A Method for Constructing an Equivalent Mathematical Model for a Large-Capacity Grid-Based Energy Storage Power Station

Abstract

This invention discloses a method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station, belonging to the field of power system energy storage technology. The specific steps are as follows: using sub-modules—connecting cables—as the basic hardware unit, a coupling strategy of segmented equivalent modeling and centralized modeling is adopted. First, for long-distance connecting cables, segments are formed according to the critical length of electrical characteristics. The distributed resistance, inductance, and capacitance of each segment are equivalently represented by a π-type circuit to accurately characterize the segmented transmission characteristics of the cable during transient processes. For sub-modules, a lumped parameter model of voltage / current source + internal resistance is used to retain the core electrical characteristics. This invention can significantly reduce transient simulation errors, thereby providing a reliable model basis for transient stability analysis of power stations and reducing power grid safety risks caused by simulation errors.

Description

Technical Field

[0001] This invention relates to the field of power system energy storage technology, specifically to a method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station. Background Technology

[0002] With the advancement of the "dual carbon" target, the proportion of new energy sources such as wind power and photovoltaics in the power system continues to increase. The volatility and intermittency of their output significantly increase the difficulty of controlling grid voltage / frequency stability. Grid-connected energy storage power stations, with their coordinated regulation capabilities across the generation, grid, load, and storage systems, can actively support voltage and participate in frequency regulation, becoming core equipment for mitigating the impact of new energy grid integration. Among them, large-capacity grid-connected energy storage power stations (such as 500MW and 1000MW units) have gradually become backbone regulation resources for the power grid due to their large regulation capacity and wide coverage.

[0003] Equivalent mathematical models are core tools for the design, commissioning, and operational optimization of grid-connected energy storage power stations, and their accuracy directly determines the reliability of grid simulation analysis results. However, current equivalent modeling methods for large-capacity grid-connected energy storage power stations still have the following key problems, making it difficult to meet the needs of engineering applications:

[0004] I. Large transient simulation error: Ignoring the influence of distributed parameters

[0005] Large-capacity grid-connected energy storage power stations typically consist of multiple sub-modules (such as energy storage converter sub-modules and battery cluster sub-modules), which are connected by long-distance cables (often 0.5-2km in length). The distributed resistance, inductance, and capacitance parameters of these cables have a significant impact on current / voltage changes during transient processes (such as voltage sags and short-circuit faults). However, existing equivalent models often use a "lumped parameter" simplification, treating the cable as a single resistor and inductor, neglecting the segmented characteristics of the distributed parameters.

[0006] II. Inaccurate dynamic response characterization: Simplified converter control strategy

[0007] The dynamic response (such as frequency regulation and voltage recovery) of grid-connected energy storage power stations mainly relies on the closed-loop control of the converter (including current loop, voltage loop, and power loop), and the control strategy includes detailed characteristics such as dead-zone compensation, saturation limitation, and frequency regulation droop coefficient adjustment. To reduce computational complexity, existing equivalent models often simplify the converter control loop, such as ignoring the saturation characteristics of the current loop and fixing the frequency regulation droop coefficient (without considering adaptive adjustment due to load changes).

[0008] III. Loss of Multi-Submodule Collaboration Characteristics: Imbalance Between Equivalence Accuracy and Complexity

[0009] Large-capacity grid-connected energy storage power stations typically have a large number of sub-modules (usually ≥10), and there are collaborative control logics among these sub-modules, such as power distribution and voltage balancing. Existing modeling methods either employ "full-detail modeling" (modeling each sub-module separately), resulting in extremely high model complexity (computation time is 5-8 times that of equivalent models), which cannot meet the requirements of real-time simulation; or "single sub-module equivalence" (using one representative sub-module to replace all sub-modules), completely ignoring the capacity differences, impedance differences, and collaborative control logic among the sub-modules.

[0010] Based on the above, a method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station is invented. Summary of the Invention

[0011] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0012] A method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station includes the following specific steps:

[0013] S1. Constructing a distributed parameter-lumped parameter coupled equivalent circuit model: Using submodules and connecting cables as basic hardware units, a coupling strategy of segmented equivalent modeling + lumped modeling is adopted. First, for long-distance connecting cables, segments are formed according to the critical length of electrical characteristics. The distributed resistance, inductance, and capacitance of each segment are equivalently represented by a π-type circuit to accurately characterize the segmented transmission characteristics of the cable during transient processes. For submodules, a lumped parameter model of voltage / current source + internal resistance is adopted to retain the core electrical characteristics. Finally, the distributed parameter model of the segmented cable is coupled in series with the lumped parameter model of the submodule to form a hardware unit equivalent model that both restores the distributed characteristics and controls the modeling complexity, ensuring that the simulation error of current and voltage in transient scenarios is ≤5%.

[0014] S2. Establish a hierarchical equivalent model for converter control: Centered on the control function hierarchy of the converter—switching execution, closed-loop regulation, and scheduling response—an equivalent strategy is adopted that preserves key characteristics and simplifies redundant details. First, the bottom layer replaces the full-switching model with an average switching model, introducing a dead-zone correction coefficient to accurately reproduce the switching characteristics of the IGBT. Next, the middle layer retains the PI control structure of the current and voltage loops and introduces dynamic saturation limits. Finally, the top layer introduces adaptive control parameters, retaining scheduling-level characteristics such as frequency modulation dead zone and power limiting. The three-layer control model is interconnected layer by layer, forming a complete dynamic control equivalent mechanism from switching execution to scheduling response.

[0015] S3 proposes a multi-submodule grouping and collaborative equivalence strategy: based on the consistency of submodule characteristics, a hierarchical strategy of grouping clustering + weighted equivalence + dynamic collaboration is adopted. First, N submodules are divided into M groups according to their capacity and outlet impedance characteristics to balance group representativeness and computational complexity. Then, for submodules in the same group, the group-level equivalent parameters are calculated using a capacity-weighted average to retain the core characteristics within the group. Afterward, an inter-group collaboration coefficient is introduced to correct the inter-group power interaction characteristics. In fault scenarios, the group weighting coefficient and control saturation limit are updated in real time to ensure that the collaborative characteristics are not lost after the submodule fault is cleared. Finally, a multi-module equivalence strategy is formed that simplifies the complexity of grouping and restores the interaction characteristics of collaboration.

[0016] S4: Model Integration and Parameter Verification: The coupled equivalent circuit of S1, the hierarchical control model of S2, and the grouping coordination strategy of S3 are integrated to form a complete equivalent mathematical model of a large-capacity grid-connected energy storage power station; and the model parameters are iteratively corrected by comparing the field test data with the model simulation results until the simulation error is ≤5%.

[0017] As a preferred embodiment of the method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station according to the present invention, the specific steps of S1 are as follows:

[0018] S11, Equivalent Value of Cable Distribution Parameters in Segments: The long-distance cable between sub-modules is segmented according to length, and the distributed resistance of each segment is equivalently valued using a π-type circuit. Distributed inductance Distributed capacitance And calculate the parameters for each segment using a formula;

[0019] S12, Submodule lumped parameter modeling: The energy storage unit of each submodule is equivalent to a voltage source + internal resistance R_b, and the converter is equivalent to a controlled current source;

[0020] S13, Coupling Connection: Connect the segmented cable π-type equivalent circuit in series with the submodule lumped parameter model to form a distributed-lumped coupled submodule unit equivalent circuit;

[0021] S14, Environmental Adaptive Parameter Correction: First, environmental factors are identified and data is collected. Then, environmental correction coefficients are introduced, and correction formulas are established. Finally, dynamic correction is performed.

[0022] As a preferred embodiment of the method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station as described in this invention, the formula in S11 is:

[0023] ,in The resistance per unit length of the cable. Let k be the length of the cable segment.

[0024] ,in, The inductance per unit length of the cable;

[0025] ,in, Capacitance per unit length of cable.

[0026] As a preferred embodiment of the method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station according to the present invention, the specific steps of S14 are as follows:

[0027] S141, Environmental Factor Identification and Data Acquisition: First, collect the real-time temperature T of the cable laying area of ​​the submodule; then collect the relative humidity H inside the energy storage compartment; finally, collect the altitude Alt of the power station location.

[0028] S142, Environmental correction formulas for key parameters: Based on the calculated cable parameters and submodule battery internal resistance R_b, an environmental correction coefficient is introduced to establish correction formulas for cable distributed resistance, battery internal resistance and converter heat dissipation.

[0029] S143, Dynamic Correction Mechanism: First, the triggering conditions are corrected in real time. When the change in environmental factors exceeds the threshold, parameter correction is automatically triggered. Then, historical data is iteratively optimized to establish an environmental-parameter error database. The correction coefficient is optimized in reverse through field test data and updated once a quarter to improve the field adaptability of the correction formula.

[0030] As a preferred embodiment of the method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station as described in this invention, the cable distributed resistance correction formula is as follows:

[0031] For every 1°C change in temperature, the correction formula is:

[0032]

[0033] in, 25 represents the temperature coefficient of the cable conductor, T represents the real-time field temperature, and 25 represents the standard ambient temperature.

[0034] As a preferred embodiment of the method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station as described in this invention, the battery internal resistance correction formula is as follows:

[0035] R_b'=R_b×[1+0.005×(H-60)+0.02×(25-T)], when T<25℃;

[0036] Where 0.005 is the humidity correction factor, 0.02 is the low temperature internal resistance correction factor, and H is the real-time humidity at the site.

[0037] As a preferred embodiment of the method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station as described in this invention, the converter heat dissipation correction formula is:

[0038] When the altitude is greater than 1000m, the maximum allowable current of the converter IGBT decreases, and the current loop saturation value in S2 needs to be corrected. The formula is as follows:

[0039] I_sat' = I_sat × [1 - 0.0001 × (Alt - 1000)], when Alt > 1000m;

[0040] Where 0.0001 is the altitude correction factor, and Alt is the site altitude, to ensure that the switching characteristics are equivalent to the heat dissipation limitations at high altitudes.

[0041] As a preferred embodiment of the method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station according to the present invention, the specific steps of S2 are as follows:

[0042] S21, bottom layer: The average switching model is used to replace the IGBT full switching model, the switching function S is introduced, and the dead time τ_d is taken into account to correct the switching function;

[0043] S22, Middle Layer: Retains the PI control characteristics of the current loop and voltage loop, introduces saturation limit, and obtains PI parameters through the step response curve of field test;

[0044] S23, Top layer: Introduce an adaptive FM droop coefficient K_f and retain the FM dead zone characteristics.

[0045] As a preferred embodiment of the method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station according to the present invention, the specific steps of S3 are as follows:

[0046] S31, Submodule Grouping: Based on the submodule capacity and impedance characteristics, divide the N submodules into M groups, with each group accounting for ≥10% of the total capacity;

[0047] S32, Intra-group weighted equivalence: For sub-modules within the same group, the group equivalence parameters are calculated using a capacity-weighted average.

[0048] S33, Inter-group Cooperative Correction: Introducing a cooperative coefficient α_ij to correct the equivalent current of the group;

[0049] S34, Fault Coordination Equivalence: When a submodule in a group is fault-cleared, the capacity weighting coefficient of the group is updated in real time, and the saturation limit value of the middle layer control is adjusted synchronously.

[0050] Compared with existing technologies:

[0051] 1. By constructing a distributed parameter-lumped parameter coupled equivalent circuit model, transient simulation errors can be greatly reduced, thereby providing a reliable model basis for power plant transient stability analysis and reducing power grid safety risks caused by simulation errors.

[0052] 2. By establishing a hierarchical converter control equivalent model, the simulation error of dynamic response can be greatly reduced, thereby providing accurate model support for the optimization of power plant frequency regulation and voltage regulation strategies, and ensuring the grid support capability of grid-connected energy storage power plants under the scenario of new energy fluctuations.

[0053] 3. By proposing a multi-submodule grouping and collaborative equalization strategy, it can achieve a balance between high precision and low complexity, significantly shortening the computation time compared to full-detail modeling, while also greatly reducing simulation errors in scenarios such as harmonics. This enables it to meet the dual requirements of real-time simulation and high-precision evaluation, and is suitable for power plants of different capacities from 200MW to 1000MW. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below.

[0055] This invention provides a method for constructing an equivalent mathematical model of a large-capacity grid-connected energy storage power station, including the following specific steps:

[0056] S1. Constructing a distributed parameter-lumped parameter coupled equivalent circuit model: Using sub-modules and connecting cables as basic hardware units, a coupling strategy of segmented equivalent modeling + lumped modeling is adopted. First, for long-distance connecting cables, segments are formed according to the critical length of electrical characteristics. The distributed resistance, inductance, and capacitance of each segment are equivalently represented by a π-type circuit to accurately characterize the segmented transmission characteristics of the cable during transient processes. For sub-modules such as energy storage converters and battery clusters, a lumped parameter model of voltage source / current source + internal resistance is adopted to retain the core electrical characteristics. Finally, the distributed parameter model of the segmented cable is coupled in series with the lumped parameter model of the sub-module to form a hardware unit equivalent model that both restores the distributed characteristics and controls the modeling complexity, ensuring that the simulation error of current and voltage in transient scenarios is ≤5%.

[0057] S2. Establish a hierarchical equivalent model for converter control: Centered on the control function hierarchy of the converter—switching execution, closed-loop regulation, and scheduling response—an equivalent strategy is adopted that preserves key characteristics and simplifies redundant details. First, the bottom layer replaces the full-switching model with an average switching model, introducing a dead-zone correction coefficient to accurately reproduce the switching characteristics of the IGBT. Next, the middle layer retains the PI control structure of the current and voltage loops and introduces dynamic saturation limits. Finally, the top layer introduces adaptive control parameters, retaining scheduling-level characteristics such as frequency modulation dead zone and power limiting. The three-layer control model is interconnected layer by layer, forming a complete dynamic control equivalent mechanism from switching execution to scheduling response.

[0058] S3 proposes a multi-submodule grouping and collaborative equivalence strategy: based on the consistency of submodule characteristics, a hierarchical strategy of grouping clustering + weighted equivalence + dynamic collaboration is adopted. First, N submodules are divided into M groups according to their capacity and outlet impedance characteristics to balance group representativeness and computational complexity. Then, for submodules in the same group, the group-level equivalent parameters are calculated using a capacity-weighted average to retain the core characteristics within the group. Afterward, an inter-group collaboration coefficient is introduced to correct the inter-group power interaction characteristics. In fault scenarios, the group weighting coefficient and control saturation limit are updated in real time to ensure that the collaborative characteristics are not lost after the submodule fault is cleared. Finally, a multi-module equivalence strategy is formed that simplifies the complexity of grouping and restores the interaction characteristics of collaboration.

[0059] S4: Model Integration and Parameter Verification: The coupled equivalent circuit of S1, the hierarchical control model of S2, and the grouping coordination strategy of S3 are integrated to form a complete equivalent mathematical model of a large-capacity grid-connected energy storage power station; and the model parameters are iteratively corrected by comparing the field test data (such as transient current peak value, frequency recovery time, and harmonic content) with the model simulation results (such as adjusting the number of cable segments and the coordination coefficient α_ij) until the simulation error is ≤5%.

[0060] The specific steps of S1 are as follows:

[0061] S11, Equivalent Value of Cable Distribution Parameters in Segments: The long-distance cable between submodules is segmented by length (each segment ≤ 0.3km), and the distributed resistance of each segment is equivalently valued using a π-type circuit. Distributed inductance Distributed capacitance And calculate the parameters for each segment using a formula;

[0062] S12, Submodule lumped parameter modeling: The energy storage unit (such as lithium battery cluster) of each submodule is equivalent to a voltage source + internal resistance R_b (the voltage source voltage changes dynamically with SOC, and the internal resistance is a linear function of SOC), and the converter is equivalent to a controlled current source (the average model retains the IGBT switching characteristics and considers the dead time τ_d).

[0063] S13, Coupling Connection: Connect the segmented cable π-type equivalent circuit in series with the submodule lumped parameter model to form a distributed-lumped coupled submodule unit equivalent circuit;

[0064] S14, Environmental Adaptive Parameter Correction: First, environmental factors are identified and data is collected. Then, environmental correction coefficients are introduced, and correction formulas are established. Finally, dynamic correction is performed.

[0065] The formula in S11 is:

[0066] ,in The resistance per unit length of the cable (Ω / km). The length of the k-th cable segment (km);

[0067] ,in, The inductance per unit length of the cable (mH / km);

[0068] ,in, Capacitance per unit length of cable (μF / km).

[0069] The specific steps of S14 are as follows:

[0070] S141, Environmental Factor Identification and Data Acquisition: First, collect the real-time temperature T of the cable laying area of ​​the submodule (such as cable trench, outdoor cable tray), with the acquisition frequency synchronized with the model simulation step size (recommended ≤50ms); then, collect the relative humidity H inside the energy storage compartment (the area where the battery cluster and converter are located), focusing on the impact of humidity on the battery internal resistance; after that, collect the altitude Alt of the power station location, which mainly affects the air insulation strength and indirectly affects the heat dissipation efficiency of the converter switching tubes (heat dissipation efficiency decreases when altitude >1000m).

[0071] S142, Environmental correction formulas for key parameters: Based on the calculated cable parameters and submodule battery internal resistance R_b, an environmental correction coefficient is introduced to establish correction formulas for cable distributed resistance, battery internal resistance and converter heat dissipation.

[0072] S143, Dynamic Correction Mechanism: First, the triggering conditions are corrected in real time. When the change in environmental factors exceeds the threshold (temperature ΔT≥5℃, humidity ΔH≥10%RH, altitude ΔAlt≥500m), parameter correction is automatically triggered. Then, historical data is iteratively optimized to establish an environmental-parameter error database. The correction coefficient is then optimized in reverse using field test data (such as transient current at different temperatures). The humidity correction factor is updated quarterly to improve the field adaptability of the correction formula.

[0073] The cable distributed resistance correction ( The formula is:

[0074] For every 1°C change in temperature, the resistance of a copper core cable changes by approximately 0.393%. The correction formula is as follows:

[0075]

[0076] in, T represents the temperature coefficient of the cable conductor (0.00393 / ℃ for copper core), T represents the real-time ambient temperature (℃), and 25 represents the standard ambient temperature.

[0077] The battery internal resistance correction ( The formula is:

[0078] R_b'=R_b×[1+0.005×(H-60)+0.02×(25-T)], when T<25℃;

[0079] Where 0.005 is the humidity correction factor, 0.02 is the low temperature internal resistance correction factor, and H is the real-time humidity at the site.

[0080] The formula for the converter heat dissipation correction (related to the switching characteristics of S2) is as follows:

[0081] When the altitude is greater than 1000m, the maximum allowable current of the converter IGBT decreases, and the current loop saturation value in S2 needs to be corrected. The formula is as follows:

[0082] I_sat' = I_sat × [1 - 0.0001 × (Alt - 1000)], when Alt > 1000m;

[0083] Where 0.0001 is the altitude correction factor, and Alt is the site altitude, to ensure that the switching characteristics are equivalent to the heat dissipation limitations at high altitudes.

[0084] The specific steps of S2 are as follows:

[0085] S21, bottom layer (equivalent to switching characteristics): The average switching model is used to replace the IGBT full switching model, and a switching function S is introduced (s = 1 when conducting, S = 0 when turning off), and the dead time τ_d (usually 0.3-0.8μs) is taken into account to correct the switching function;

[0086] S22, Middle Layer (Closed-Loop Control Equivalent): Retains the PI control characteristics of the current loop and voltage loop, and introduces saturation limits (the saturation value of the current loop is 1.5-2 times the rated current, and the saturation value of the voltage loop is 0.9-1.1 times the rated voltage). The PI parameters (Kp, Ki) are identified and obtained through the step response curves of field tests.

[0087] S23, Top-level (Schedule Control Equivalent): Introduce an adaptive frequency modulation droop coefficient K_f (K_f is dynamically adjusted with the grid frequency deviation Δf. The larger Δf is, the smaller K_f is, satisfying the formula K_f=K_f0×(1-0.2×|Δf|), where K_f0 is the rated droop coefficient), and retain the frequency modulation dead zone characteristics (dead zone range 0.01-0.03Hz).

[0088] The specific steps of S3 are as follows:

[0089] S31, Submodule Grouping: Based on the submodule capacity and impedance characteristics (submodules with an output impedance deviation ≤10% belong to the same group), divide N submodules into M groups (M≤5, to ensure controllable complexity), with each group accounting for ≥10% of the capacity;

[0090] S32, Intra-group weighted equivalent: For sub-modules within the same group, the group equivalent parameters are calculated using a capacity-weighted average (e.g., group equivalent voltage U_g=Σ(U_i×P_i) / ΣP_i, where U_i is the voltage of the i-th sub-module and P_i is the capacity of the i-th sub-module).

[0091] S33, Inter-group Coordination Correction: Introduce a coordination coefficient α_ij (characterizing the power interaction intensity between the i-th group and the j-th group, α_ij is obtained by fitting the inter-group power flow data from field tests, ranging from 0.6 to 1.0), and correct the equivalent current of the group (I_g_i=I_g_i0×(1+0.1×α_ij), where I_g_i0 is the equivalent current of the group without considering coordination).

[0092] S34, Fault Coordination Equivalence: When a submodule in a group is disconnected due to a fault, the capacity weighting coefficient of the group is updated in real time (e.g., after disconnecting a 50MW submodule, the group capacity drops from 300MW to 250MW, and the weighting coefficient is corrected from 0.6 to 0.5), and the saturation limit value of the middle layer control is adjusted synchronously (the saturation value is increased to twice the rated value after the fault).

[0093] In practical applications, the following examples are included, but are not limited to:

[0094] A 500MW grid-connected energy storage power station (containing 12 sub-modules: 6 50MW and 6 33.3MW, with a 1.5km cable length between sub-modules, cable parameters...) =0.012Ω / km =0.22mH / km Taking (e.g., 0.11 μF / km) as an example, the specific implementation steps are as follows:

[0095] (I) Step 1: Constructing a coupled equivalent circuit model

[0096] 1.1 Cable Segmentation: The 1.5km cable is divided into 5 segments (0.3km each). Parameter calculations for each segment are as follows:

[0097]

[0098] 1.2 Submodule lumped parameters: Energy storage unit voltage varies with SOC (U=3.2kV, internal resistance R_b=0.05Ω when SOC=80%), converter dead time τ_d=0.5μs.

[0099] 1.3 Coupling connection: The 5-segment π-type cable circuit is connected in series with the lumped parameter model of the submodule to form the equivalent circuit of the submodule unit.

[0100] 1.4.1 Environmental data: The temperature at the data collection site was 15℃, the humidity was 75%, and the altitude was 1500m;

[0101] 1.4.2 Parameter Correction:

[0102] cable The original calculated value was 0.0036Ω, and the corrected value is now...

[0103] The original calculated value of battery R_b was 0.05Ω. After correction, R_b' = 0.05 × [1 + 0.005 × (75 - 60) + 0.02 × (25 - 15)] ≈ 0.06375Ω.

[0104] The original current loop saturation value was 1.8 pu, and after correction, I_sat' = 1.8 × [1 - 0.0001 × (1500 - 1000)] = 1.71 pu;

[0105] 1.4.3 Simulation verification: After correction, the calculated peak current value for the scenario "voltage temporarily drops to 0.7 pu" is 2.13 pu, and the field test shows 2.15 pu, with an error of 0.9% (the original scheme had an error of 1.4%). The accuracy has been further improved, and it is seamlessly connected with the hierarchical control model in the subsequent step 2 (the corrected current loop saturation value is directly used for the equivalent value of the middle layer control).

[0106] (II) Step 2: Establish a hierarchical control model

[0107] 2.1 Bottom layer: Average switching model, after the switching function S corrects for the dead time, the conduction time accounts for 99.8%;

[0108] 2.2 Middle layer: Current loop Kp=0.6, Ki=12, saturation value 1.8pu; Voltage loop Kp=0.3, Ki=6, saturation value 1.05pu;

[0109] 2.3 Top layer: Rated droop coefficient K_f0=4% / Hz, adaptive correction formula K_f=4% / Hz×(1-0.2×|Δf|), frequency modulation dead zone 0.02Hz.

[0110] (III) Step 3: Multi-submodule grouping and collaborative equivalence

[0111] 3.1 Grouping: Group 1 consists of 6 50MW sub-modules (total capacity 300MW), and Group 2 consists of 6 33.3MW sub-modules (total capacity 200MW).

[0112] 3.2 Intra-group weighting: Equivalent voltage of group 1 U_g1=Σ(U_i×50) / 300, equivalent voltage of group 2 U_g2=Σ(U_i×33.3) / 200;

[0113] 3.3 Inter-group coordination: Through field testing, the coordination coefficient α_12 = 0.85, and the equivalent current of group 1 I_g1 = I_g10 × (1 + 0.1 × 0.85) = 1.085 × I_g10;

[0114] 3.4 Fault Coordination: If Group 1 removes one 50MW submodule, the capacity of Group 1 becomes 250MW. The weighting coefficient is corrected from 300 / 500=0.6 to 250 / 450≈0.56, and the current loop saturation value is increased to 2.0pu.

[0115] (iv) Model Validation

[0116] Transient simulation: The simulation showed a voltage drop to 0.7 pu. The model calculated a peak current of 2.12 pu, while the field test showed 2.15 pu, with an error of 1.4%.

[0117] Frequency modulation simulation: Simulating a "sudden load increase of 100MW", the model calculated the frequency recovery time to be 2.05s, while the field test showed 2.1s, with an error of 2.4%.

[0118] Harmonic simulation: The simulated grid-connected third harmonic content is 1.72%, while the field test shows 1.8%, with an error of 4.4%.

[0119] All verification results showed an error of ≤5%, meeting the engineering requirements.

[0120] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for constructing an equivalent mathematical model of a large-capacity network-constructed energy storage power station, characterized in that, The specific steps include the following: S1, constructing a distributed parameter-centralized parameter coupling equivalent circuit model: taking a sub-module-connection cable as a basic hardware unit, adopting a coupling strategy of segmented equivalent+centralized modeling; first, the long-distance connection cable is segmented according to the critical length of electrical characteristics, and the segmented transmission characteristics of the cable in the transient process are characterized by using a π-type circuit to equivalently represent the distributed resistance, inductance and capacitance of each segment; for the sub-module, a centralized parameter model of a voltage source / current source+internal resistance is adopted to retain the core electrical characteristics; finally, the distributed parameter model of the segmented cable and the centralized parameter model of the sub-module are coupled in series to form a hardware unit equivalent model that restores the distributed characteristics and controls the modeling complexity; S2, establishing a hierarchical converter control equivalent model: taking the control function level of the switching execution-loop regulation-scheduling response of the converter as the core, adopting an equivalent strategy of simplifying redundant details while retaining key characteristics; first, the bottom layer uses an average switching model to replace the full switching model, and introduces a dead zone correction coefficient to restore the switching action characteristics of the IGBT; then, the middle layer retains the PI control structure of the current loop and the voltage loop, and introduces dynamic saturation limitation; after that, the top layer introduces adaptive control parameters to retain the frequency modulation dead zone and power limit scheduling level characteristics; the three-layer control model is connected layer by layer to form a complete dynamic control equivalent mechanism from switching execution to scheduling response; S3, proposing a multi-sub-module grouping coordination equivalent strategy: taking the consistency of sub-module characteristics as the grouping basis, adopting a hierarchical strategy of grouping clustering+weighted equivalent+dynamic coordination; first, N sub-modules are divided into M groups according to the capacity and outlet impedance characteristics of the sub-modules to balance the representativeness and computational complexity of the groups; then, for sub-modules in the same group, the group-level equivalent parameters are calculated by using capacity weighted average to retain the core characteristics within the group; after that, an inter-group coordination coefficient is introduced to correct the power interaction characteristics between groups, and the grouping weighting coefficients and control saturation limits are updated in real time in the fault scenario, and the coordination characteristics are not lost after the sub-module fault is removed; finally, a multi-module equivalent strategy is formed that simplifies the complexity of the groups and restores the interaction characteristics of the coordination; S4: model integration and parameter verification: integrating the coupling equivalent circuit of S1, the hierarchical control model of S2 and the grouping coordination strategy of S3 to form a complete equivalent mathematical model of a large-capacity grid-connected energy storage power station; and comparing the field test data with the simulation results of the model, iteratively correcting the model parameters until the simulation error is ≤5%.

2. The equivalent mathematical model construction method of a large-capacity network-constructed energy storage power station according to claim 1, characterized in that, The specific steps of S1 are as follows: S11, Equivalent Value of Cable Distribution Parameters in Segments: The long-distance cable between sub-modules is segmented according to length, and the distributed resistance of each segment is equivalently valued using a π-type circuit. Distributed inductance Distributed capacitance And calculate the parameters for each segment using a formula; S12, centralized parameter modeling of sub-modules: the energy storage unit of each sub-module is equivalent to a voltage source+internal resistance R_b, and the converter is equivalent to a controlled current source; S13, coupling connection: the segmented cable π-type equivalent circuit and the centralized parameter model of the sub-module are connected in series to form a distributed-centralized coupled sub-module unit equivalent circuit; S14, environment adaptive parameter correction: first, identify and collect environmental factors, then introduce an environmental correction coefficient to establish a correction formula; and then perform dynamic correction.

3. The equivalent mathematical model construction method of a large-capacity network-constructed energy storage power station according to claim 2, characterized in that, The formula in S11 is: wherein R is the cable unit length resistance, Lk is the length of the kth cable section; wherein, L is the inductance per unit length of the cable; wherein, C is the cable capacitance per unit length.

4. The equivalent mathematical model construction method of a large-capacity network-constructed energy storage power station according to claim 2, characterized in that, The specific steps of S14 are as follows: S141, environmental factor identification and data collection: first, collect the real-time temperature T of the cable laying area of the sub-module; then collect the relative humidity H in the energy storage cabin; then collect the altitude Alt of the power station site; S142, environmental correction formula for key parameters: for the calculated cable parameters and sub-module battery internal resistance R_b, an environmental correction coefficient is introduced to establish the cable distribution resistance, battery internal resistance and converter heat dissipation correction formula; S143, dynamic correction mechanism: first, real-time correction trigger condition, when the environmental factor change exceeds the threshold, automatically trigger parameter correction; then, historical data iteration optimization, establish environment-parameter error database, through the reverse optimization of correction coefficient by field test data, update once every quarter.

5. The equivalent mathematical model construction method of a large-capacity network-constructed energy storage power station according to claim 4, characterized in that, The cable distribution resistance correction formula is: The correction formula is: ， wherein, Tc = Tc0 + Tc25(T - 25) (1) where Tc0 is the cable conductor temperature at standard ambient temperature, T is the real-time temperature at the site, and 25 is the standard ambient temperature.

6. The equivalent mathematical model construction method of a large-capacity network-constructed energy storage power station according to claim 4, characterized in that, The battery internal resistance correction formula is: R_b'=R_b×[1+0.005×(H-60)+0.02×(25-T)], when T<25℃; Where 0.005 is the humidity correction coefficient, 0.02 is the low temperature internal resistance correction coefficient, and H is the real-time humidity.

7. The equivalent mathematical model construction method of a large-capacity network-constructed energy storage power station according to claim 4, characterized in that, The converter heat dissipation correction formula is: When the altitude is greater than 1000m, the maximum allowable current of the converter IGBT decreases, and the current loop saturation value in S2 needs to be corrected, the formula is: I_sat'=I_sat×[1-0.0001×(Alt-1000)], when Alt>1000m; Where I_sat is the current loop saturation value, 0.0001 is the altitude correction coefficient, Alt is the field altitude, and the switching characteristic is equivalent to the high altitude heat dissipation limit.

8. The equivalent mathematical model construction method of a large-capacity network-constructed energy storage power station according to claim 1, characterized in that, The specific steps of S2 are as follows: S21, bottom layer: use average switching model instead of IGBT full switching model, introduce switching function S, and count in the correction of dead time τ_d to switching function; S22, middle layer: retain the PI control characteristics of current loop and voltage loop, introduce saturation limit, and obtain PI parameters through step response curve identification of field test; S23, top layer: introduce adaptive frequency modulation droop coefficient K_f, and retain frequency modulation dead zone characteristics.

9. The equivalent mathematical model construction method of a large-capacity network-constructed energy storage power station according to claim 1, characterized in that, The specific steps of S3 are as follows: S31, sub-module grouping: according to the capacity and impedance characteristics of sub-modules, N sub-modules are divided into M groups, and the capacity proportion of each group is ≥10%; S32, group weighted equivalence: for sub-modules in the same group, the group equivalent parameters are calculated by capacity weighted average; S33, group correction: introduce cooperation coefficient α_ij to correct group equivalent current; S34, fault cooperative equivalence: when a sub-module in a group is fault removed, the capacity weighting coefficient of the group is updated in real time, and the saturation limit value of the middle layer control is adjusted synchronously.

Images