatic generation control, and the sudden adjustment events include rapid power support requirements for primary frequency regulation. According to the technical specifications for grid connection of new energy power plants, a set of standard voltage disturbance waveform templates are defined, including symmetrical voltage drop waveforms, asymmetrical voltage drop waveforms, and phase jump waveforms. Based on actual power grid operation statistics, a set of typical frequency change rate curves are defined, including linear decreasing curves, exponential recovery curves, and oscillation decay curves. The periodic adjustment mode is combined with the standard voltage disturbance waveform template to generate a planned voltage adaptability test scenario. The sudden adjustment event is combined with the typical frequency change rate curve to generate an unplanned frequency stability test scenario. Random noise components are inserted into the voltage adaptability test scenario and the frequency stability test scenario. These random noise components are used to simulate measurement errors and background harmonic interference. The voltage adaptability test scenarios and frequency stability test scenarios are sorted according to the test objectives to form a sequence of disturbance scenarios with clear test objectives. Each scenario is labeled with the expected disturbance type, intensity parameter and duration.

4. The real-time simulation test modeling method for a new energy power station according to claim 3, characterized in that, The step of deploying the device model library on the real-time simulation platform, injecting the disturbance scene sequence, and starting a closed-loop simulation with mixed time scales specifically involves: The executable code module of the device model library is loaded into the multi-core processor of the real-time simulation platform, the multi-core processor containing a core dedicated to microsecond-level calculation and a core dedicated to millisecond-level calculation; The computational kernel of the detailed electromagnetic transient model is assigned to the core dedicated to microsecond-level computation, and a fixed-step interrupt service routine is configured. The mathematical description of the simplified electromechanical transient model is assigned to the core dedicated to millisecond-level calculations, and a task scheduler with variable step size is configured. Initialize the shared memory data area and assign initial values ​​to all interface variables based on the steady-state operating parameters; The description information of the first test scenario is read from the disturbance scenario sequence, and the type and intensity parameters of the disturbance to be applied are parsed out. According to the type of disturbance, the corresponding disturbance generation subroutine is invoked, and the disturbance generation subroutine generates a voltage or frequency excitation signal that meets the standard requirements; The voltage or frequency excitation signal is superimposed on the grid connection point boundary condition of the simplified electromechanical transient model, and the fault initiation flag of the detailed electromagnetic transient model is triggered simultaneously. The computational tasks of the core dedicated to microsecond-level calculations and the core dedicated to millisecond-level calculations are initiated, so that the detailed electromagnetic transient model and the simplified electromechanical transient model begin to synchronously advance the simulation time. The two-way data interaction channel enables real-time data exchange between the two models, forming a hybrid time-scale closed-loop simulation that includes the rapid electromagnetic processes of the equipment and the slow electromechanical processes of the station.

5. The real-time simulation test modeling method for a new energy power station according to claim 4, characterized in that, During the closed-loop simulation, the internal electrical waveforms of the detailed electromagnetic transient model and the global operating state variables of the simplified electromechanical transient model are simultaneously acquired, specifically: Multiple non-destructive probes are inserted into the calculation kernel of the detailed electromagnetic transient model. These non-destructive probes record the pulse state of the power electronic switching device, the instantaneous value of the filter inductor current, and the ripple component of the DC capacitor voltage. The data recorded by the non-destructive probe is saved at a sampling rate no less than the model calculation step size to form a high-fidelity original record of the device-level electrical waveform. Key state variables are marked in the mathematical description of the simplified electromechanical transient model. These key state variables include the smoothed value of the active power output of the power station, the cumulative value of the reactive power, and the integral of the frequency deviation. The numerical changes of the key state variables are recorded at a rate synchronized with the model calculation step size to form a time series of station-level state variables at equal intervals. A synchronous clock source is configured in the real-time simulation platform. The synchronous clock source provides a unified time reference for the acquisition of the high-fidelity equipment-level electrical waveform raw records and the recording of the equally spaced station-level state quantity time series. Add a timestamp and a model identifier to the high-fidelity equipment-level electrical waveform raw record. The model identifier is used to distinguish which specific power generation unit model the data comes from. Add timestamps and scenario identifiers to the equally spaced station-level state quantity time series, wherein the scenario identifiers are used to associate the current data with the disturbance test scenario to which it belongs; The high-fidelity original records of equipment-level electrical waveforms with added identification information and the time series of station-level state variables at equal intervals are stored in the distributed buffer of the real-time simulation platform.

6. The real-time simulation test modeling method for a new energy power station according to claim 5, characterized in that, The process involves extracting features from the internal electrical quantity waveforms to obtain a device-level transient response feature set, and simultaneously performing sequence analysis on the global operating state quantities to obtain a station-level dynamic response feature set. Specifically: Extract current and voltage data within a specific time window before and after the disturbance from the high-fidelity original record of the device-level electrical waveform; Perform a sliding window fast Fourier transform on the current and voltage data within the specific time window to calculate the fundamental amplitude, phase, and harmonic distortion rate of each window; The maximum overshoot, settling time, and steady-state error values ​​in the fundamental amplitude variation curve are extracted as characteristic indicators to characterize the transient process of equipment voltage and current. Identify the switching frequency change patterns and latch-up events in the pulse state records of the power electronic switching device, and count the number of pulse loss and the duration of abnormal conduction. The characteristic indicators representing the transient process of device voltage and current, the switching frequency change mode, and the latching event are packaged together to form the device-level transient response feature set describing the response characteristics of a single device; The active power sequence, reactive power sequence, and grid connection point voltage sequence are separated from the equally spaced station-level state quantity time series. The root mean square value of the tracking error of the active power sequence relative to the dispatch command is calculated as a measure of the power control accuracy of the power station. Calculate the integral area of ​​the reactive power sequence during voltage disturbances as a measure of the reactive power support capability of the power station; The recovery rate of the grid connection point voltage sequence after fault clearance is calculated as a measure of the station voltage stability. The measurement of power control accuracy, reactive power support capability, and voltage stability of the power station are combined to form the power station-level dynamic response feature set that describes the overall behavior of the power station.

7. The real-time simulation test modeling method for a new energy power station according to claim 6, characterized in that, The process of integrating the equipment-level transient response feature set and the station-level dynamic response feature set to generate a dynamic response panoramic map reflecting the overall situation of the new energy power station is as follows: A unique spatial location code is assigned to the equipment-level transient response feature set of each power generation unit, and the spatial location code corresponds to the actual position of the power generation unit in the electrical wiring diagram of the new energy power station; The site-level dynamic response feature set is treated as a global layer and placed as the background layer of the dynamic response panoramic map. According to the spatial location encoding, the device-level transient response feature set of each power generation unit is placed in the corresponding position of the global layer in the form of icons; Within each icon, the fill color represents the maximum overshoot magnitude among the characteristic indicators representing the transient process of the device's voltage and current, and the icon border thickness represents the frequency of abnormal events in the switching frequency variation mode. On the global layer, the change trajectories of the active power sequence, reactive power sequence, and grid connection point voltage sequence in the field-level dynamic response feature set are plotted in the form of curves. Add markers at the key inflection points of the changing trajectory. The markers display the simulation time and specific feature values ​​corresponding to the key inflection points. Establish an interactive link between the icon and the change trajectory. When an icon is selected, highlight the impact traces caused by the action time point of the device unit on the field-level sequence curve. The global layer, the set of icons with visual encoding, the change trajectory curve, and the interactive links are integrated into a unified graphical view framework, which supports zooming of the timeline and switching of layer displays. The final rendered output is the dynamic response panoramic map.

8. The real-time simulation test modeling method for a new energy power station according to claim 7, characterized in that, The step of matching the dynamic response panorama with pre-stored historical typical response patterns to output a model confidence assessment report for the current simulation experiment is as follows: Retrieve historical typical response patterns from the simulation case database. The historical typical response patterns include compliant response patterns that have been verified by previous simulations and real fault response patterns recorded in field tests. Extract the feature vector of the historical typical response pattern. The feature vector is composed of historical equipment-level feature indicators and historical site-level measurement parameters combined according to the same rules. The equipment-level transient response feature set and the station-level dynamic response feature set contained in the dynamic response panoramic map are transformed into the feature vector of the current test according to the same rules. Calculate the Euclidean distance between the feature vector of the current experiment and the feature vector of each historical typical response pattern; Set a similarity threshold, and filter out historical typical response patterns whose Euclidean distance is less than the similarity threshold as a candidate set of similar patterns; Analyze the category to which each pattern belongs in the candidate set of similar patterns, and count the number of times each compliance response pattern and real fault response pattern appear; If the frequency of the compliant response pattern is dominant, the response behavior of the current simulation model is determined to be in line with expectations, and a higher confidence score is assigned. If the frequency of occurrence of the actual fault response mode is dominant, it is determined that the response behavior of the current simulation model is close to the actual fault waveform, but it is necessary to verify whether it is the expected fault ride-through behavior. Based on the combined confidence scores and pattern category analysis results, a structured model confidence assessment report is generated. The report includes a list of similarity matching results, descriptions of major deviation features, and the overall confidence level.

9. A real-time simulation test modeling system for a new energy power station, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the real-time simulation test modeling method for a new energy power station as described in any one of claims 1 to 8.

Images