A simulation method for dynamic characteristics of new energy power grid

Abstract

The application provides a simulation method for dynamic characteristics of a new energy power grid, and belongs to the technical field of new energy power grids, and comprises the following steps: connecting according to an actual topological structure of the new energy power grid to form a new energy power grid model; distributing to different CPU simulation cores for simulation; enabling an adaptive step algorithm to adjust a simulation step in real time according to state changes of the CPU simulation cores, and sequentially obtaining dynamic characteristics of corresponding sub-models after each adjustment of the simulation step; using a Kalman filtering algorithm to pre-process actual power grid operation data to obtain a parameter estimation set of each power grid unit under the new energy power grid model; analyzing dynamic characteristics of corresponding sub-models according to all parameter estimation sets under each sub-model, and determining a regulation and control strategy according to a power grid operation control target. The application provides a reliable basis for power grid planning and operation control, and guarantees safe and stable operation of the power grid.

Description

Technical Field

[0001] This invention relates to the field of new energy power grid technology, and in particular to a method for simulating the dynamic characteristics of new energy power grids. Background Technology

[0002] As the proportion of new energy sources in the power system continues to increase, the scale of new energy power grids is expanding and their structure is becoming increasingly complex. Traditional power grid models have limitations in handling the complex dynamic characteristics after large-scale integration of new energy sources, making it difficult to accurately simulate the intermittency and volatility of new energy power generation. For example, wind and solar power generation is affected by natural conditions, resulting in random variations in power output, which alters the dynamic characteristics of power flow distribution, voltage frequency, and other aspects of the power grid, and traditional models cannot effectively address this. Furthermore, the computational demands of simulating large-scale new energy power grids are enormous, and existing computational resource allocation and simulation algorithms struggle to balance computational efficiency and accuracy, failing to meet the needs of real-time operation analysis and control, and reducing the stability of power grid operation.

[0003] Therefore, this invention proposes a method for simulating the dynamic characteristics of new energy power grids. Summary of the Invention

[0004] This invention provides a method for simulating the dynamic characteristics of new energy power grids to solve the aforementioned technical problems.

[0005] This invention provides a method for simulating the dynamic characteristics of a new energy power grid, comprising:

[0006] Collect detailed electric field information of new energy power plants, establish the operation model of each power grid unit in simulation software, and connect them according to the actual topology of the new energy power grid to form a new energy power grid model;

[0007] The new energy power grid model is decomposed into multiple sub-models according to pre-set rules and assigned to different CPU simulation cores for simulation.

[0008] The adaptive step size algorithm is enabled to adjust the simulation step size in real time according to the state changes of the CPU simulation core, and the dynamic characteristics of the corresponding sub-model are obtained after each adjustment of the simulation step size.

[0009] The actual grid operation data of the new energy grid is acquired in real time, and the actual grid operation data is preprocessed using the Kalman filter algorithm to obtain the parameter estimation set of each grid unit under the new energy grid model.

[0010] The dynamic characteristics of each sub-model are analyzed based on the parameter estimation set of each sub-model, and the control strategy is determined according to the power grid operation control objectives.

[0011] Preferably, the detailed information about the electric field includes: the model, installation location, and operating conditions of each power grid unit.

[0012] Preferably, a new energy power grid model is formed, including:

[0013] The execution function and execution scale are determined based on the operating conditions of the power grid unit, and the design level is determined in combination with the connection relationship under the installation location of the power grid unit, wherein the connection relationship includes existing interconnection and parallel relationships;

[0014] Design the model structure based on the aforementioned design hierarchy and the type of power grid source;

[0015] Various physical phenomena that match the model of the power grid unit are obtained from the model-physical model lookup table, and then coupled and applied to the corresponding model structure to obtain the operating model.

[0016] The information interaction mechanism between different design levels is determined, and the operation models under different design levels are connected according to the actual topology of the new energy power grid to obtain the new energy power grid model.

[0017] Preferably, the adaptive step-size algorithm is enabled to adjust the simulation step-size in real time according to the state changes of the CPU simulation core, including:

[0018] A monitoring mechanism is activated for each CPU emulation core to perform real-time monitoring. The real-time monitoring data is related to CPU emulation core utilization, memory usage, cache hit rate, and task queue length.

[0019] The real-time monitoring data is input into the pre-established CPU simulation core state evaluation model to obtain the state evaluation coefficients at the corresponding time and determine the first error adjustment step size at N consecutive time points.

[0020] The second error adjustment step size is determined based on the dynamic changes of the corresponding CPU simulation core over N consecutive time periods.

[0021] When the first error adjustment step size and the second error adjustment step size are in the same direction, the first unit step size adjustment amount is determined;

[0022]

[0023] Where D1 is the first unit step size adjustment; These represent the first error adjustment step size and the second error adjustment step size, respectively.

[0024] When the directions of the first error adjustment step size and the second error adjustment step size are not consistent, calculate the absolute value of the difference between the first error adjustment step size and the second error adjustment step size. At the same time, determine the data compliance ratio of the corresponding CPU simulation core at N consecutive time points. ,in, This represents the sum of parameters of the corresponding CPU simulation core within a set range over N consecutive time points, and ;

[0025] The correction value Xz is obtained by multiplying the absolute value of the difference by the value obtained by subtracting the data compliance ratio from 1.

[0026] Based on the correction value, the first error adjustment step size, and the second error adjustment step size, determine the second unit step size adjustment amount;

[0027] The step size is adjusted based on the state change coefficient between the previous time step and the current time step, and based on the determined unit step size adjustment amount of the CPU simulation core. The update period of the unit step size adjustment amount is T, and the number of consecutive time steps involved in the update period T is greater than or equal to N.

[0028] Preferably, determining the second unit step size adjustment amount based on the correction value, the first error adjustment step size, and the second error adjustment step size includes:

[0029]

[0030] in, This indicates the second unit step size adjustment; , It is a symbolic function, and , .

[0031] Preferably, the actual power grid operation data is preprocessed using the Kalman filter algorithm, including:

[0032] The actual power grid operation data is converted into a data sequence;

[0033] The historical operation data of the new energy power grid is obtained, and the compliance coefficient and expected coefficient of each historical data group are analyzed. The process noise covariance matrix and measurement noise covariance matrix are adjusted, and the Kalman filter algorithm is updated.

[0034] The updated algorithm is used to filter noise from the data sequence to obtain a new sequence.

[0035] The parameter estimation set of the corresponding power grid unit is obtained based on the new sequence at M consecutive time points.

[0036] Preferably, the dynamic characteristics of the corresponding sub-model are analyzed based on the parameter estimation set of all sub-models, including:

[0037] For each parameter estimation set under the sub-model, perform in-set global analysis and out-of-set global analysis to obtain in-set error and out-of-set error;

[0038] Minimize the error of all in-set errors under the sub-model and compare it with the out-of-set errors to obtain the reference error for each parameter;

[0039] All parameter estimates are sequentially input into the first characteristic analysis model to obtain the nonlinear mapping relationship between each parameter and the dynamic characteristics of the corresponding sub-model;

[0040] Sensitivity analysis is performed on each parameter in the parameter estimation set, and the influence of each parameter on the dynamic characteristics is quantified by combining nonlinear mapping relationships. Furthermore, the key parameter combinations that affect the changes in dynamic characteristics are identified by combining reference errors.

[0041] By combining the power grid topology, the dynamic characteristic correlation between sub-models at different locations is analyzed, the combination of key parameters is verified, and parameter weights are assigned to each key parameter.

[0042] Preferably, the control strategy is determined based on the power grid operation control objectives, including:

[0043] The power grid operation control objectives are described using quantitative indicators;

[0044] The description results, key parameters, and parameter weights are combined into an input vector and input into the strategy analysis model to obtain the control strategy.

[0045] Preferably, after inputting all parameter estimates into the first characteristic analysis model sequentially, the method further includes:

[0046] Obtain the work log of the first feature analysis model, and match each work item in the work log with the corresponding item traceability list to obtain the parameter description of the corresponding work item;

[0047] Retrieve a capture tool that matches the parameter description from the description-tool database, and capture the working path that matches the parameter description and generate path data based on the working path in real time according to the capture tool;

[0048] The generated data from the path is subjected to indicator feature extraction to obtain a generation vector;

[0049] The generated vector is input into the vector analysis model to obtain the early warning regulations, which include: early warning type and early warning information;

[0050] The warning regulations for each work item are compared and analyzed with the standard regulations to identify work anomaly factors;

[0051] The rationality of the work is analyzed by combining abnormal factors and parameter combinations;

[0052] If the reasonableness of the work is greater than the preset reasonableness, the working process of the first characteristic analysis model is determined to be reasonable;

[0053] Otherwise, based on the anomaly factors, parameter combinations, and working path, determine the reasons for optimizing the first characteristic analysis model.

[0054] Compared with the prior art, the beneficial effects of this application are as follows:

[0055] This solution, through the establishment of detailed models, parallel computing, and adaptive step size adjustment, can quickly and accurately simulate the dynamic characteristics of the power grid, providing a reliable basis for power grid planning and operation control, and ensuring the safe and stable operation of the power grid. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0057] Figure 1 This is a flowchart of a method for simulating the dynamic characteristics of a new energy power grid in an embodiment of the present invention. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0059] This invention provides a method for simulating the dynamic characteristics of new energy power grids, such as... Figure 1 As shown, it includes:

[0060] Step 1: Collect detailed electric field information of new energy power plants, establish the operation model of each power grid unit in simulation software, and connect them according to the actual topology of the new energy power grid to form a new energy power grid model;

[0061] Step 2: Decompose the new energy power grid model into multiple sub-models according to pre-set rules, and assign them to different CPU simulation cores for simulation;

[0062] Step 3: Enable the adaptive step size algorithm to adjust the simulation step size in real time according to the state changes of the CPU simulation core, and obtain the dynamic characteristics of the corresponding sub-model after each adjustment of the simulation step size.

[0063] Step 4: Acquire the actual grid operation data of the new energy grid in real time, and use the Kalman filter algorithm to preprocess the actual grid operation data to obtain the parameter estimation set of each grid unit under the new energy grid model;

[0064] Step 5: Analyze the dynamic characteristics of the corresponding sub-model based on the parameter estimation set of each sub-model, and determine the control strategy according to the power grid operation control objectives.

[0065] Preferably, the detailed information about the electric field includes: the model, installation location, and operating conditions of each power grid unit.

[0066] In this embodiment, the new energy power plant is a large-scale wind power plant with hundreds of wind turbines or photovoltaic array panels. Taking a certain region's new energy power grid as an example, the following data are collected: wind turbine model, installation location (latitude and longitude coordinates), and historical wind speed data for the local wind farm; photovoltaic panel model, installation tilt angle, and solar radiation data for the photovoltaic power station. Using PSCAD / EMTDC electric field simulation software, the operation model of each power grid unit is established based on equipment parameters and physical principles, and then connected according to the actual topology to form a new energy power grid model.

[0067] The power grid unit includes new energy power generation equipment (wind turbines, photovoltaic panels), power electronic converters, transmission lines, transformers, and various loads (industrial and residential loads).

[0068] Operational models, such as the output power model of photovoltaic panels, calculate power based on sunlight and temperature; the aerodynamic model of wind turbines determines power generation based on wind speed; and the new energy power grid model is an overall model constructed by combining power grid unit models such as photovoltaic power plants, wind farms, transmission lines, and load centers within a region according to their actual connection relationships. Sub-models are parts of the new energy power grid model after being broken down, such as separate wind power sub-models, photovoltaic sub-models, and transmission line sub-models. The above power grid model is divided into wind power sub-models, photovoltaic sub-models, transmission line sub-models, etc., according to region and function. These sub-models interact with each other through a message passing interface (MPI) to achieve parallel simulation.

[0069] The adaptive step-size algorithm employs the variable-step-size Runge-Kutta algorithm, with a system error threshold set during the process. When the system state changes slightly, the step size is increased to speed up the calculation; when the changes are drastic, the step size is decreased to maintain accuracy. The dynamic characteristics of the sub-model, such as voltage and frequency, are recorded in real time after the step size is adjusted.

[0070] In this embodiment, the actual power grid operation data consists of voltage, current, power, frequency, and other data acquired in real-time by power grid monitoring. The parameter estimation set is a collection of estimated values ​​for power grid unit parameters, such as the estimated values ​​for transmission line resistance, inductance, capacitance, current, and voltage. Voltage and current data are collected in real-time by the SCADA system within the power grid, once per second. The data is input into MATLAB and processed using a written Kalman filter program to obtain the estimated values ​​for each power grid unit parameter; for example, the estimated resistance of a certain transmission line segment is 0.5Ω.

[0071] In this embodiment, the control strategy comprises control measures formulated to achieve the power grid operation goals, such as adjusting the active / reactive power of generators and switching reactive power compensation equipment. MATLAB data analysis tools are used to analyze the parameter estimation sets of each sub-model to determine the dynamic characteristics of the sub-models, such as stability and response speed. Combined with the goal of safe and stable power grid operation, control strategies are formulated; for example, when the voltage in a certain area is low, it is recommended to activate capacitor banks.

[0072] In this embodiment, the model number represents uniqueness, such as the wind turbine model Vestas V164-9.5MW and the photovoltaic panel model Jinko TigerNeo series.

[0073] The installation location is expressed in latitude and longitude, such as a photovoltaic power station located at 30° North latitude and 110° East longitude.

[0074] Operating conditions include ambient temperature, light intensity, wind speed, humidity, etc., such as an average annual wind speed of 8 m / s for a certain wind farm.

[0075] When collecting information on new energy power plants, detailed records are kept of the model, installation location, and operating conditions of each grid unit. For example, after recording the wind turbine model, parameters such as its rated power and blade size can be obtained; the connection relationship with other equipment can be determined based on the installation location; and the equipment performance can be estimated based on the operating conditions. This information is used for subsequent model building to make the model more realistic.

[0076] The model, installation location, and operating conditions of power grid units directly affect their performance and role in the power grid. Accurately considering these factors enables the established model to more realistically reflect the actual situation of the power grid, improves simulation accuracy, and provides more reliable support for power grid operation analysis and decision-making.

[0077] The beneficial effects of the above technical solution are: by establishing detailed models, parallel computing, and adaptive step size adjustment, this solution can quickly and accurately simulate the dynamic characteristics of the power grid, providing a reliable basis for power grid planning and operation control, and ensuring the safe and stable operation of the power grid.

[0078] This invention provides a method for simulating the dynamic characteristics of a new energy power grid, forming a new energy power grid model, including:

[0079] The execution function and execution scale are determined based on the operating conditions of the power grid unit, and the design level is determined in combination with the connection relationship under the installation location of the power grid unit, wherein the connection relationship includes existing interconnection and parallel relationships;

[0080] Design the model structure based on the aforementioned design hierarchy and the type of power grid source;

[0081] Various physical phenomena that match the model of the power grid unit are obtained from the model-physical model lookup table, and then coupled and applied to the corresponding model structure to obtain the operating model.

[0082] The information interaction mechanism between different design levels is determined, and the operation models under different design levels are connected according to the actual topology of the new energy power grid to obtain the new energy power grid model.

[0083] In this embodiment, operating condition data of the grid unit is collected. For example, for photovoltaic panels, data such as light intensity, temperature, and humidity are collected; for wind turbines, data such as wind speed, wind direction, and air density are collected. Taking photovoltaic panels as an example, based on the light intensity-power characteristic curve (generally provided by the photovoltaic panel manufacturer or determined experimentally), when the light intensity reaches a certain threshold, its function of converting light energy into electrical energy is determined. Based on the specifications of the photovoltaic panel (such as single-panel power and quantity) and operating conditions such as light intensity, its power generation is calculated, thereby determining the scale of operation. The installation location of the grid unit is clarified, and its connection relationship with surrounding equipment is analyzed. For example, in a new energy power generation park, photovoltaic panels and inverters are interconnected in series, and multiple such series combinations are connected in parallel. Based on these connections, individual devices such as photovoltaic panels and inverters are considered as component-level design levels; photovoltaic panel-inverter combinations are considered as subsystem-level design levels; and the entire park's power generation, transmission, and transformation systems are considered as system-level design levels. Operating conditions directly affect the functional realization and scale of the grid unit; for example, insufficient light intensity will limit the power generation of the photovoltaic panel. The installation location and interconnection relationships determine the position and role of the power grid unit within the entire system, thus influencing the design hierarchy. A proper determination of these elements provides an accurate foundation for subsequent model construction. By collecting and analyzing data on operating conditions and interconnection relationships, the functional scope, scale, and design hierarchy of the power grid unit are determined. For example, if a photovoltaic power station, under current illumination conditions, performs the function of converting solar energy into electrical energy, with a daily power generation of X megawatt-hours, the design hierarchy might include component-level (photovoltaic panels, inverters, etc.) and subsystem-level (photovoltaic panel-inverter combinations). This result provides a clear direction for the subsequent design model structure.

[0084] In this embodiment, the grid power supply model is considered based on the design level determined in the first step. For example, for a wind farm using a certain type of wind turbine, if the design level is at the subsystem level, the connection relationship between the wind turbine and components such as the tower and foundation, as well as the interaction logic with the control system, needs to be considered. In simulation software (such as MATLAB / Simulink, PSCAD / EMTDC, etc.), the corresponding model structure is built using the modules and tools provided by the software. For example, in PSCAD, the electrical component module is used to build the electrical model of the wind turbine, and the mechanical module is used to build the mechanical model of the wind turbine, and they are connected according to the actual physical connection relationship. The design level determines the complexity and components of the model, while the grid power supply model affects the selection of specific modules and parameters. A reasonable model structure design can accurately reflect the actual operation of the grid unit, facilitating subsequent simulation and analysis. In the simulation software, the model structure is built step by step according to the design level and power supply model requirements. For example, the model structure of a certain wind power electronic system is completed, including the wind turbine body model, the control system model, and the connection line model between them. This model structure can initially simulate the operating state of the wind turbine under different wind speeds and other conditions, preparing for further addition of physical phenomena and model improvement.

[0085] In this embodiment, a model-physical model lookup table is established, recording the physical phenomena corresponding to different grid unit models, such as the photoelectric conversion and heating phenomena of photovoltaic panels, and the aerodynamic and mechanical vibration phenomena of wind turbines. Based on the grid unit model, the corresponding physical phenomena are obtained from the lookup table. Then, in simulation software, the multiphysics coupling function (such as COMSOL Multiphysics software, which can achieve multiple physical field couplings) is used to couple the obtained physical phenomena. For example, for photovoltaic panels, the current generated by photoelectric conversion is coupled with the temperature change caused by heating, considering the impact of temperature on photoelectric conversion efficiency. The coupled physical phenomena are applied to the model structure designed in the second step to obtain the operating model. Different grid unit models have different physical characteristics and phenomena. Accurately obtaining these physical phenomena through the model-physical model lookup table and coupling them allows the model to more realistically reflect the actual operating behavior of the grid unit. Because in actual operation, multiple physical phenomena interact, such as the increase in photovoltaic panel temperature reducing power generation efficiency. By looking up the lookup table and performing coupling operations, multiple physical phenomena are incorporated into the model structure to obtain the operating model. For example, a photovoltaic panel operating model considering photoelectric conversion, heating, and their coupling relationship is obtained. This model can more accurately simulate the output power changes of photovoltaic panels under different light and temperature conditions, and the simulation accuracy is significantly improved compared with models that do not consider coupling relationships.

[0086] In this embodiment, the data flow and interaction requirements between different design levels are analyzed to determine the information interaction mechanism. For example, in a power grid system containing new energy power generation units, transmission lines, and loads, the component-level power generation unit model needs to transmit power data to the subsystem-level transmission line model, and the subsystem-level model needs to transmit voltage, current, and other data to the system-level model. The communication protocol (such as Modbus, TCP / IP, etc.) and data format for information interaction are determined. Then, based on the actual topology of the new energy power grid, the operating models at different design levels are connected in the simulation software. For example, in the power system simulation software DIgSILENT, the power generation unit model, transmission line model, and load model are connected according to the node connection relationship of the actual power grid. Information interaction between models at different design levels is necessary to accurately simulate the operation of the entire power grid system. Determining a suitable information interaction mechanism ensures accurate and timely data transmission. Connecting models according to the actual topology ensures that the constructed new energy power grid model is consistent with the actual power grid, improving the model's practicality and accuracy. By determining the information interaction mechanism and connecting model operations, a new energy power grid model is constructed. For example, a regional renewable energy power grid model can be constructed, which includes multiple renewable energy power generation units, transmission lines, and loads. This model can simulate the operating status of the entire regional power grid under different operating conditions, such as the changes in grid voltage and frequency when renewable energy power generation fluctuates, providing an effective tool for power grid operation analysis and regulation strategy formulation.

[0087] The beneficial effects of the above technical solution are: by clearly defining the execution functions and scale, designing reasonable levels and structures, coupling physical phenomena and establishing interaction mechanisms, a more accurate and comprehensive power grid model can be constructed, effectively simulating the dynamic characteristics of the power grid and meeting the needs of power grid analysis and control.

[0088] This invention provides a method for simulating the dynamic characteristics of new energy power grids, which enables an adaptive step-size algorithm to adjust the simulation step size in real time according to the state changes of the CPU simulation core, including:

[0089] A monitoring mechanism is activated for each CPU emulation core to perform real-time monitoring. The real-time monitoring data is related to CPU emulation core utilization, memory usage, cache hit rate, and task queue length.

[0090] The real-time monitoring data is input into the pre-established CPU simulation core state evaluation model to obtain the state evaluation coefficients at the corresponding time and determine the first error adjustment step size at N consecutive time points.

[0091] The second error adjustment step size is determined based on the dynamic changes of the corresponding CPU simulation core over N consecutive time periods.

[0092] When the first error adjustment step size and the second error adjustment step size are in the same direction, the first unit step size adjustment amount is determined;

[0093]

[0094] Where D1 is the first unit step size adjustment; These represent the first error adjustment step size and the second error adjustment step size, respectively.

[0095] When the directions of the first error adjustment step size and the second error adjustment step size are not consistent, calculate the absolute value of the difference between the first error adjustment step size and the second error adjustment step size. At the same time, determine the data compliance ratio of the corresponding CPU simulation core at N consecutive time points. ,in, This represents the sum of parameters of the corresponding CPU simulation core within a set range over N consecutive time points, and ;

[0096] The correction value Xz is obtained by multiplying the absolute value of the difference by the value obtained by subtracting the data compliance ratio from 1.

[0097] Based on the correction value, the first error adjustment step size, and the second error adjustment step size, determine the second unit step size adjustment amount;

[0098] The step size is adjusted based on the state change coefficient between the previous time step and the current time step, and based on the determined unit step size adjustment amount of the CPU simulation core. The update period of the unit step size adjustment amount is T, and the number of consecutive time steps involved in the update period T is greater than or equal to N.

[0099] In this embodiment, tools or interfaces provided by the operating system are used to monitor the CPU emulation core. For example, in Linux systems, the `top` or `ps` commands can be used to obtain information such as CPU utilization and memory usage; in Windows systems, PerformanceMonitor can be used to obtain relevant data. Scripts or programs are written to periodically (e.g., once per second) collect data such as CPU emulation core utilization, memory usage (the proportion of used memory to total memory), cache hit rate (the proportion of times the CPU successfully reads data from the cache), and task queue length (the number of tasks waiting for the CPU to process). This monitoring data reflects the real-time load and operating status of the CPU emulation core. CPU utilization reflects the CPU's workload; memory usage reflects the utilization of memory resources, and excessively high usage may lead to decreased system performance; cache hit rate affects the speed at which the CPU retrieves data, and a low hit rate may mean frequent data retrieval from slow storage; task queue length indicates the backlog of tasks waiting to be processed. By monitoring these data in real time, the status of the CPU emulation core can be understood promptly, providing a basis for subsequent step size adjustments. During program execution, the monitoring mechanism continuously collects data. For example, data is acquired once per second, resulting in a series of time-series data. For instance, in the first second, CPU utilization is 30%, memory usage is 60%, cache hit rate is 70%, and the task queue length is 5. In the second second, these data will change based on the CPU's performance. This real-time monitoring data will serve as input for subsequent steps.

[0100] In this embodiment, a CPU emulation core status assessment model is pre-established. This model can be based on machine learning algorithms (such as neural networks, decision trees, etc.) or traditional mathematical statistical methods (such as linear regression, multiple regression, etc.). The real-time monitoring data collected in the first step is input into this model. Through internal calculations and processing, the model outputs a status assessment coefficient for the corresponding time point. This coefficient can be a numerical value used to quantify the state of the CPU emulation core. For example, a larger value indicates a worse state, possibly signifying excessive load or resource strain. Then, based on the status assessment coefficient and pre-defined rules (e.g., calculating based on a comparison of the status assessment coefficient with a threshold, combined with a proportionality coefficient), the first error adjustment step size for N consecutive time points is determined. Assuming N is 10, the first error adjustment step size is calculated based on the status assessment coefficients of the past 10 time points. The CPU emulation core status assessment model is the core component; it comprehensively considers multiple monitoring indicators and evaluates the CPU state through a specific algorithm. The status assessment coefficient provides a quantitative basis for subsequent step size adjustments. The determination of the first error adjustment step size is based on the analysis of the CPU's state over a past period, aiming to initially determine the adjustment direction and magnitude of the step size based on the CPU's load. After real-time monitoring data is input into the model, the model calculates the state evaluation coefficient. For example, the calculated state evaluation coefficient at a certain moment is 0.8 (assuming it is between 0 and 1, with the closer to 1 indicating a worse state). According to pre-set rules, the first error adjustment step size is calculated to be 0.05 (assuming the step size adjustment unit is seconds). This yields the first error adjustment step size for N consecutive moments, which is used for subsequent step size adjustment decisions.

[0101] In this embodiment, the dynamic changes of the corresponding CPU emulator core over N consecutive time periods are analyzed. These dynamic changes can be reflected by calculating the difference or rate of change of monitoring data between adjacent time periods. For example, the rate of change of CPU utilization between adjacent time periods, or the difference in memory usage, etc. Based on these dynamic change data, combined with a pre-set algorithm (e.g., calculated based on the slope of the trend and an adjustment factor), a second error adjustment step size is determined. This step size reflects the step adjustment amount determined based on the dynamic change trend of the CPU emulator core. Considering only the state evaluation coefficient is insufficient; it is also necessary to analyze the dynamic change trend of the CPU emulator core. Even if the current state evaluation coefficients are the same, if one shows a rapid upward trend and the other a slow downward trend, the step size adjustment strategy should be different. The second error adjustment step size is designed to capture the impact of this dynamic change on the step size adjustment. Dynamic change analysis is performed on the monitoring data over N consecutive time periods. Assume that the CPU utilization shows an upward trend over these N time periods, with a rate of change of 0.02 per second. Based on a pre-defined algorithm, the second error adjustment step size is calculated to be -0.03 (the negative sign indicates that the step size may need to be reduced; the unit is also assumed to be seconds). This second error adjustment step size will be used together with the first error adjustment step size in subsequent step size adjustment decisions.

[0102] In this embodiment, it is determined whether the directions of the first error adjustment step size and the second error adjustment step size are consistent. This can be determined by comparing the signs of the two step sizes. If the signs of the two step sizes are the same, it means that their adjustment directions are consistent. When the directions of the two error adjustment step sizes are consistent, it means that the direction of the step size adjustment is clear based on the CPU's state evaluation and dynamic change trend. The first unit step size adjustment amount is calculated using this formula, which comprehensively considers the size of the two step sizes and the time span (represented by N), making the step size adjustment more reasonable.

[0103] Assumption It is 0.05. With a value of 0.03 and N = 10, D1 is 0.004. This result will be used for subsequent step size adjustment operations.

[0104] The set range can be CPU utilization of 0-100%, memory usage of 0-100%, etc., and the total sum of parameters Wn within these ranges is calculated.

[0105] When the directions of the two error adjustment step sizes are inconsistent, it indicates that the CPU's state assessment and dynamic change trend are providing different step size adjustment signals. In this case, more factors need to be considered to determine the step size adjustment amount. The absolute value of the difference reflects the degree of difference between the two step sizes, while the data compliance ratio reflects the compliance of the CPU parameters over N consecutive time periods. By calculating the correction value and combining it with the other two step sizes, the step size adjustment amount can be determined more accurately.

[0106] In this embodiment, the state change coefficient between the previous and current moments is calculated. This coefficient can be obtained by calculating the difference or rate of change of the state evaluation coefficient between adjacent moments. For example, the state evaluation coefficient of the previous moment is subtracted from the state evaluation coefficient of the current moment, and then divided by the time interval. Then, step size adjustment is performed according to the determined unit step size adjustment amount of the CPU simulation core (the first unit step size adjustment amount is when the two error adjustment step sizes are in the same direction, and the second unit step size adjustment amount is when they are not in the same direction). The update period of the unit step size adjustment amount is T, and the number of consecutive moments involved in the update period T is greater than or equal to N. For example, T can be set to 10 seconds, that is, the step size adjustment amount is updated every 10 seconds according to the above calculation results, and the step size adjustment operation is performed. The state change coefficient reflects the change of the state of the CPU simulation core in a short period of time. Combining the unit step size adjustment amount with step size adjustment can make the step size adjustment more timely and accurate to adapt to the state changes of the CPU. Setting the update period T is to balance the amount of computation and the timeliness of step size adjustment, and to avoid adjusting the step size too frequently or too sparsely. Assuming the state evaluation coefficient was 0.7 at the previous moment and 0.6 at the current moment, with a time interval of 1 second, the state change coefficient is (0.7 − 0.6) ÷ 1 = 0.1. If the previously determined unit step size adjustment is 0.01 (assuming it's the second unit step size adjustment), then the simulation step size is adjusted based on these values. For example, the current simulation step size is increased by 0.01 (assuming the step size unit is seconds). After adjustment, the simulation will continue to run according to the new step size to better adapt to the state changes of the CPU simulation core.

[0107] The beneficial effects of the above technical solution are: by monitoring the core status in real time and dynamically adjusting the step size, computing resources can be fully utilized, avoiding the impact of excessive load on a certain core on the overall simulation efficiency and accuracy, and ensuring the efficient and accurate operation of large-scale power grid simulation.

[0108] This invention provides a method for simulating the dynamic characteristics of a new energy power grid, which determines a second unit step adjustment amount based on the correction value, a first error adjustment step size, and a second error adjustment step size, including:

[0109]

[0110] in, This indicates the second unit step size adjustment; , It is a symbolic function, and , .

[0111] The beneficial effects of the above technical solution are as follows: In complex power grid dynamic simulations, this method can adjust the step size more accurately. Compared with simple averaging or single-judgment methods, the accuracy of step size adjustment is improved by 20%, further enhancing simulation efficiency and accuracy. When adjusting the step size by comprehensively considering CPU core states and system dynamic changes, accurately determining the adjustment direction and magnitude is crucial. The introduction of the sign function and the calculation of specific formulas make the step size adjustment more scientific and reasonable, adapting to complex and ever-changing simulation scenarios and ensuring the reliability of simulation results.

[0112] This invention provides a method for simulating the dynamic characteristics of a new energy power grid, which utilizes the Kalman filter algorithm to preprocess actual power grid operation data, including:

[0113] The actual power grid operation data is converted into a data sequence;

[0114] The historical operation data of the new energy power grid is obtained, and the compliance coefficient and expected coefficient of each historical data group are analyzed. The process noise covariance matrix and measurement noise covariance matrix are adjusted, and the Kalman filter algorithm is updated.

[0115] The updated algorithm is used to filter noise from the data sequence to obtain a new sequence.

[0116] The parameter estimation set of the corresponding power grid unit is obtained based on the new sequence at M consecutive time points.

[0117] In this embodiment, the real-time collected actual power grid operation data is organized into a data sequence according to time order.

[0118] The beneficial effects of the above technical solution are: the Kalman filter algorithm, by utilizing historical data to update algorithm parameters, can effectively remove noise and accurately estimate grid unit parameters. This provides an accurate data foundation for subsequent model analysis and control strategy formulation, and is a key step in improving simulation accuracy.

[0119] This invention provides a method for simulating the dynamic characteristics of a new energy power grid. It analyzes the dynamic characteristics of each sub-model based on the parameter estimation sets of all sub-models, including:

[0120] For each parameter estimation set under the sub-model, perform in-set global analysis and out-of-set global analysis to obtain in-set error and out-of-set error;

[0121] Minimize the error of all in-set errors under the sub-model and compare it with the out-of-set errors to obtain the reference error for each parameter;

[0122] All parameter estimates are sequentially input into the first characteristic analysis model to obtain the nonlinear mapping relationship between each parameter and the dynamic characteristics of the corresponding sub-model;

[0123] Sensitivity analysis is performed on each parameter in the parameter estimation set, and the influence of each parameter on the dynamic characteristics is quantified by combining nonlinear mapping relationships. Furthermore, the key parameter combinations that affect the changes in dynamic characteristics are identified by combining reference errors.

[0124] By combining the power grid topology, the dynamic characteristic correlation between sub-models at different locations is analyzed, the combination of key parameters is verified, and parameter weights are assigned to each key parameter.

[0125] In this embodiment, actual power grid operation data typically comes from various monitoring devices, such as substation measurement and control devices and power plant data acquisition systems. This data may be stored and transmitted in different formats, such as CSV and JSON. First, these data files need to be read, and then sorted according to the order of their timestamps. For example, using the pandas library in Python, the `read_csv()` function can be used to read CSV format power grid operation data files, and then the `sort_values()` function can be used to sort the data in ascending order by the time column, thus organizing the data into a time-ordered sequence. Converting the actual power grid operation data into a data sequence facilitates subsequent processing. Power grid operation data changes dynamically over time; arranging it in chronological order better reflects the sequential correlation of the data, meets the requirements of the Kalman filter algorithm for time-series data processing, and facilitates the analysis of data changes over time, as well as noise filtering and parameter estimation. After reading and sorting, the originally chaotic data is organized into an ordered data sequence. For example, a set of power grid voltage data collected every 5 minutes is processed to form a data sequence arranged from the earliest collection time to the latest collection time. Each data point corresponds to a specific collection time, providing a standardized input data format for subsequent steps.

[0126] In this embodiment, historical operating data of the new energy power grid is queried and extracted from the power grid's data storage system, such as a database (MySQL, Oracle, etc.). SQL queries can be written to filter the required historical data based on conditions such as time range and data type. For example, querying daily power, voltage, and current data for the past year. For each historical data group, a certain compliance standard is defined, such as voltage within ±5% of the rated value being considered compliant. The proportion of compliant data in each data group is statistically analyzed and used as the compliance coefficient. The expected coefficient can be determined by calculating statistics such as the mean and median of the historical data, reflecting the average level or central tendency of the data. For example, the average value of historical power data is calculated as the expected coefficient for the power data. The process noise covariance matrix Q and the measurement noise covariance matrix R are adjusted based on the compliance coefficient and the expected coefficient. In the Kalman filter algorithm, these two matrices are used to describe the statistical characteristics of system noise and measurement noise. The adjustment method can be determined based on empirical formulas or machine learning algorithms. For example, if the compliance coefficient is low, it indicates high data noise; the values ​​of the Q and R matrices are increased accordingly to enhance the algorithm's noise processing capability. After adjusting the matrix, the new matrix parameters are substituted into the Kalman filter algorithm to complete the algorithm update. Historical operating data contains past operating characteristics and noise properties of the power grid. By analyzing the compliance coefficient and expected coefficient, the quality and distribution of the data can be understood. The process noise covariance matrix and the measurement noise covariance matrix are key parameters of the Kalman filter algorithm, which determine how the algorithm estimates and processes noise. Adjusting these two matrices based on the analysis results of historical data allows the Kalman filter algorithm to better adapt to the characteristics of power grid data, improving the filtering effect and the accuracy of parameter estimation. Through a series of operations, the Kalman filter algorithm is updated. For example, after analyzing historical voltage data, it was found that the compliance coefficient was low, indicating that the voltage data had significant noise. Therefore, the values ​​of voltage-related elements in the measurement noise covariance matrix R were increased. The updated Kalman filter algorithm can more effectively handle noise in voltage data, providing a more reliable tool for subsequent data filtering and parameter estimation.

[0127] In this embodiment, the data sequence obtained in the first step is input into the updated Kalman filter algorithm. The Kalman filter algorithm, based on the system state equation and measurement equation, processes the data through two steps: prediction and updating. In the prediction step, the state at the current moment is predicted based on the state estimate from the previous moment and the system state transition matrix. In the updating step, the predicted state is corrected using the current measurement value and the measurement matrix to obtain the optimal state estimate for the current moment. By iterating through these two steps, each data point in the data sequence is processed to remove the influence of noise, resulting in a new sequence after noise filtering. The Kalman filter algorithm can be implemented using relevant functions in MATLAB software, such as the `kalman()` function, which takes the data sequence as input parameters for calculation. Power grid operation data is inevitably subject to various noise interferences during acquisition and transmission, such as electromagnetic interference and measurement equipment errors. The Kalman filter algorithm can utilize the system's dynamic model and noise statistical characteristics to estimate and compensate for these noises, thereby extracting more realistic signals. By filtering the data sequence for noise, the quality of the data can be improved, laying the foundation for accurate estimation of power grid unit parameters. After the data sequence is input into the updated Kalman filter algorithm, a new sequence is obtained through the algorithm's processing. For example, for a set of noisy grid current data sequences, after Kalman filtering, the data fluctuations become smoother, outliers caused by noise are effectively removed, and a new sequence that better reflects the actual current change trend is obtained.

[0128] In this embodiment, data from M consecutive time points are selected from the new sequence after noise filtering. The value of M can be determined based on actual needs and data characteristics. For example, if the dynamic changes of the power grid unit are relatively slow, M can be a larger value; if the changes are rapid, M can be a smaller value. For each power grid unit, its parameters are estimated using this data from the M consecutive time points, based on its physical model and related electrical principles. For example, for a transmission line unit, its parameters may include resistance, inductance, and capacitance. The values ​​of resistance, inductance, and capacitance can be calculated using Ohm's law, the law of electromagnetic induction, and the voltage and current data from these M time points, combined with parameter estimation methods such as the least squares method, thus obtaining the parameter estimation set for the transmission line unit. The parameters of the power grid unit are key indicators describing its electrical characteristics and operating behavior. By analyzing and processing the noise-filtered data, these parameters can be estimated more accurately. Selecting data from the M consecutive time points is to comprehensively consider the operating conditions of the power grid unit over a period of time, avoiding the influence of the randomness of data from a single time point on the parameter estimation results. Using appropriate parameter estimation methods, parameter values ​​that best reflect the actual operating conditions of the power grid unit can be inferred from the data. By processing and calculating the new sequence data over M consecutive time periods, a parameter estimation set for the corresponding power grid unit is obtained. For example, for a certain transmission line, the calculated estimated resistance is 0.5Ω, the estimated inductance is 0.01H, and the estimated capacitance is 0.0001F. These parameter values ​​constitute the parameter estimation set for this transmission line unit, providing an important basis for subsequent analysis of the dynamic characteristics of the transmission line and the operating status of the entire power grid.

[0129] The beneficial effects of the above technical solution are: by comprehensively analyzing the parameter estimation set, determining key parameters and weights, important factors can be highlighted, the analysis process can be simplified, the efficiency and accuracy of dynamic characteristic analysis can be improved, and strong support can be provided for formulating effective control strategies.

[0130] This invention provides a method for simulating the dynamic characteristics of a new energy power grid, which determines a control strategy based on the grid operation control objectives, including:

[0131] The power grid operation control objectives are described using quantitative indicators;

[0132] The description results, key parameters, and parameter weights are combined into an input vector and input into the strategy analysis model to obtain the control strategy.

[0133] In this embodiment, based on the previous analysis of the dynamic characteristics of the power grid sub-model, key parameters that significantly affect the power grid operation control objectives are identified. For example, in the new energy power generation unit, irradiance, temperature, and inverter control parameters may be key parameters; in the transmission line, line impedance, transmission power, and reactive power compensation may be key parameters. Appropriate methods, such as the Analytic Hierarchy Process (AHP) and entropy weighting, are used to determine the weights of each key parameter to represent their importance to the power grid operation control objectives. The quantitative index data obtained in the first step is combined with the key parameters and their weights to form an input vector. For example, assuming there are 6 quantitative indicators (voltage deviation, frequency, active power deficit ratio, power factor, grid loss rate, and new energy power generation utilization rate) and 4 key parameters (irradiance, temperature, line impedance, and reactive power compensation) and their corresponding weights, then the dimension of the input vector is 6+4+4 (parameter weights). These data are arranged into a vector form in a certain order. The generated input vector is then input into a pre-established strategy analysis model. Strategy analysis models can be based on artificial intelligence algorithms, such as deep reinforcement learning (DRL) and neural networks (NN), or on traditional optimization algorithms, such as linear programming (LP) and integer programming (IP). By learning and analyzing input vectors, the model utilizes its internal algorithms and rules to output control strategies tailored to the current power grid operating state. For example, a strategy analysis model based on deep reinforcement learning will infer from the input vectors within its trained strategy network, outputting control strategies such as adjusting generator active power and switching reactive power compensation equipment.

[0134] In this embodiment, key parameters and their weights reflect the factors and their importance that significantly influence the power grid operation control objectives. Combining these with quantitative indicators into an input vector allows for the comprehensive input of relevant power grid operation information into the strategy analysis model. By processing and analyzing the input vector, the strategy analysis model can generate appropriate control strategies based on its learned knowledge and rules to achieve the power grid operation control objectives. Different strategy analysis models have different characteristics and advantages; selecting the appropriate model is crucial for generating effective control strategies. After the combined input vector is input into the strategy analysis model, the model performs calculations and inferences, outputting control strategies. For example, the control strategy output by the strategy analysis model might be: increase the active power of a generator by 10MW and connect a set of capacitors to compensate for reactive power. These control strategies are generated based on the quantitative indicators of the current power grid operation state, as well as key parameters and weights, aiming to optimize power grid operation and guide it towards the set control objectives.

[0135] The beneficial effects of the above technical solution are: the strategy analysis model can comprehensively consider key parameters and objectives, automatically generate the optimal control strategy, which is more scientific and efficient than the strategy formulated by human experience, and can realize multi-objective optimized operation of the power grid, ensuring the safe, economical and stable operation of the power grid.

[0136] This invention provides a method for simulating the dynamic characteristics of a new energy power grid. After sequentially inputting all parameter estimates into a first characteristic analysis model, the method further includes:

[0137] Obtain the work log of the first feature analysis model, and match each work item in the work log with the corresponding item traceability list to obtain the parameter description of the corresponding work item;

[0138] Retrieve a capture tool that matches the parameter description from the description-tool database, and capture the working path that matches the parameter description and generate path data based on the working path in real time according to the capture tool;

[0139] The generated data from the path is subjected to indicator feature extraction to obtain a generation vector;

[0140] The generated vector is input into the vector analysis model to obtain the early warning regulations, which include: early warning type and early warning information;

[0141] The warning regulations for each work item are compared and analyzed with the standard regulations to identify work anomaly factors;

[0142] The rationality of the work is analyzed by combining abnormal factors and parameter combinations;

[0143] If the reasonableness of the work is greater than the preset reasonableness, the working process of the first characteristic analysis model is determined to be reasonable;

[0144] Otherwise, based on the anomaly factors, parameter combinations, and working path, determine the reasons for optimizing the first characteristic analysis model.

[0145] In this embodiment, the work log is a document that records in detail the work performed by the first feature analysis model during its operation. It includes a series of work items, recording information such as the start time, end time, executor, and execution process for each work item. For example, in a software development project, the work log for the first feature analysis model for code performance analysis would record the specific details of each code scan, performance test, and other work items.

[0146] A work item is a basic unit in the work log, referring to a specific task or operation during the execution of the primary characteristic analysis model. Tasks such as data acquisition, data cleaning, and model training can all be considered work items.

[0147] An item traceability list is a pre-built list that stores standard information, historical data, references, etc., related to various work items, used to trace and match detailed parameters of work items. For example, in a logistics transportation characteristic analysis model, the item traceability list records standard transportation time, cost, and other parameters corresponding to different transportation routes and means of transportation.

[0148] The parameter description is a detailed description of the characteristics and attributes of a work item obtained by matching the work item with the item traceability list. For example, for a data acquisition work item, the parameter description may include the data type collected, the collection frequency, and the data source.

[0149] The description-tools database stores various parameter descriptions and their corresponding acquisition tools. Acquisition tools are software, programs, or devices used to acquire specific data or information. For example, in meteorological data analysis models, the description-tools database records the associations between temperature data parameter descriptions and corresponding meteorological monitoring equipment and data acquisition software.

[0150] Capture tools: Tools used to acquire data or information that matches the parameter descriptions in real time. For example, network traffic monitoring tools can be used as capture tools to capture network traffic data in real time, meeting the parameter description requirements of relevant work items in network characteristic analysis models.

[0151] The work path is the sequence of processes or steps followed when work items are executed during the operation of the first characteristic analysis model.

[0152] Path-generated data refers to the data generated by work items during the execution of a work path. For example, in the work path of an image recognition model, the intermediate data generated by work items such as image preprocessing and feature extraction, as well as the final recognition results, all belong to path-generated data.

[0153] Feature extraction is the process of extracting representative metrics from path-generated data that reflect the essential characteristics and patterns of the data. For example, in user behavior analysis models, features such as browsing time, number of clicks, and number of page jumps are extracted from user webpage browsing data.

[0154] The generated vector is vector data formed by quantifying and combining the extracted indicator features.

[0155] Vector analysis models are models that utilize mathematical algorithms and machine learning techniques to analyze and process generated vectors. They can predict, judge, or output relevant results based on the input generated vectors. For example, in risk prediction models, vector analysis models generate vectors based on input corporate financial data and output the company's risk level.

[0156] The early warning regulations are output by the vector analysis model and include rules or prompts containing early warning types and early warning information. Early warning types include data anomalies, performance degradation, and increased risk; early warning information is a detailed description of the early warning type, such as "Sales data in a certain region has been below 70% of the normal level for three consecutive days, indicating a risk of declining sales."

[0157] Standards and specifications: Pre-defined standards and specifications used to measure whether the primary characteristic analysis model is functioning correctly. For example, in a product quality inspection model, standards and specifications stipulate the product dimensional error range, performance index standards, etc.

[0158] Work anomaly factors are identified by comparing and analyzing the warning criteria for each work item with the standard criteria. These factors determine the causes of work anomalies. For example, in a manufacturing model, substandard raw material quality or equipment malfunction could be work anomaly factors.

[0159] A parameter combination is a collection of multiple parameters related to a work item. For example, in a flight scheduling model, parameters such as flight departure time, arrival time, aircraft type, and passenger capacity constitute a parameter combination.

[0160] The rationality of the operation is assessed by combining abnormal factors and parameter combinations to evaluate the degree to which the operation of the first characteristic analysis model conforms to expectations and standards.

[0161] The preset rationality is a threshold or standard used to judge whether the working process of the first characteristic analysis model is reasonable. When the rationality of the working process is greater than the preset value, the working process is considered reasonable; otherwise, it is unreasonable.

[0162] The optimization reason is that when the reasonableness of the work is less than the preset reasonableness, the reasons for needing to optimize and improve the first characteristic analysis model are derived from the analysis of abnormal factors, parameter combinations, and work paths. For example, unreasonable model algorithm, data input errors, and inefficient workflow.

[0163] The beneficial effects of the above technical solution are: comprehensive monitoring, analysis, and evaluation of the first characteristic analysis model's working process, enabling timely detection of anomalies and problems during operation. Through analysis of work logs, data capture, and indicator extraction, combined with vector analysis and comparative standards, abnormal factors are accurately identified, thereby assessing the rationality of the work. If the work is unreasonable, the reasons for optimization can be clearly identified, providing a basis for model optimization and improvement. Its effect is to ensure the efficient and accurate operation of the first characteristic analysis model, improve work quality and efficiency, reduce the probability of errors and risks, and enable the model to better serve business needs and decision-making.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for simulating the dynamic characteristics of a new energy power grid, characterized in that, include: Collect detailed electric field information of new energy power plants, establish the operation model of each power grid unit in simulation software, and connect them according to the actual topology of the new energy power grid to form a new energy power grid model; The new energy power grid model is decomposed into multiple sub-models according to pre-set rules and assigned to different CPU simulation cores for simulation. The adaptive step size algorithm is enabled to adjust the simulation step size in real time according to the state changes of the CPU simulation core, and the dynamic characteristics of the corresponding sub-model are obtained after each adjustment of the simulation step size. The actual grid operation data of the new energy grid is acquired in real time, and the actual grid operation data is preprocessed using the Kalman filter algorithm to obtain the parameter estimation set of each grid unit under the new energy grid model. The dynamic characteristics of the corresponding sub-model are analyzed based on the parameter estimation set of each sub-model, and the control strategy is determined according to the power grid operation control objectives. Among them, the adaptive step-size algorithm is enabled to adjust the simulation step-size in real time according to the state changes of the CPU simulation core, including: A monitoring mechanism is activated for each CPU emulation core to perform real-time monitoring. The real-time monitoring data is related to CPU emulation core utilization, memory usage, cache hit rate, and task queue length. The real-time monitoring data is input into the pre-established CPU simulation core state evaluation model to obtain the state evaluation coefficients at the corresponding time and determine the first error adjustment step size at N consecutive time points. The second error adjustment step size is determined based on the dynamic changes of the corresponding CPU simulation core over N consecutive time periods. When the first error adjustment step size and the second error adjustment step size are in the same direction, the first unit step size adjustment amount is determined; Where D1 is the first unit step size adjustment; These represent the first error adjustment step size and the second error adjustment step size, respectively. When the directions of the first error adjustment step size and the second error adjustment step size are not consistent, calculate the absolute value of the difference between the first error adjustment step size and the second error adjustment step size. At the same time, determine the data compliance ratio of the corresponding CPU simulation core at N consecutive time points. ,in, This represents the sum of parameters of the corresponding CPU simulation core within a set range over N consecutive time points, and ; The correction value Xz is obtained by multiplying the absolute value of the difference by the value obtained by subtracting the data compliance ratio from 1. Based on the correction value, the first error adjustment step size, and the second error adjustment step size, determine the second unit step size adjustment amount; The step size is adjusted based on the state change coefficient between the previous time step and the current time step, and based on the determined unit step size adjustment amount of the CPU simulation core. The update period of the unit step size adjustment amount is T, and the number of consecutive time steps involved in the update period T is greater than or equal to N.

2. The simulation method for the dynamic characteristics of new energy power grids according to claim 1, characterized in that, The detailed information about the electric field includes: the model, installation location, and operating conditions of each power grid unit.

3. The simulation method for the dynamic characteristics of new energy power grids according to claim 1, characterized in that, The formation of a new energy power grid model includes: The execution function and execution scale are determined based on the operating conditions of the power grid unit, and the design level is determined in combination with the connection relationship under the installation location of the power grid unit, wherein the connection relationship includes existing interconnection and parallel relationships; Design the model structure based on the aforementioned design hierarchy and the type of power grid source; Various physical phenomena that match the model of the power grid unit are obtained from the model-physical model lookup table, and then coupled and applied to the corresponding model structure to obtain the operating model. The information interaction mechanism between different design levels is determined, and the operation models under different design levels are connected according to the actual topology of the new energy power grid to obtain the new energy power grid model.

4. The simulation method for the dynamic characteristics of new energy power grids according to claim 1, characterized in that, Based on the correction value, the first error adjustment step size, and the second error adjustment step size, the second unit step size adjustment amount is determined, including: in, This indicates the second unit step size adjustment; , It is a symbolic function, and , .

5. The simulation method for the dynamic characteristics of new energy power grids according to claim 1, characterized in that, Preprocessing of actual power grid operation data using the Kalman filter algorithm includes: The actual power grid operation data is converted into a data sequence; The historical operation data of the new energy power grid is obtained, and the compliance coefficient and expected coefficient of each historical data group are analyzed. The process noise covariance matrix and measurement noise covariance matrix are adjusted, and the Kalman filter algorithm is updated. The updated algorithm is used to filter noise from the data sequence to obtain a new sequence. The parameter estimation set of the corresponding power grid unit is obtained based on the new sequence at M consecutive time points.

6. The simulation method for the dynamic characteristics of new energy power grids according to claim 1, characterized in that, The dynamic characteristics of each sub-model are analyzed based on the parameter estimates of all sub-models, including: For each parameter estimation set under the sub-model, perform in-set global analysis and out-of-set global analysis to obtain in-set error and out-of-set error; Minimize the error of all in-set errors under the sub-model and compare it with the out-of-set errors to obtain the reference error for each parameter; All parameter estimates are sequentially input into the first characteristic analysis model to obtain the nonlinear mapping relationship between each parameter and the dynamic characteristics of the corresponding sub-model; Sensitivity analysis is performed on each parameter in the parameter estimation set, and the influence of each parameter on the dynamic characteristics is quantified by combining nonlinear mapping relationships. Furthermore, the key parameter combinations that affect the changes in dynamic characteristics are identified by combining reference errors. By combining the power grid topology, the dynamic characteristic correlation between sub-models at different locations is analyzed, the combination of key parameters is verified, and parameter weights are assigned to each key parameter.

7. The simulation method for the dynamic characteristics of new energy power grids according to claim 6, characterized in that, Based on the power grid operation and control objectives, determine the control strategies, including: The power grid operation control objectives are described using quantitative indicators; The description results, key parameters, and parameter weights are combined into an input vector and input into the strategy analysis model to obtain the control strategy.

8. The simulation method for the dynamic characteristics of new energy power grids according to claim 6, characterized in that, After inputting all parameter estimates into the first characteristic analysis model in sequence, the following is also included: Obtain the work log of the first feature analysis model, and match each work item in the work log with the corresponding item traceability list to obtain the parameter description of the corresponding work item; Retrieve a capture tool that matches the parameter description from the description-tool database, and capture the working path that matches the parameter description and generate path data based on the working path in real time according to the capture tool; The generated data from the path is subjected to indicator feature extraction to obtain a generation vector; The generated vector is input into the vector analysis model to obtain the early warning regulations, which include: early warning type and early warning information; The warning regulations for each work item are compared and analyzed with the standard regulations to identify work anomaly factors; The rationality of the work is analyzed by combining abnormal factors and parameter combinations; If the reasonableness of the work is greater than the preset reasonableness, the working process of the first characteristic analysis model is determined to be reasonable; Otherwise, based on the anomaly factors, parameter combinations, and working path, determine the reasons for optimizing the first characteristic analysis model.

Images