A source network load storage integrated feature modeling method, system, device and medium

Abstract

This invention discloses an integrated characteristic modeling method, system, equipment, and medium for power system modeling, grid, load, and storage, belonging to the field of power system modeling and simulation analysis technology. It includes generating a cluster of wind and solar power output curves based on meteorological reanalysis data to form a power output characteristic vector; analyzing air conditioning thermal inertia, electric vehicle participation willingness, and industrial process flexibility to form a load flexibility characteristic vector and source-load correlation data; constructing net load based on the source-load correlation data to form an energy storage operation characteristic vector and source-load-storage correlation data; determining DC reverse power, virtual impedance, and tie-line dynamic transmission margin to form a grid response characteristic vector; and generating a multi-dimensional characteristic database by spatiotemporally and structurally associating the source-load-storage correlation data and grid response characteristic vector. This invention introduces a measured deviation feedback correction mechanism across four time scales: annual planning, monthly assessment, day-ahead scheduling, and real-time verification, improving the proactive adjustment capability and simulation assessment accuracy for source-load fluctuations.

Description

Technical Field

[0001] This invention relates to the field of power system modeling and simulation analysis technology, specifically to a method, system, equipment, and medium for integrated source-grid-load-storage characteristic modeling. Background Technology

[0002] With a high proportion of renewable energy being integrated into provincial power systems, problems such as power output fluctuations, dispersed load-side flexibility resources, and bidirectional changes in grid power flow are becoming increasingly prominent. Existing modeling methods typically establish separate models for power sources, loads, and energy storage, lacking a unified feature representation and cross-link coupling mechanism. Furthermore, they are often limited to a single time scale, making it difficult to support the rolling connection between annual planning, monthly assessments, day-ahead scheduling, and real-time verification.

[0003] Existing methods tend to oversimplify the characterization of energy storage health status, temperature constraints, electric vehicle participation willingness, and industrial load regulation capabilities. They also fail to fully consider the power flow regulation capabilities and dynamic cross-sectional transmission margins of AC / DC hybrid power grids, making it difficult for the source-grid-load-storage collaborative analysis results to accurately reflect the actual operating status of the system. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention provides a method, system, equipment and medium for integrated source-grid-load-storage feature modeling.

[0005] Therefore, the technical problem solved by this invention is: how existing methods can simplify the characterization of energy storage health status, temperature constraints, electric vehicle participation willingness and industrial load regulation capability, and do not fully consider the power flow regulation capability and dynamic cross-sectional transmission margin of AC / DC hybrid power grids, resulting in the source-grid-load-storage collaborative analysis results failing to accurately reflect the actual operating status of the system.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a source-grid-load-storage integrated feature modeling method, comprising: generating a cluster of wind and solar power output curves based on meteorological reanalysis data; identifying risk zones according to ramp rate and extreme output duration to form a power output feature vector; analyzing air conditioning thermal inertia, electric vehicle participation willingness, and industrial process flexibility to form a load flexibility feature vector and source-load correlation data; constructing net load based on source-load correlation data; adjusting the filter window according to risk zones to determine the energy storage target power and reserve capacity; adjusting the charging and discharging efficiency and minimum allowable state of charge according to health status and ambient temperature to form an energy storage operation feature vector and source-load-storage correlation data; calculating power flow based on source-load-storage correlation data; determining DC reverse power, virtual impedance, and tie-line dynamic transmission margin to form a grid response feature vector; and constructing a collaborative evolution feature matrix by spatiotemporally and structurally associating source-load-storage correlation data and grid response feature vector, establishing multi-timescale state transmission and deviation feedback relationships, and generating a multi-dimensional feature database.

[0007] As a preferred embodiment of the integrated source-grid-load-storage characteristic modeling method of the present invention, the step of generating a cluster of wind and solar power output curves based on meteorological reanalysis data, identifying risk intervals according to the ramp rate and the duration of extreme power output, and forming a power output feature vector includes: fitting the probability distribution of wind resources and the probability distribution of solar resources based on meteorological reanalysis data to generate wind resource sequences and solar resource sequences with uniform time resolution; converting the wind resource sequences and solar resource sequences into wind power output sequences and photovoltaic power output sequences, and organizing them into a cluster of wind and solar power output curves according to the time continuity relationship; calculating the ramp rate of the wind and solar power output curve cluster using a sliding window, and determining the ramp risk conditions based on the ramp rate distribution; statistically analyzing the duration of continuous periods of extreme power output, identifying risk intervals based on the dual criteria of ramp risk conditions and extreme power output duration conditions, marking the risk intervals, and forming a power output feature vector.

[0008] This preferred solution, through the joint identification of climbing risk and the duration of extreme power output, enables the determination of the fluctuation range of wind and solar power output to simultaneously possess amplitude and persistence characteristics.

[0009] As a preferred embodiment of the integrated source-grid-load-storage characteristic modeling method described in this invention, the analysis of air conditioning thermal inertia, electric vehicle participation willingness, and industrial process flexibility to form a load flexibility feature vector and source-load correlation data includes: determining the interruptible time window of the air conditioning load based on the indoor temperature change relationship and temperature deviation tolerance, and determining the air conditioning load reduction power based on the interruptible time window; determining the participation willingness based on the historical response data of electric vehicle users, and using the participation willingness to weight and aggregate the technically dispatchable power of electric vehicles to obtain the aggregated dispatchable power of electric vehicles and the dispatch uncertainty; distinguishing rigid tasks and flexible tasks according to the industrial process chain, and determining the industrial load shifting period and shifting power based on the process connection relationship of flexible tasks; organizing the air conditioning load reduction power, the aggregated dispatchable power of electric vehicles, and the industrial load shifting power to form a load flexibility feature vector, and associating the load flexibility feature vector with the power output feature vector according to time tags and spatial nodes to form source-load correlation data.

[0010] This preferred solution enables distributed load resources to form associative source-load characteristics by uniformly characterizing the regulation capabilities of air conditioning, electric vehicles, and industrial loads, based on time tags and spatial nodes.

[0011] As a preferred embodiment of the integrated source-grid-load-storage feature modeling method described in this invention, the following steps are included: constructing net load based on source-load correlation data, adjusting the filtering window according to risk intervals, determining the target power and reserve capacity of energy storage, and adjusting the charging and discharging efficiency and minimum allowable state of charge according to health status and ambient temperature to form an energy storage operation feature vector and source-load-storage correlation data. This includes: processing the source-load correlation data to form a net load considering load flexibility; configuring the filtering window according to risk intervals to perform frequency band decomposition on the net load, obtaining the low-frequency component and high-frequency component of the net load; determining the target power of energy storage by combining the high-frequency component of the net load with the duration of extreme output, and determining the reserve capacity by combining the aggregated dispatchable power of electric vehicles with dispatch uncertainty; embedding the health status into the charging and discharging efficiency function, dynamically setting the minimum allowable state of charge according to ambient temperature, and switching the energy storage operation mode according to the priority of dispatch instructions under the constraints of state of charge boundary and reserve capacity to form an energy storage operation feature vector; and associating the energy storage operation feature vector with the source-load correlation data according to time tags and spatial nodes to form source-load-storage correlation data.

[0012] This preferred scheme selects the filter window by driving the risk range, and combines the health status and ambient temperature constraints to keep the energy storage target power, reserve capacity and operating boundary consistent.

[0013] As a preferred embodiment of the integrated source-grid-load-storage characteristic modeling method of the present invention, the step of calculating power flow based on source-load-storage correlation data, determining DC reverse power, virtual impedance, and tie-line dynamic transmission margin, and forming a grid response characteristic vector includes: extracting power output, load flexibility dispatch results, and energy storage operation status from source-load-storage correlation data, and forming node injection power according to spatial nodes; substituting the node injection power into the power flow calculation formula to obtain node voltage, node phase angle, and key section power flow; substituting the total net load fluctuation and the maximum allowable power change rate of DC lines into the DC reverse power calculation formula to obtain DC reverse power; substituting the risk range and electric vehicle dispatch uncertainty into the virtual impedance calculation formula to obtain virtual impedance; and substituting the power output fluctuation, energy storage response power, and tie-line operating limits into the dynamic section transmission margin calculation formula to obtain the tie-line dynamic section transmission margin, thus forming a grid response characteristic vector.

[0014] This preferred scheme incorporates DC power flow reversal, virtual impedance, and tie-line dynamic transmission margin into power flow calculation, enabling the grid response characteristics to reflect the regulation capability under source-load-storage fluctuations.

[0015] As a preferred embodiment of the integrated source-grid-load-storage feature modeling method described in this invention, the following steps are included: Associating source-grid-storage related data and grid response feature vectors according to spatiotemporal and link characteristics to construct a collaborative evolution feature matrix, establish multi-timescale state transmission and deviation feedback relationships, and generate a multi-dimensional feature database. This includes: configuring time tags, spatial node identifiers, and link type identifiers for source-grid-storage related data and grid response feature vectors; extracting source-grid-storage related data and grid response feature vectors from the same time segment according to the time tags, and aligning various features under the same spatial node according to the spatial node identifiers; organizing power output characteristics, load flexibility characteristics, energy storage operation characteristics, and grid response characteristics according to the link type identifiers to construct a collaborative evolution feature matrix; configuring unified feature codes and data dictionaries for annual planning, monthly assessment, day-ahead scheduling, and real-time verification, establishing a multi-timescale rolling connection relationship for upper-level output transmission to lower-level and lower-level deviation feedback to upper-level; and arranging the collaborative evolution feature matrices of each time segment according to the basic time step to generate a multi-dimensional feature database.

[0016] This preferred scheme constructs a collaborative evolution feature matrix by linking time, space, and links, enabling source-grid-load-storage features to be invoked across time scales under a unified encoding.

[0017] As a preferred embodiment of the source-grid-load-storage integrated feature modeling method described in this invention, the establishment of a multi-timescale rolling connection relationship between upper-level output and lower-level deviation feedback includes: using the annual planning output as the initial state for monthly assessment, the monthly assessment output as the initial state for day-ahead scheduling, and the day-ahead scheduling output as the initial state for real-time verification; acquiring real-time verification observations and day-ahead scheduling prediction observations to generate real-time verification observation deviations; using a Kalman gain matrix to process the real-time verification observation deviations and correct the day-ahead scheduling state; feeding back the state deviations formed in the next time scale to the previous time scale to correct the processing parameters of the previous time scale; and updating the co-evolution feature matrix based on the corrected state and processing parameters.

[0018] This preferred scheme enables real-time verification results to correct day-ahead scheduling, monthly assessment, and annual planning parameters through hierarchical status transmission and deviation feedback.

[0019] This invention provides an integrated feature modeling system for source-grid-load-storage systems.

[0020] To address the aforementioned technical problems, this invention provides the following technical solution: an integrated feature modeling system for power generation, grid, load, and energy storage, comprising: a power source feature construction module, a load feature construction module, an energy storage feature construction module, a grid response construction module, and a collaborative evolution construction module; the power source feature construction module is used to generate a cluster of wind and solar power output curves based on meteorological reanalysis data, identify risk intervals according to the ramp rate and extreme output duration, and form a power source output feature vector; the load feature construction module is used to analyze air conditioning thermal inertia, electric vehicle participation willingness, and industrial process flexibility, forming a load flexibility feature vector and source-load correlation data; the energy storage feature module is used to analyze source-load correlation data... The system constructs net load data, adjusts the filter window according to risk intervals, determines the target power and reserve capacity of energy storage, and adjusts the charging and discharging efficiency and minimum allowable state of charge based on health status and ambient temperature, forming an energy storage operation feature vector and source-load-storage correlation data. The grid response construction module is used to calculate power flow based on the source-load-storage correlation data, determine DC reverse power, virtual impedance, and tie-line dynamic transmission margin, forming a grid response feature vector. The collaborative evolution construction module is used to construct a collaborative evolution feature matrix by spatiotemporally and link-wise associating source-load-storage correlation data and grid response feature vector, establishing multi-timescale state transmission and deviation feedback relationships, and generating a multi-dimensional feature database.

[0021] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the integrated source-grid-load-storage feature modeling method.

[0022] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the integrated source-grid-load-storage feature modeling method.

[0023] The beneficial effects of this invention are as follows: This invention effectively solves the problem of inaccurate assessment of the adaptability of high-proportion renewable energy access in power systems due to the fragmented models of the power source, grid, load, and storage links, the single time scale, and the coarse characterization of flexibility features. By constructing a progressive coupled modeling link of power output feature vector, load flexibility feature vector, energy storage operation feature vector, grid response feature vector, and co-evolution feature matrix, non-electrical physical quantities and behavioral quantities such as energy storage health state decay, temperature adaptive boundary, and user participation willingness are embedded in the models of each link in the form of explicit formulas. This achieves the collaborative expression of all links in the same data structure. Furthermore, a measured deviation feedback correction mechanism is introduced between the four time scales of annual planning, monthly assessment, day-ahead scheduling, and real-time verification, enabling the same modeling framework to have both long-term evolution analysis and short-term operation simulation capabilities, thereby improving the proactive adjustment capability and simulation evaluation accuracy of power source and load fluctuations. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 The above is a flowchart of an integrated source-grid-load-storage feature modeling method provided in one embodiment of the present invention.

[0026] Figure 2 This invention provides a load flexibility feature construction and source-load correlation diagram for an integrated source-grid-load-storage feature modeling method, as an embodiment of the present invention.

[0027] Figure 3 This invention provides an embodiment of a source-grid-load-storage integrated feature modeling method for energy storage operation feature construction and correlation diagram.

[0028] Figure 4 The energy storage capacity curve of energy storage power station A is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0029] Figure 5 The energy storage capacity curve of energy storage power station B is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0030] Figure 6 The energy storage capacity curve of energy storage power station C is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0031] Figure 7 The energy storage power curve of an energy storage power station D is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0032] Figure 8 The energy storage capacity curve of an energy storage power station E is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0033] Figure 9 The energy storage power curve of energy storage power station A is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0034] Figure 10 The energy storage power curve of energy storage power station B is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0035] Figure 11The energy storage power curve of an energy storage power station C is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0036] Figure 12 The energy storage power curve of an energy storage power station D is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0037] Figure 13 The energy storage power curve of an energy storage power station E is shown in an embodiment of the present invention, which provides a feature modeling method for integrated source-grid-load-storage.

[0038] Figure 14 This is a comparison chart of total system load and energy storage capacity provided by a source-grid-load-storage integrated feature modeling method according to an embodiment of the present invention. Detailed Implementation

[0039] To make the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0040] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a source-grid-load-storage integrated feature modeling method, including: S1. Generate a cluster of wind and solar power output curves based on meteorological reanalysis data, identify risk zones according to the slope rate and extreme power output duration, and form a power output feature vector.

[0041] S2. Analyze the thermal inertia of air conditioning, the willingness of electric vehicles to participate, and the flexibility of industrial processes to form a load flexibility feature vector and source-load correlation data.

[0042] S3. Construct net load based on source-load correlation data, adjust the filter window according to risk interval, determine the energy storage target power and reserve capacity, and adjust the charging and discharging efficiency and minimum allowable state of charge according to health status and ambient temperature to form energy storage operation characteristic vector and source-load-storage correlation data.

[0043] S4. Calculate power flow based on source-load-storage correlation data, determine DC reverse power, virtual impedance and tie-line dynamic transmission margin, and form a grid response characteristic vector.

[0044] S5. Based on the spatiotemporal and link-related source-load-storage correlation data and power grid response feature vector, construct a collaborative evolution feature matrix, establish multi-timescale state transmission and deviation feedback relationships, and generate a multi-dimensional feature database.

[0045] It should be noted that the operating characteristics of the power generation, grid, load, and storage components differ. Power output is affected by meteorological conditions, load regulation capabilities vary, energy storage operation is constrained by its health status and ambient temperature, and grid power flow also changes with the status of the power generation, load, and storage. When each component is modeled independently, it is difficult to form a unified characteristic representation and cross-timescale correlation.

[0046] Therefore, in this embodiment, the power output characteristic vector, source-load correlation data, source-load-storage correlation data and grid response characteristic vector are formed sequentially through S1-S4. Then, the spatiotemporal and link correlation is completed through S5 to establish the multi-timescale state transmission and deviation feedback relationship and generate a multi-dimensional characteristic database integrating source, grid, load and storage.

[0047] Example 2, refer to Figure 2 and Figure 3 As an embodiment of the present invention, based on the previous embodiment, a source-grid-load-storage integrated feature modeling method is provided, including: Furthermore, in step S1, a cluster of wind and solar power output curves is generated based on meteorological reanalysis data. Risk zones are identified according to the gradient rate and the duration of extreme power output, forming a power output feature vector, including the following steps A1-A4: A1. Based on meteorological reanalysis data, fit the probability distributions of wind resources and solar resources to generate wind resource sequences and solar resource sequences with uniform time resolution.

[0048] A2. Convert the wind resource sequence and solar resource sequence into wind power output sequence and photovoltaic power output sequence, and organize them into a cluster of wind and solar power output curves according to the time continuity relationship.

[0049] A3. The slope rate of the wind and solar power output curve cluster is calculated using a sliding window, and the slope risk conditions are determined based on the slope rate distribution.

[0050] A4. Statistically determine the duration of continuous periods of extreme power output, identify risk intervals based on the dual criteria of ramp risk conditions and extreme power output duration conditions, mark the risk intervals, and form a power output feature vector.

[0051] Based on meteorological reanalysis data-driven mixture Gaussian distribution method, the probability density function of wind and solar resources is fitted, generating a typical power output curve cluster with a time resolution of 15 minutes and covering 8760 hours throughout the year. The risk range of rapid power fluctuation is automatically identified by sliding window ramp rate and dual criteria, and the power output feature vector containing volatility and ramp risk labels is output. Specifically, a Gaussian mixture distribution method driven by meteorological reanalysis data is used to fit the probability density functions of wind and solar resources: in, Let be the probability density function. For wind speed or irradiance, No. The mixture weights of Gaussian components satisfy the following conditions: , The first The mean and variance of each component, The number of Gaussian components. To sum the model parameters, the expectation-maximization algorithm was used to estimate the parameters from historical meteorological reanalysis data. This represents a normal distribution.

[0052] Generate a typical power output curve cluster with a time resolution of 15 minutes, covering 8760 hours throughout the year, with a sliding window width of [missing information]. Under the conditions, Climb rate at any time for: in, The first in the power output curve cluster The curves in The output power at any given moment.

[0053] The dual criteria for the risk range of rapid power change are: in, The threshold for the ramp rate is represented by the 95th quantile of the absolute value sequence of ramp rates, which is sorted in descending order. The 95th quantile is chosen because this quantile corresponds to an event that occurs approximately once every 18 days in a probabilistic sense. This is a low-probability and high-impact fluctuation scenario that requires close attention. To determine the extreme output threshold, an extreme low output is defined. When the installed capacity is below 10%, the power supply almost stops generating electricity, resulting in extremely high output. The power output is close to 90% of the installed capacity. These two limits are determined by the physical output range of the power supply itself, and the threshold for the duration of extreme output. The determination is based on all data in the statistical historical data that satisfy the following conditions. or The duration of consecutive time periods is used to form a duration histogram, and the 90th quantile is taken as the duration. .

[0054] When both conditions are met, the time period will be... The risk zone is marked as such. The duration of extreme output is used to form a climbing risk indicator function. .

[0055] Output power characteristic vector: in, The power output of the power supply. For the standard deviation of output, As a risk indicator for climbing, For geographic grid numbers.

[0056] Furthermore, in step S2, the thermal inertia of air conditioning, the willingness of electric vehicles to participate, and the flexibility of industrial processes are analyzed to form a load flexibility feature vector and source-load correlation data, including the following steps B1-B4: B1. Determine the interruptible time window of the air conditioning load based on the indoor temperature change relationship and temperature deviation tolerance, and determine the power that can be reduced for the air conditioning load according to the interruptible time window.

[0057] B2. Based on historical response data of electric vehicle users, determine the willingness to participate, and use the willingness to participate to weight and aggregate the schedulable power of electric vehicles to obtain the aggregated schedulable power of electric vehicles and the scheduling uncertainty.

[0058] B3. Differentiate between rigid and flexible tasks according to the industrial process chain, and determine the industrial load transfer period and transfer power based on the process connection relationship of flexible tasks.

[0059] B4. Organize the power that can be reduced by air conditioning load, the power that can be aggregated and dispatched by electric vehicles, and the power that can be shifted by industrial load to form a load flexibility feature vector. Associate the load flexibility feature vector with the power output feature vector according to time tags and spatial nodes to form source-load correlation data.

[0060] Load controllability analysis is performed. For air conditioning loads, interruptible time windows are derived based on thermal inertia and temperature deviation tolerance. For electric vehicle loads, participation willingness coefficients are generated based on historical user response data, and vehicle-to-grid (V2G) dispatchable power is weighted and aggregated. For industrial loads, flexible task shifting periods are identified according to the process chain. A comprehensive load flexibility feature vector is output. Figure 2 As shown, the process of extracting flexibility features of air conditioning load, electric vehicle and industrial load, and the correlation between load flexibility feature vector and power output feature vector is presented.

[0061] For the air conditioning load, a first-order thermodynamic equivalent model is used: in, The equivalent heat capacity of indoor air. Equivalent thermal resistance of building envelope. Indoor temperature, Outdoor ambient temperature The rated cooling / heating power of the air conditioner. It is in start / stop state.

[0062] User-set temperature and tolerance offset Down, Default ±2℃, window width when air conditioning can be interrupted in cooling mode. From room temperature - Change to The inverse solution of the thermal inertia equation yields: Air conditioner power reduction during outages Within that time period Calculate the average value.

[0063] For electric vehicle load, the first definition is... Dynamic participation willingness coefficient of individual users: in, This represents the total number of scheduling invitations for this user within the statistical period. This represents the actual number of times a user accepts a dispatch. The charging start time, target state of charge, and technically dispatchable power for each vehicle are generated through Monte Carlo simulation. The cluster is weighted and aggregated to obtain the expected value of V2G schedulable power. and the upper bound of the 95% confidence interval : in, For mathematical expectation, Represents a grid The total number of electric vehicles connected to the platform.

[0064] For industrial loads, tasks are divided into rigid and flexible categories according to the production process chain. The allowable translation power for flexible tasks is... ,exist Execution within hours.

[0065] Output load flexibility feature vector: in, This represents the unadjusted, original load active power.

[0066] in, For grid Total number of indoor air conditioners For the first air conditioner during the time period The average start-stop status within the time frame; if the air conditioning is interrupted during this period. ,otherwise Take the duty cycle during normal operation.

[0067] Furthermore, in step S3, the net load is constructed based on the source-load correlation data, the filter window is adjusted according to the risk interval, the target power and reserve capacity of energy storage are determined, and the charge-discharge efficiency and minimum allowable state of charge are adjusted according to the health status and ambient temperature to form the energy storage operation feature vector and source-load-storage correlation data, including the following steps C1-C5: C1. Process source-load related data to form a net load that takes into account load flexibility.

[0068] C2. Configure the filter window according to the risk interval, and perform frequency band decomposition on the net load to obtain the low-frequency component and high-frequency component of the net load.

[0069] C3. Determine the target power of energy storage by combining the high-frequency component of net load and the duration of extreme output, and determine the reserve capacity by combining the aggregated dispatchable power of electric vehicles and dispatch uncertainty.

[0070] C4. Embed the health status into the charge and discharge efficiency function, dynamically set the minimum allowable state of charge based on the ambient temperature, and switch the energy storage operation mode according to the priority of the dispatch instructions under the constraints of the state of charge boundary and the reserve capacity, forming an energy storage operation feature vector.

[0071] C5. Based on time tags and spatial nodes, associate the energy storage operation feature vector with the source-load associated data to form source-load-storage associated data.

[0072] Using the power output characteristic vector from step S1 and the load flexibility characteristic vector from step S2 as inputs, the distribution of net load fluctuation amplitude and duration is calculated to determine the power fluctuation frequency band and reserve requirements that the energy storage device needs to smooth. Based on this, an energy storage operation model is constructed, dynamically embedding the State of Health (SOH) into the charge / discharge efficiency function and linearly adjusting the minimum allowable state of charge according to ambient temperature. Simultaneously, a mode-switching state machine considering dispatch command priority is set up to output the energy storage operation characteristic vector. For example... Figure 3 As shown, the process of net load decomposition, determination of energy storage target power and reserve capacity, adjustment of energy storage operation constraints, and formation of source-load-storage correlation data is presented. The charge / discharge efficiency functions include a charging efficiency calculation function and a discharging efficiency calculation function. The charging efficiency calculation function is used to correct the baseline charging efficiency based on the health status of the energy storage battery, and the discharging efficiency calculation function is used to correct the baseline discharging efficiency based on the health status of the energy storage battery.

[0073] Specifically, from extract , , ;from extract , , , , , Construct the net load sequence after considering load flexibility call: The net load is decomposed into low-frequency components using a moving average filter. and high frequency components : Slide window width Depend on control: in, To achieve a fast response filtering window, the value is set to 4 time steps, i.e., 1 hour, with 15-minute steps. The standard filtering window is set to 12 time steps, or 3 hours.

[0074] The target power that energy storage needs to smooth out and Related: in, Maximum allowable net load fluctuation. Energy storage reserve capacity. Determined based on the width of the V2G confidence interval: in, A conservative coefficient was determined through backtesting based on historical dispatch deviations of electric vehicles and reserve capacity coverage requirements. Energy storage charging and discharging efficiency and health status. The attenuation relationship is as follows: in, , The charging and discharging efficiency is the factory-specified efficiency. The attenuation coefficient is determined by fitting test data of the energy storage battery's charge and discharge efficiency under different health conditions. For charging efficiency; This refers to the discharge efficiency.

[0075] Ambient temperature was obtained from the same meteorological data source. Dynamically adjust the minimum permissible state of charge: Real-time charging and discharging power of energy storage Pick And decomposed into charging power and discharge power The state of charge is recursively derived as follows: Operational constraints include And spare capacity must be reserved, that is, the following must be satisfied: in, Indicates the rated capacity of the energy storage system Output energy storage operation feature vector: Furthermore, in step S4, power flow is calculated based on source-load-storage correlation data to determine DC reverse power, virtual impedance, and tie-line dynamic transmission margin, forming a grid response characteristic vector, including the following steps D1-D5: D1. Extract power output, load flexibility call results and energy storage operation status from source-load-storage correlation data, and form node injection power according to spatial nodes.

[0076] D2. Substitute the injected power at the nodes into the power flow calculation formula to obtain the node voltage, node phase angle, and power flow at the critical section.

[0077] D3. Substitute the total net load fluctuation and the maximum allowable power change rate of the DC line into the DC reverse power calculation formula to obtain the DC reverse power.

[0078] D4. Substitute the risk interval and the uncertainty of electric vehicle scheduling into the virtual impedance calculation formula to obtain the virtual impedance.

[0079] D5. Substitute the power output fluctuation, energy storage response power, and tie line operating limits into the dynamic cross-sectional transmission margin calculation formula to obtain the tie line dynamic cross-sectional transmission margin, forming the grid response characteristic vector.

[0080] Using the power output characteristic vector, load flexibility characteristic vector, and energy storage operation characteristic vector output from steps S1, S2, and S3 as inputs, the system power flow distribution is calculated. In the power grid model, DC power flow rapid reversal, flexible AC transmission virtual impedance control, and inter-provincial tie line dynamic cross-sectional transmission margin are configured, and the power grid response characteristic vector associated with the source-load-storage characteristics is output.

[0081] Specifically, from Pick , , ;from Pick , , , ;from Pick , , , Define the active power of a node with rigid loads: Assuming reactive power and active power remain constant, the power factor angle is... ,but , Reactive power storage Determined by the converter control range, it is simplified to zero here. The node injected power is: in, The power factor angle on the power supply side is taken as follows: if no reactive power regulator is installed, then... Delay, The load-side power factor angle is taken as follows: Lag. The above power factor can be adjusted based on actual operating data, reactive power compensation status, and the reactive power regulation capability of the equipment.

[0082] Solving for voltage amplitude using power flow equations and phase angle : in, This is the active power deviation vector at the nodes. This is the reactive power deviation vector at the nodes. Jacobian matrix calculated for power flow. This is the vector of nodal phase angle corrections. This is the vector for the correction of node voltage magnitude.

[0083] line - The meritorious trend: in, For the line - The meritorious trend, For nodes voltage amplitude, For nodes voltage amplitude, For the line - The conductivity component corresponding to admittance, For nodes With nodes The voltage phase angle difference between them For the line - The susceptance component corresponding to the admittance.

[0084] Perform DC power flow fast reversal control and define the total net load fluctuation of the system. When its absolute value exceeds the threshold and the line is a DC line, it is adjusted according to the maximum allowable rate of change within 200ms. Adjust the target power: in, for The total net load fluctuation of the system at any given moment; for Time-based geographic grid Net load; To control the elapsed time after triggering At that time, DC line - Target power; This is a sign function used to determine the direction of fluctuations in the system's total net load. This represents the maximum permissible power change rate of a DC line, taken as 20% / second of the line's rated power. The value is determined based on the DC converter equipment's permissible continuous power change capability and the grid frequency and voltage withstand capability, and is used to prevent excessively rapid changes in power commands.

[0085] Virtual impedance control for flexible AC transmission, virtual impedance according to of and V2G confidence dynamically adjusted: in, The reference virtual impedance is taken as 10% of the impedance magnitude of the controlled line, so that the virtual impedance can provide a certain damping, while avoiding excessive virtual impedance causing line voltage drop and power flow deviation. Indicates the impedance increment in the climbing risk scenario. ; The impedance increment under load uncertainty is taken as .

[0086] Inter-provincial connecting lines dynamic cross-sectional transmission margin, defined This is the maximum response power that energy storage can provide when switching from the current mode to the frequency regulation support mode, and its value is based on... SOC, Mode, and rated power calculations: in, This is an indicative function. When the energy storage is in black-start standby mode, it does not participate in frequency regulation response. Similarly, if the energy storage is not currently in black-start standby mode, the value of this function is 1. Equal to the front The calculated value indicates that energy storage can provide fast-response power.

[0087] Indicates the rated power of the energy storage converter. This indicates the required frequency regulation response time for energy storage, which is set to 2 seconds. It is determined based on the power response capability of the energy storage converter and the response time requirement of the frequency regulation command, so that the energy storage can complete the response during second-level power fluctuations.

[0088] The dynamic stability limit is: in, The tie line stability limit under rated operating conditions is determined by offline stability simulation. The sensitivity coefficient of power supply fluctuation to the stability limit is set to 0.05. It is determined by performing offline stability simulation by changing the standard deviation of power supply output fluctuation, and fitting the relationship between power supply fluctuation and tie line stability limit. It represents the degree to which power supply fluctuation weakens the stability limit. The factor representing the improvement of the energy storage response on the stability limit is set to 0.8. This factor is determined by conducting offline stability simulations by changing the energy storage response power and fitting the relationship between the energy storage response and the tie-line stability limit. The value of 0.8 indicates that after considering the impact of response delay, conversion loss, and connection location, 80% of the energy storage response power is included in the improvement of the stability limit. Take the mean of the standard deviation of the power supply for the entire system.

[0089] Cross-sectional transmission margin is updated every 15 minutes: in, For cross-sectional transmission margin, Indicates the thermal stability limit power of the tie line. Ground state power flow under the current topology and power generation plan.

[0090] Output power grid response feature vector: in, For the key cross-section of the tidal current, The DC reversal response time is set to 200ms.

[0091] Furthermore, in step S5, based on the spatiotemporal and link-related source-load-storage correlation data and grid response feature vector, a co-evolution feature matrix is ​​constructed to establish multi-timescale state transmission and deviation feedback relationships, generating a multi-dimensional feature database, including the following steps E1-E5: E1. Configure time tags, spatial node identifiers, and link type identifiers for source-load-storage correlation data and grid response feature vectors.

[0092] E2. Extract source-load-storage correlation data and grid response feature vectors for the same time segment according to time labels, and align various features under the same spatial node according to spatial node identifiers.

[0093] E3. Organize the power output characteristics, load flexibility characteristics, energy storage operation characteristics and grid response characteristics according to the link type, and construct a collaborative evolution characteristic matrix.

[0094] E4. Configure unified feature codes and data dictionaries for annual planning, monthly evaluation, day-ahead scheduling and real-time verification, and establish a multi-time-scale rolling connection relationship for the transmission of upper-level output to lower-level and the feedback of lower-level deviations to upper-level.

[0095] E5. Arrange the co-evolutionary feature matrices of each time segment according to the basic time step to generate a multi-dimensional feature database.

[0096] Furthermore, in step E4, a multi-timescale rolling connection relationship is established for the transmission of upper-layer output to lower-layer and the feedback of lower-layer deviations to upper-layer, including the following steps E41-E45: E41. Use the annual planning output as the initial state for monthly evaluation, the monthly evaluation output as the initial state for day-ahead scheduling, and the day-ahead scheduling output as the initial state for real-time verification.

[0097] E42. Obtain real-time verification observations and day-ahead scheduling prediction observations, and generate real-time verification observation bias.

[0098] E43. Use Kalman gain matrix to process and verify observation deviations in real time to correct day-ahead scheduling status.

[0099] E44. Feed back the state deviation formed in the next time scale to the previous time scale and correct the processing parameters of the previous time scale.

[0100] E45. Update the co-evolutionary feature matrix based on the corrected state and processing parameters.

[0101] The feature vectors generated in steps S1 to S4 at the same time segment are correlated and mapped according to time label, spatial node, and stage type to construct a co-evolution feature matrix. A multi-timescale rolling connection mechanism is set up, from annual planning, monthly assessment, day-ahead scheduling to real-time verification. Each level shares a unified feature code. A multi-dimensional feature database characterizing the multi-timescale collaborative characteristics of source-grid-load-storage is generated by using upper-level outputs as the initial state of lower levels and lower-level deviations as feedback to correct upper-level parameters. Specifically, annual planning, monthly assessment, day-ahead scheduling, and real-time verification are arranged in descending order of time scale, forming upper and lower layers, with adjacent longer time scales representing the upper layer and shorter time scales representing the lower layer. Upper-level outputs serve as the initial state of lower layers, and lower-level deviations are fed back to the upper layer to correct processing parameters.

[0102] The feature vectors from the same time section are aligned according to spatial nodes to form a co-evolutionary feature matrix: in, It is a set of all modeled nodes within the province.

[0103] A four-tiered time scale is established: annual planning (typical week), monthly evaluation (daily), day-ahead scheduling (1-hour rolling), and real-time verification (5-minute). Each tier shares a unified feature code and data dictionary. The feedback correction equation for day-ahead estimates based on real-time verification is as follows: in, For state vectors, For observation purposes, This is the Kalman gain matrix. The final multidimensional feature database is generated as follows: This database provides a standardized input interface for subsequent planning evaluation and operational optimization.

[0104] Example 3, referring to Figures 4-14 This invention provides a method for integrated source-grid-load-storage feature modeling, and to verify the beneficial effects of the invention, scientific demonstration is carried out through experiments.

[0105] A simulation example was constructed using a large-scale regional power grid. This example retains the multi-voltage level structure and complex topology of the actual power grid, including voltage levels of 500kV, 220kV, 110kV, 35kV, 10kV, and 6kV, with a total of 560 bus nodes and 756 branches. These include 219 transmission lines, 131 branches with two-winding transformers, 223 branches with equivalent three-winding transformers, and 183 switch branches, thus comprehensively reflecting the operational characteristics of lines, transformers, and switchgear in a large-scale power grid.

[0106] The power supply side includes 27 conventional generator units with a total installed capacity of 2520.86MW; the new energy side includes one wind farm and nine photovoltaic power stations, with a wind power installed capacity of 49.50MW and a photovoltaic installed capacity of 427.12MW, for a total new energy installed capacity of 476.62MW.

[0107] The load side uses a typical 24-hour daily operating curve, with each scheduling period lasting 1 hour, divided into 24 periods. Based on positive load statistics, the system's baseline active load is 1269.76 MW and reactive load is 247.33 Mvar. The load for each period is obtained by multiplying the nodal baseline load by a normalized load factor, with the load factor ranging from 0.36 to 1.00, peaking in the 21st period. Wind power and solar power output are determined based on their respective typical daily output factors, with the wind power output factor ranging from 0.30 to 1.00 and the solar power output factor ranging from 0 to 1.00, thus reflecting the temporal variation characteristics of load demand and renewable energy output within the day.

[0108] like Figure 4 As shown, energy storage station A has an initial capacity of 200 MWh, with upper and lower limits of 400 MWh and 40 MWh, respectively. During periods 0-2, the energy storage capacity remains at the initial level; subsequently, it is charged during periods of lower system load, reaching the upper limit of 400 MWh in period 4. During periods 11-12, energy storage station A discharges intensively, reducing its capacity from approximately 400 MWh to 40 MWh. During periods 14-15, as system regulation demands change, the energy storage is recharged to the upper limit; during periods 20-21, it discharges again to the minimum allowable capacity, recovering to 200 MWh before the end of the scheduling cycle. This energy storage station has a large capacity and power, primarily undertaking system-level peak shaving and valley filling, and power balance regulation tasks.

[0109] like Figure 5 As shown, the maximum capacity of energy storage station B is 5MWh, and the minimum and initial capacity are both 1.5MWh. During periods 0-3, the energy storage capacity gradually increases and reaches 5MWh; thereafter, it remains fully charged until period 10-11 when it discharges according to system power demand, reducing the capacity to 1.5MWh. During periods 14-15, the energy storage absorbs energy again and recovers to its upper capacity limit; during periods 20-21, it discharges again to its initial capacity and maintains it until the end of the scheduling cycle. This result indicates that energy storage station B mainly participates in local power regulation through two charge-discharge processes, while consistently meeting the upper and lower capacity limits and the capacity constraint at the end of the cycle.

[0110] like Figure 6As shown, energy storage station C has a maximum capacity of 5MWh, and a minimum and initial capacity of 0.5MWh. During periods 0-1, the energy storage is at its minimum capacity. During periods 2-4, it gradually charges, reaching 5MWh in period 4. During periods 11-12, the energy storage capacity rapidly decreases to 0.5MWh; during periods 15-16, it recharges to its maximum capacity; and during periods 20-21, it discharges again to the minimum allowable capacity and maintains this level until the end of the dispatch cycle. Compared to energy storage station B, energy storage station C has a larger range of available energy for adjustment, resulting in a higher depth of charge and discharge, enabling it to provide more sufficient energy support during periods of supply-demand imbalance at its node.

[0111] like Figure 7 As shown, the initial capacity of energy storage station D is 10MWh, the maximum capacity is 20MWh, and the minimum capacity is 1MWh. During periods 0-2, the energy storage capacity remains at the initial level; during periods 3-4, it is rapidly charged to approximately 20MWh and remains fully charged for several subsequent periods. During periods 10-11, the energy storage is concentratedly discharged to the minimum capacity; during periods 14-15, it is recharged to 20MWh; during periods 20-21, the second discharge process is completed, and it recovers to 10MWh during period 24. This operating curve exhibits a relatively obvious "two-charge, two-discharge" characteristic, indicating that energy storage station D can transfer electrical energy between different load phases during the day, playing a role in peak shaving and valley filling and alleviating power pressure on transmission lines.

[0112] like Figure 8 As shown, the initial capacity of energy storage station E is 4MWh, the maximum capacity is 8MWh, and the minimum capacity is 0.8MWh. During periods 0-2, the energy storage is charged from its initial state to 8MWh and remains fully charged until period 10; in period 11, it rapidly discharges to 0.8MWh. In period 15, the energy storage is recharged to its maximum capacity and then discharged again to its minimum capacity in period 20, finally recovering to 4MWh in period 24. The energy storage station exhibits rapid capacity changes, with clear characteristics of full charging and deep discharging, primarily used to respond to power shortages and renewable energy consumption demands at different times of the system. Throughout the entire dispatch process, the energy storage capacity did not exceed the allowable range and met the capacity recovery requirements at the end of the cycle.

[0113] like Figures 9-14As shown, the system load exhibits a clear peak-valley variation over 24 hours, with the lowest load at approximately 457.11 MW occurring in the third period and the highest load at approximately 1269.76 MW occurring in the 21st period. During periods 1-6, the system is in a low load phase, and energy storage stations utilize their remaining power capacity for charging, resulting in an overall upward trend in energy storage capacity. During periods 11-13, the system enters its first load peak, and energy storage capacity decreases significantly, indicating that energy storage provides power support to the system through discharge. During periods 14-15, as the system load decreases and photovoltaic output increases, energy storage stations recharge and recover to higher capacity levels. During periods 19-22, the system experiences a second load peak, with the load reaching its maximum for the day in the 21st period. Energy storage stations again discharge intensively, thereby reducing the power supply pressure on conventional power sources and transmission lines during peak hours. At the end of the dispatch cycle, the capacity of each energy storage station has recovered to its initial level, meeting the end-of-cycle capacity constraint. Because energy storage power station A has a significantly higher capacity than the other energy storage power stations, its power output change is the most pronounced in the graph, while the curves of the other energy storage stations are concentrated in the lower part of the graph due to the influence of a unified coordinate scale. Overall, each energy storage power station can coordinately adjust its charging and discharging state according to changes in system load and renewable energy output, realizing the transfer of electricity from low-load periods to high-load periods. This demonstrates the role of the proposed source-grid-load-storage coordinated optimization method in peak shaving and valley filling and improving the renewable energy absorption capacity.

[0114] Example 4 is an embodiment of the present invention. This embodiment provides an integrated feature modeling system for source-grid-load-storage, including a power source feature construction module, a load feature construction module, an energy storage feature construction module, a grid response construction module, and a collaborative evolution construction module.

[0115] The power characteristic construction module is used to generate a cluster of wind and solar power output curves based on meteorological reanalysis data, identify risk zones according to the slope rate and the duration of extreme power output, and form a power output characteristic vector.

[0116] The load characteristic construction module is used to analyze air conditioning thermal inertia, electric vehicle participation willingness, and industrial process flexibility to form load flexibility feature vectors and source-load correlation data.

[0117] The energy storage feature construction module is used to construct net load based on source-load correlation data, adjust the filter window according to the risk interval, determine the target power and reserve capacity of energy storage, and adjust the charging and discharging efficiency and minimum allowable state of charge according to the health status and ambient temperature, so as to form energy storage operation feature vector and source-load-storage correlation data.

[0118] The grid response construction module is used to calculate power flow based on source-load-storage correlation data, determine DC reverse power, virtual impedance and tie-line dynamic transmission margin, and form a grid response feature vector.

[0119] The co-evolution construction module is used to construct a co-evolution feature matrix by associating source-load-storage correlation data and power grid response feature vectors according to time, space and links, establishing multi-timescale state transmission and deviation feedback relationships, and generating a multi-dimensional feature database.

[0120] This embodiment also provides a computer device applicable to a source-grid-load-storage integrated feature modeling method, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the source-grid-load-storage integrated feature modeling method proposed in the above embodiment.

[0121] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements a source-grid-load-storage integrated feature modeling method as proposed in the above embodiment.

[0122] The storage medium proposed in this embodiment and the method for implementing a source-grid-load-storage integrated feature modeling method proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0123] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for integrated source-grid-load-storage characteristic modeling, characterized in that, include: Based on meteorological reanalysis data, a cluster of wind and solar power output curves is generated, and risk zones are identified according to the slope rate and the duration of extreme power output, forming a power output feature vector. Analyze the thermal inertia of air conditioning, the willingness of electric vehicles to participate, and the flexibility of industrial processes to form a load flexibility feature vector and source-load correlation data; Based on the source-load correlation data, the net load is constructed, the filter window is adjusted according to the risk interval, the target power and reserve capacity of energy storage are determined, and the charging and discharging efficiency and minimum allowable state of charge are adjusted according to the health status and ambient temperature to form the energy storage operation characteristic vector and source-load-storage correlation data. Based on the source-load-storage correlation data, the power flow is calculated, the DC reverse power, virtual impedance and tie line dynamic transmission margin are determined, and a grid response characteristic vector is formed. By constructing a collaborative evolution feature matrix based on spatiotemporal and link-related source-load-storage data and power grid response feature vectors, a multi-timescale state transmission and deviation feedback relationship is established, generating a multi-dimensional feature database.

2. The integrated source-grid-load-storage feature modeling method as described in claim 1, characterized in that, The generation of wind and solar power output curve clusters based on meteorological reanalysis data, the identification of risk zones according to the slope rate and the duration of extreme power output, and the formation of a power output feature vector include: Based on meteorological reanalysis data, wind resource probability distribution and solar resource probability distribution are fitted to generate wind resource sequence and solar resource sequence with uniform time resolution. The wind resource sequence and solar resource sequence are converted into wind power output sequence and photovoltaic power output sequence, and then organized into a cluster of wind and solar power output curves according to the time continuity relationship. The slope rate of the wind and solar power output curve cluster is calculated using a sliding window, and the slope risk conditions are determined based on the slope rate distribution. The duration of continuous periods of extreme power output is statistically analyzed. Risk intervals are identified based on the dual criteria of ramp risk conditions and extreme power output duration conditions. These risk intervals are then marked to form a power output feature vector.

3. The integrated source-grid-load-storage feature modeling method as described in claim 2, characterized in that, The analysis of air conditioning thermal inertia, electric vehicle participation willingness, and industrial process flexibility forms a load flexibility feature vector and source-load correlation data, including: Based on the indoor temperature change relationship and temperature deviation tolerance, determine the interruptible time window of the air conditioning load, and determine the power that can be reduced for the air conditioning load according to the interruptible time window; Based on historical response data of electric vehicle users, the willingness to participate is determined. The willingness to participate is used to weight and aggregate the schedulable power of electric vehicles to obtain the aggregated schedulable power of electric vehicles and the scheduling uncertainty. The industrial process chain is used to distinguish between rigid and flexible tasks. The time period and power of industrial load transfer are determined based on the process connection relationship of flexible tasks. The load flexibility feature vector is formed by organizing the power that can be reduced by air conditioning load, the power that can be aggregated and dispatched by electric vehicles, and the power that can be shifted by industrial load. The load flexibility feature vector is then associated with the power output feature vector according to time tags and spatial nodes to form source-load correlation data.

4. The integrated source-grid-load-storage feature modeling method as described in claim 3, characterized in that, The process involves constructing net load based on source-load correlation data, adjusting the filter window according to risk intervals, determining the energy storage target power and reserve capacity, and adjusting charge-discharge efficiency and minimum allowable state of charge based on health status and ambient temperature. This results in the formation of an energy storage operation characteristic vector and source-load-storage correlation data, including: Process source-load correlation data to form a net load that takes into account load flexibility deployment; Configure the filter window according to the risk interval, and decompose the net load into frequency bands to obtain the low-frequency component and high-frequency component of the net load. The target power of energy storage is determined by combining the high-frequency component of net load and the duration of extreme output, and the reserve capacity is determined by combining the aggregated dispatchable power of electric vehicles and dispatch uncertainty. The health status is embedded into the charge and discharge efficiency function, the minimum allowable state of charge is dynamically set according to the ambient temperature, and the energy storage operation mode is switched according to the priority of the dispatch command under the constraints of the state of charge boundary and the reserve capacity, forming an energy storage operation feature vector. By associating energy storage operation feature vectors with source-load data based on time tags and spatial nodes, source-load-storage correlation data is formed.

5. The integrated source-grid-load-storage feature modeling method as described in claim 4, characterized in that, The process of calculating power flow based on source-load-storage correlation data, determining DC reverse power, virtual impedance, and tie-line dynamic transmission margin, and forming a grid response characteristic vector includes: Extract power output, load flexibility dispatch results and energy storage operation status from source-load-storage correlation data, and form node injection power according to spatial nodes; Substituting the injected power at the nodes into the power flow calculation formula yields the node voltage, node phase angle, and power flow at the critical section. Substituting the total net load fluctuation and the maximum allowable power change rate of the DC line into the DC reverse power calculation formula, we obtain the DC reverse power. Substituting the risk range and the uncertainty of electric vehicle scheduling into the virtual impedance calculation formula, we obtain the virtual impedance; By substituting power output fluctuations, energy storage response power, and tie line operating limits into the dynamic cross-sectional transmission margin calculation formula, the dynamic cross-sectional transmission margin of the tie line is obtained, forming the grid response characteristic vector.

6. The integrated source-grid-load-storage feature modeling method as described in claim 5, characterized in that, The method involves associating source-load-storage data and grid response feature vectors according to spatiotemporal and link relationships to construct a collaborative evolution feature matrix, establish multi-timescale state transfer and deviation feedback relationships, and generate a multi-dimensional feature database, including: Configure time tags, spatial node identifiers, and link type identifiers for source-load-storage correlation data and grid response feature vectors; Extract source-load-storage correlation data and grid response feature vectors from the same time segment according to time tags, and align various features under the same spatial node according to spatial node identifiers; Organize power output characteristics, load flexibility characteristics, energy storage operation characteristics, and grid response characteristics according to the segment type identifier, and construct a collaborative evolution characteristic matrix; Configure unified feature codes and data dictionaries for annual planning, monthly evaluation, day-ahead scheduling and real-time verification, and establish a multi-time-scale rolling connection relationship for the transmission of upper-level output to lower-level and the feedback of lower-level deviations to upper-level; The co-evolutionary feature matrices of each time segment are arranged according to the basic time step to generate a multi-dimensional feature database.

7. The integrated source-grid-load-storage feature modeling method as described in claim 6, characterized in that, The establishment of a multi-timescale rolling connection relationship between upper-layer output and lower-layer deviation and upper-layer feedback includes: The annual planning output is used as the initial state for monthly evaluation, the monthly evaluation output is used as the initial state for day-ahead scheduling, and the day-ahead scheduling output is used as the initial state for real-time verification. Acquire real-time verification observations and day-ahead scheduling prediction observations, and generate real-time verification observation biases; Kalman gain matrix is ​​used to process and verify observation bias in real time, and correct day-ahead scheduling status; The state deviation formed in the next time scale is fed back to the previous time scale to correct the processing parameters of the previous time scale; The co-evolutionary feature matrix is ​​updated based on the corrected state and processing parameters.

8. A source-grid-load-storage integrated feature modeling system, employing the source-grid-load-storage integrated feature modeling method as described in any one of claims 1 to 7, characterized in that, include: Power source characteristic construction module, load characteristic construction module, energy storage characteristic construction module, grid response construction module, and collaborative evolution construction module; The power characteristic construction module is used to generate a cluster of wind and solar power output curves based on meteorological reanalysis data, identify risk intervals according to the slope rate and the duration of extreme power output, and form a power output characteristic vector. The load feature construction module is used to analyze air conditioning thermal inertia, electric vehicle participation willingness, and industrial process flexibility to form load flexibility feature vectors and source-load correlation data. The energy storage feature construction module is used to construct net load based on source-load correlation data, adjust the filter window according to risk interval, determine the energy storage target power and reserve capacity, and adjust the charging and discharging efficiency and minimum allowable state of charge according to health status and ambient temperature, so as to form energy storage operation feature vector and source-load-storage correlation data. The power grid response construction module is used to calculate power flow based on source-load-storage correlation data, determine DC reverse power, virtual impedance and tie line dynamic transmission margin, and form a power grid response feature vector. The co-evolution construction module is used to construct a co-evolution feature matrix by associating source-load-storage correlation data and power grid response feature vectors according to spatiotemporal and link relationships, establish multi-timescale state transmission and deviation feedback relationships, and generate a multi-dimensional feature database.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the integrated source-grid-load-storage feature modeling method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the integrated source-grid-load-storage feature modeling method according to any one of claims 1 to 7.

Images