A method for VSG parameter tuning for grid-type energy storage suitable for weak power grids

By using differentiated modeling and sensitivity matrix search, the VSG parameters were tuned, which solved the problem of unreasonable VSG control algorithm parameter tuning under weak power grids and improved system stability and power quality.

CN122782602BActive Publication Date: 2026-11-10STATE GRID HEILONGJIANG ELECTRIC POWER CO LTD SHUANGYASHAN POWER SUPPLY CO
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611264163.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-08-20
Publication Date
2026-11-10
Estimated Expiration
2046-08-20

AI Technical Summary

Technical Problem

In weak power grid environments, the existing VSG control algorithm parameter tuning methods cannot match complex operating conditions, resulting in an inability to effectively play a role in stabilizing the grid and easily causing faults such as low-frequency power oscillations and voltage instability.

Method used

By collecting line parameters, a voltage-power sensitivity matrix is ​​constructed, candidate nodes are screened, and mathematical models of three types of interconnected components are constructed differently. The steady-state operating point is solved, the system state matrix is ​​obtained, the modal damping ratio and oscillation frequency of complex eigenvalues ​​are calculated, key eigenvalues ​​are screened, a sensitivity matrix is ​​constructed, a stability boundary is searched along the sensitivity matrix, damping ratio constraints are applied, and parameters are tuned to maximize the safety margin.

Benefits of technology

It improves the VSG's ability to resist disturbances and parameter drift under weak power grids, enhances system stability and power quality, and solves the problem of unreasonable parameter tuning in traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122782602B_ABST
    Figure CN122782602B_ABST
Patent Text Reader

Abstract

This application relates to the field of VSG parameter tuning technology for energy storage, and proposes a method for tuning the parameters of grid-connected VSGs suitable for weak power grids. The method includes: collecting line parameters of the target distribution substation area and screening candidate nodes; setting up and constructing mathematical models of three types of interconnected elements, solving for the steady-state operating point, solving for the partial derivatives of the mathematical models of the three types of interconnected elements at the steady-state operating point, constructing the system state matrix, constructing a standard state-space model and solving for complex eigenvalues, screening key eigenvalues, and constructing a sensitivity matrix; constructing a stable feasible region for the combination of tuning parameters, and completing the tuning of the parameters based on the damping ratios of all tuning parameters in the combination of tuning parameters within the stable feasible region. This application can improve the grid stabilization performance of VSGs.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of VSG parameter tuning technology for energy storage, specifically to a method for tuning VSG parameters of grid-type energy storage suitable for weak power grids. Background Technology

[0002] The large-scale distributed integration of renewable energy into distribution networks has led to a continuous increase in the penetration rate of new energy in the power grid. Some distribution areas are gradually exhibiting typical characteristics of weak power grids, such as small short-circuit capacity, large equivalent impedance, and weak system strength. Generally, a power grid with a short-circuit ratio (SCR) below 3 is defined as a weak power grid. Under such grid conditions, traditional grid-connected inverters heavily rely on phase-locked loops (PLLs) to track the grid voltage phase, which easily leads to power quality and system stability problems such as PLL instability and harmonic oscillations, making them unsuitable for the operational demands of grids with high proportions of renewable energy integration. Grid-connected energy storage converters can achieve autonomous voltage and frequency stabilization independently of external grid voltage and frequency signals. By relying on the Virtual Synchronous Generator (VSG) control algorithm to simulate the second-order swing equation of a synchronous generator and introducing control parameters such as virtual inertia, damping coefficient, and active power droop coefficient, they provide inertia support and damping response to the grid, effectively optimizing the grid's dynamic characteristics. This is a core technical means to solve the stability problems of weak power grids.

[0003] However, the VSG control performance of the VSG control algorithm relies heavily on key parameters such as parasitic resistance, filter capacitor, virtual moment of inertia, damping coefficient, and line equivalent impedance. Especially in weak grid environments, the grid equivalent impedance cannot be ignored, and various control parameters form a strong coupling effect with the grid impedance. Existing fixed parameter tuning methods cannot match the complex operating conditions of weak grids. Not only is it difficult to determine reasonable inertia support and damping response, thus failing to realize the grid stabilization function of VSG, but it is also very easy to induce faults such as low-frequency power oscillation and voltage instability. Summary of the Invention

[0004] This application provides a method for tuning the parameters of a grid-connected energy storage system (VSG) suitable for weak power grids, in order to solve the problem that the preset values ​​of the control system parameters in the VSG control algorithm are unreasonable, thus failing to give full play to the grid stabilization function of the VSG. The specific technical solution adopted is as follows:

[0005] One embodiment of this application provides a method for tuning the parameters of a grid-connected energy storage (VSG) system suitable for weak power grids. The method includes the following steps:

[0006] Collect line parameters of the target distribution area, calculate the short-circuit ratio (SCR) of each node, construct a voltage-power sensitivity matrix, and screen candidate nodes based on the short-circuit ratio;

[0007] Three types of interconnecting elements are set up, and mathematical models of the three types of interconnecting elements are constructed differently. The steady-state operating point of the target distribution substation is solved. The partial derivatives of the mathematical models of the three types of interconnecting elements are solved at the steady-state operating point to obtain the sub-matrices of each type of interconnecting element and the coupling sub-matrices between interconnecting elements. Then, the system state matrix is ​​constructed. The distribution system containing VSG, power grid and distributed power source is linearized according to the steady-state operating point. A standard state-space model is constructed and the complex eigenvalues ​​are solved. The modal damping ratio, oscillation frequency and participation factor of the complex eigenvalues ​​are calculated. Based on the modal damping ratio, oscillation frequency and participation factor, the key eigenvalues ​​in the complex eigenvalues ​​are selected, the sensitivity corresponding to the key eigenvalues ​​is calculated, and the sensitivity matrix is ​​constructed.

[0008] The stability boundary is searched along the sensitivity direction within the sensitivity matrix, and damping ratio constraints are applied to construct the stable feasible region of the combination of tuning parameters. Using the principle of maximizing safety margin, the tuning parameters are tuned according to the damping ratio of all tuning parameters in the combination of tuning parameters within the stable feasible region.

[0009] Furthermore, the method for selecting candidate nodes is as follows:

[0010] The comprehensive score index of a node is calculated based on the short-circuit ratio (SCR) of the node and the sum of the column elements of the corresponding node in the voltage-power sensitivity matrix.

[0011] The first preset number of nodes with the largest comprehensive score index are recorded as candidate nodes.

[0012] Furthermore, the formula for calculating the comprehensive scoring index of the node is as follows:

[0013] Represents a node The comprehensive scoring indicators; Represents a node The short-circuit ratio of SCR; This represents the preset threshold for the short-circuit ratio; This represents the maximum value normalization function; , These represent the first weight and the second weight, respectively, and the sum of the first weight and the second weight is 1. Represents the voltage-power sensitivity matrix of the first... The cumulative sum of column elements; This represents the maximum value among the cumulative sums of the elements in each column of the voltage-power sensitivity matrix.

[0014] Furthermore, the three types of interconnecting components specifically include: VSG, power grid network components, and distributed power sources.

[0015] Furthermore, the mathematical models for the three types of interconnecting elements are solved by partial derivatives at the steady-state operating point to obtain the sub-matrices of each type of interconnecting element and the coupling sub-matrices between interconnecting elements, thereby constructing the system state matrix. The specific methods include:

[0016] The sub-matrices of VSG, grid network elements, and distributed generation are respectively denoted as... , and ;

[0017] The coupling sub-matrix between interconnected elements includes the coupling sub-matrix of network state to VSG state, the coupling sub-matrix of VSG state to network state, and the coupling sub-matrix of network state to distributed generation, which are denoted as follows: , and ;

[0018] System state matrix .

[0019] Furthermore, the standard state-space model is as follows:

[0020] In this process, the state vector is obtained by sequentially arranging the VSG state variables, grid state variables, and distributed generation state variables as the first, second, and third row elements, respectively. Represents the state vector; Represents the system state matrix; The matrix representing the partial derivatives of the system's nonlinear equations with respect to the input vector; Represents the perturbation input vector; Indicates time.

[0021] Furthermore, the modal damping ratio of the complex eigenvalues ​​and the oscillation frequency are calculated as follows:

[0022] The arithmetic square root of the sum of the squares of the real and imaginary parts of the complex eigenvalues ​​is used as the denominator, and the opposite of the real part of the complex eigenvalues ​​is used as the numerator. The value of the fraction formed by the numerator and denominator is denoted as the modal damping ratio of the complex eigenvalue.

[0023] The imaginary part of the complex eigenvalue is combined with 2. The ratio of is denoted as the oscillation frequency of the complex eigenvalue.

[0024] Furthermore, the calculation method for the participation factor is as follows:

[0025] The product of the absolute values ​​of the same state variable in the left and right eigenvectors of the complex eigenvalues ​​is denoted as the eigenproduct of the same state variable. The ratio of the eigenproduct of the state variable to the sum of the eigenproducts of all state variables is denoted as the modal participation factor of the state variable.

[0026] The sum of the modal participation factors of the state variables corresponding to the complex eigenvalues ​​of the four state parameters of the VSG is denoted as the participation factor of the complex eigenvalues. The state parameters of the VSG include power angle, angular velocity deviation, and dq axis output current.

[0027] Furthermore, the method for selecting the key feature values ​​is as follows:

[0028] Identification condition 1: Modal damping ratio is less than the first preset threshold;

[0029] Identification condition 2: Modal oscillation frequency is greater than or equal to 0.1 Hz and less than or equal to 2 Hz;

[0030] Identification condition 3: The participation factor is greater than the second preset threshold;

[0031] The complex feature value that satisfies any one of the recognition conditions is recorded as the key feature value.

[0032] Furthermore, the tuning parameters include virtual moment of inertia, virtual damping coefficient, active droop coefficient, and reactive droop coefficient.

[0033] The beneficial effects of this application are:

[0034] This application first filters out instability parameter combinations that violate damping stability constraints, retaining only parameter points that meet stability criteria to obtain candidate nodes. It then employs a hierarchical differentiated modeling approach, performing refined modeling, simplified modeling, and a modeling approach that balances modeling accuracy and computational efficiency for three types of interconnected components. Through steady-state operating points and linearized solutions, it accurately quantifies the coupling and disturbance mechanisms of system variables, constructing a standard state-space model. Next, based on damping ratio, oscillation frequency, and parameter participation factors, it sets identification conditions to accurately screen low-frequency, weakly damped hazardous modes dominated by VSG, obtaining key eigenvalues ​​and improving the targeting of parameter tuning. This paper proposes several methods for improving the VSG's grid connection stability. First, by constructing an eigenvalue-parameter sensitivity matrix, the direction and intensity of parameter adjustment are quantified, solving the problem of blindly trying multiple parameters in traditional methods. Second, based on sensitivity-oriented search of critical boundaries, a stable and feasible region for continuous parameters is obtained through fitting, clarifying the safe tuning range of parameters. Finally, within the stable and feasible region, optimization is performed with the maximum safety margin as the objective to obtain the tuning values ​​of the tuning parameters. This enhances the VSG's ability to resist disturbances and parameter drift, improves system stability and power quality under weak grid conditions, and solves the problem that unreasonable preset values ​​of system parameters in the VSG control algorithm prevent the VSG from playing its role in grid stabilization. Attached Figure Description

[0035] Figure 1 This is a schematic flowchart illustrating a method for tuning VSG parameters for grid-type energy storage suitable for weak power grids, provided as an embodiment of this application. Detailed Implementation

[0036] Please see Figure 1 The diagram illustrates a flowchart of a method for tuning the parameters of a grid-type energy storage VSG suitable for weak power grids, according to an embodiment of this application. The method includes the following steps:

[0037] Step S100: Collect line parameters of the target distribution area, calculate the short-circuit ratio (SCR) of each node, construct a voltage-power sensitivity matrix, and screen candidate nodes based on the short-circuit ratio.

[0038] The short-circuit ratio (SCR) of an energy storage access node is the ratio of the grid short-circuit capacity at the energy storage access node to the rated capacity of the energy storage device. The lower the SCR value, the weaker the grid strength at the energy storage access node, and the more susceptible the node voltage and frequency are to power disturbances. Theoretically, this places a higher demand on the active support of grid-connected energy storage.

[0039] However, relying solely on the SCR value is insufficient to accurately determine the optimal deployment node for energy storage, as different nodes exhibit significant differences in their voltage and frequency response to power injection changes. Some nodes have low SCR values ​​and weak grid infrastructure, but due to their remote geographical location and position at the edge of the grid topology, injecting or adjusting power at these nodes has a very limited effect on improving the overall voltage quality of the entire distribution area. Conversely, some nodes have relatively high SCR values, but because they are located at the core of the grid, only a small amount of power injection is needed to effectively optimize the voltage operation quality of the entire distribution area. Therefore, it is necessary to conduct a multi-dimensional comprehensive evaluation of energy storage deployment nodes by combining the voltage-power sensitivity matrix of each node.

[0040] First, the line parameters of the target distribution area are collected. The line parameters specifically include distribution network topology data, typical daily load history data, distributed power generation capacity data, and equivalent impedance parameters of the upper-level power grid. Among them, the network topology data includes the total number of nodes, the total number of branches, the resistance and reactance of each branch, and the equivalent impedance parameters of the upper-level power grid are used to accurately calculate the short-circuit capacity of each node.

[0041] Based on the line parameters of the target distribution area, the short-circuit ratio (SCR) of all nodes is calculated one by one.

[0042] In power systems, the node short-circuit capacity is the ratio of the square of the system's rated line voltage to the node's equivalent self-impedance, and the ratio of the node's short-circuit capacity to its rated energy storage capacity is the dimensionless parameter SCR. Specifically, the industry-standard criteria are: a node with an SCR less than 3 but greater than or equal to 1.5 is considered a weak grid node; a node with an SCR less than 1.5 is considered an extremely weak grid node. These industry-standard criteria can be used as a priority for energy storage deployment.

[0043] Based on the line parameters of the target distribution substation, the Newton-Raphson power flow calculation method, which is commonly used in the power field, is adopted to extract the Jacobian matrix of the power flow calculation and invert it to construct the voltage-power sensitivity matrix.

[0044] Among them, the diagonal elements of the voltage-power sensitivity matrix of the transformer area are the self-sensitivity of the corresponding node, which characterizes the influence of node power injection on its own voltage; the off-diagonal elements are the mutual sensitivity, which characterize the coupling influence of node power injection on the voltage of other nodes; the Newton-Raphson power flow calculation method is a well-known technique and will not be described in detail.

[0045] The comprehensive score index for computing nodes is calculated using the following formula:

[0046] Represents a node The comprehensive scoring indicators; Represents a node The short-circuit ratio of SCR; The reference threshold for the short-circuit ratio is 3 in this embodiment. Represents a node The short-circuit ratio; This represents the maximum value normalization function, with the normalization range being the ratio of short-circuit ratios of all nodes; , These represent the first weight and the second weight, respectively. The sum of the first weight and the second weight is 1. In this embodiment, the values ​​of the first weight and the second weight are both 0.5. Represents the voltage-power sensitivity matrix of the first... The cumulative sum of column elements; This represents the maximum value among the cumulative sums of the elements in each column of the voltage-power sensitivity matrix.

[0047] node The short-circuit ratio is used to evaluate the node. The degree to which grid-based energy storage needs to provide active support. Used to evaluate nodes The importance of the voltage-power sensitivity matrix, the first The larger the cumulative sum of the column elements, the more significant the effect on the node. The more significant the effect of injected power on optimizing the overall voltage quality of the transformer area, the better.

[0048] The comprehensive score index of the node is calculated as formula (1).

[0049] The first preset number of nodes with the largest comprehensive score index are recorded as candidate nodes.

[0050] This embodiment takes the IEEE 13-node power system as an example. It pre-selects three candidate nodes to connect to the grid-type energy storage device controlled by the virtual synchronous machine, and the distributed power supply in the distribution area is configured as a single photovoltaic unit and a single wind turbine unit; that is, in this embodiment, the value of the first preset number is 3.

[0051] At this point, the first preset number of candidate nodes in the power system have been identified.

[0052] Step S200: Set up three types of interconnecting elements, construct mathematical models for the three types of interconnecting elements in a differentiated manner, solve for the steady-state operating point of the target distribution substation, solve for the partial derivatives of the mathematical models of the three types of interconnecting elements at the steady-state operating point, obtain the sub-matrices of each type of interconnecting element and the coupling sub-matrices between interconnecting elements, and then construct the system state matrix. Linearize the distribution system containing VSG, power grid and distributed power source according to the steady-state operating point, construct the standard state space model and solve for the complex eigenvalues, calculate the modal damping ratio, oscillation frequency and participation factor of the complex eigenvalues, select the key eigenvalues ​​among the complex eigenvalues ​​according to the modal damping ratio, oscillation frequency and participation factor, calculate the sensitivity corresponding to the key eigenvalues, and construct the sensitivity matrix.

[0053] Existing VSG parameter tuning methods generally rely on multiple sets of preset empirical parameters, adjusting them through repeated simulations. This empirical trial-and-error method lacks quantitative stability analysis support. When multiple VSGs are connected to the grid and the number of parameters to be tuned is large, it cannot clearly define the direction of parameter optimization and gain effect, resulting in low tuning efficiency, poor adaptability, and difficulty in adapting to the complex coupled conditions of weak power grids. Therefore, this application adopts a small-signal stability analysis method to conduct refined modeling of the target distribution substation area and construct state-space equations. Relying on matrix eigenvalue quantification, the overall stability characteristics of the system are determined, providing a clear optimization direction and stability domain constraints for VSG parameter tuning.

[0054] Based on Kundur's classic power system modeling theory, a state-space model of a weak power grid system with multiple VSGs is constructed. The target distribution substation system is decomposed into three types of interconnected elements for differentiated modeling, balancing modeling accuracy and computational efficiency.

[0055] The target distribution substation system is modeled as an interconnected system composed of three types of interconnected components. The first type is the VSG (Variable Sensing Grid Controller), which is the core parameter tuning object, and its internal dynamic characteristics directly affect the system sensitivity calculation results; therefore, refined dynamic modeling is adopted for the VSG. The second type is the power grid network component, which serves as a fixed constraint in the parameter tuning process; only its electrical coupling effect with the VSG needs to be accurately characterized, therefore, a simplified modeling approach is used for the power grid network component. The third type is the distributed generation source. The output dynamic response rate of the distributed generation source is much lower than the control and regulation rate of the VSG, and it belongs to the external disturbance source of the system; therefore, a quasi-steady-state assumption is adopted for simplified modeling. The specific modeling and analysis steps are as follows.

[0056] Step S201: Differentiate and construct mathematical models for the three types of interconnect components.

[0057] The first type of interconnected element is a VSG deployed at a first preset number of candidate nodes. The state parameters of each VSG include power angle, angular velocity deviation, and dq-axis output current. The VSG simulates the rotor motion characteristics and primary frequency regulation capability of a synchronous generator through a power frequency controller. Based on the second-order rotor motion equation, a virtual swing equation of the active power-frequency loop is constructed to achieve active support of the grid frequency.

[0058] The active-frequency loop virtual swing equation is denoted as calculation formula (2). The specific active-frequency loop virtual swing equation is as follows:

[0059] in, Represents the virtual moment of inertia; Indicates angular velocity deviation; Indicates time; Indicates the virtual damping coefficient; This indicates the output power deviation of the VSG inverter; Indicates the active power droop coefficient; This represents the reference angular frequency generated internally by the VSG, i.e., the rotational speed of the virtual rotor; This indicates the angular velocity corresponding to the rated synchronous angular frequency, which is the standard frequency of the power grid.

[0060] The virtual swing equation characterizes the inertia and damping regulation characteristics of the VSG. Virtual moment of inertia is used to prevent sudden changes in system frequency; the larger the virtual moment of inertia, the slower the rate of frequency change and the smoother the dynamic response of power balance. Angular velocity deviation is the difference between the actual virtual angular frequency output by the VSG and the rated synchronous angular frequency of the system. When the angular velocity deviation is greater than zero, the VSG virtual rotor is in an overspeed state, and the unit outputs active power according to the synchronous machine's frequency regulation characteristics. When the angular velocity deviation is less than zero, the VSG virtual rotor is in an underspeed state, and the unit reduces its active power output, achieving dynamic balance regulation of the grid's active power and frequency. Virtual damping is used to suppress frequency oscillation attenuation; the larger the virtual damping coefficient, the faster the frequency oscillation convergence speed, but an excessively large virtual damping coefficient will increase the system's steady-state frequency deviation. The active power droop coefficient is used to quantify the active power regulation gain of primary frequency regulation; the larger the active power droop coefficient, the more active the frequency support response, but the more likely it is to induce system power oscillations.

[0061] This is a virtual damping power term used to suppress virtual rotor speed oscillations. The power feedback term represents the output power deviation of the VSG inverter. It is calculated by multiplying the d-axis voltage and d-axis current in the dq coordinate system and the q-axis voltage and q-axis current in the dq coordinate system. The sum of the two products is the measured output power of the inverter. The difference between the measured output power of the inverter and the steady-state operating reference active power of the VSG is denoted as the output power deviation of the VSG inverter. Corresponding to the primary frequency droop adjustment stage, the active power output is corrected in real time based on the speed deviation.

[0062] Establish the state transition equation for the deviation of power angle and angular velocity, establish the correlation between power angle in the spatial domain and frequency in the time domain, and denote the state transition equation for the deviation of power angle and angular velocity as calculation formula (3). The specific state transition equation for the deviation of power angle and angular velocity is as follows:

[0063] in, The 'x' angle represents the spatial position angle of the VSG virtual rotor.

[0064] To avoid secondary voltage surges caused by drastic fluctuations in reactive power, this application abandons instantaneous proportional control and uses a first-order inertial element to construct a reactive power-voltage regulation model. The reactive power regulation response speed is constrained by a time constant, and the reactive power droop coefficient is used to quantify the reactive power regulation intensity corresponding to the voltage deviation.

[0065] The reactive power-voltage regulation model is denoted as calculation formula (4), and the specific reactive power-voltage regulation model is as follows:

[0066] This represents the reactive power regulation inertia time constant; It represents the reactive power droop factor, which determines the intensity of the reactive power response caused by voltage deviation; This indicates the voltage amplitude at the output port of the VSG inverter; This indicates the internal reference voltage used for the inverter's inner loop control. This indicates reactive power deviation.

[0067] The reactive power regulation inertia time constant is used to control the speed of reactive power regulation response; the larger the reactive power regulation inertia, the slower the reactive power regulation response. The reactive power droop coefficient is used to determine the reactive power regulation output strength corresponding to the voltage deviation. The VSG inverter output port is the PCC point. Based on the 5% voltage over-limit standard for the distribution system, the internal reference voltage used for the inverter's inner loop control is 105% of the rated voltage of the weak grid, serving as the upper limit for voltage regulation. The reactive power deviation is the difference between the actual reactive power output of the VSG and the steady-state reference reactive power.

[0068] The second type of interconnecting element is the power grid network element. This element includes transmission lines and distribution transformer loads, and only handles power transmission and loss; it has no power output. Transmission lines use standard power system technologies. The system uses equivalent circuit modeling and participates in the construction of the system state matrix through node admittance matrix and power flow Jacobian matrix. The transformer area load adopts the ZIP hybrid load model, which integrates three types of load characteristics: constant impedance, constant current, and constant power. This solves the simulation distortion problem of single load model under the scenario of large voltage fluctuation in weak power grid. The partial derivative terms of the static characteristics of load voltage participate in the linearization construction of the system state matrix.

[0069] in, The equivalent circuit and ZIP load model are both common models for power grid stability analysis and are well-known technologies. The specific modeling process will not be described in detail here.

[0070] The third type of interconnecting element consists of distributed photovoltaic and wind power sources, which serve as external disturbance sources for the system.

[0071] The output of distributed power sources changes slowly on a second or minute basis, much slower than the millisecond-level control response speed of VSG. Therefore, a quasi-steady-state power injection model is adopted. In this embodiment, a photovoltaic power source is used as an example to calculate the active power of the distributed power source. The calculation formula for the active power of the distributed power source is denoted as calculation formula (5). The calculation formula for the active power of the distributed power source is as follows:

[0072] in, Indicates the time of distributed power sources The active power; This represents the reference steady-state power of the distributed power source. Indicates time Irradiance, This represents the irradiance of a distributed power source under reference steady-state conditions. Represents the power temperature coefficient, with units of . The value of the power temperature coefficient is obtained by looking up a table according to the specifications of the distributed power source; Indicates the time of distributed power sources The battery temperature; This represents the average temperature of the battery when the distributed power source generates electricity under the reference steady-state condition.

[0073] Under quasi-steady-state conditions, the distributed generation operates with unity power factor, and its reactive power output is always zero. The real-time active power of the modeled distributed generation is used as the exogenous disturbance variable of the system to construct the input disturbance vector of the state-space equation.

[0074] Step 202: Solve for the steady-state operating point of the system.

[0075] Before linearizing the system using a nonlinear model, power flow calculations are performed on the target distribution area to determine the steady-state operating point x0 of the system. The process of power flow calculation is a well-known technique and will not be described in detail here.

[0076] The inputs for power flow calculation include: network topology and line parameters of the target distribution area, node operating parameters, distributed generation output parameters, VSG steady-state operating parameters, weak grid short-circuit ratio, and equivalent impedance parameters. Network topology and line parameters include line resistance, line open-circuit point, line-to-ground admittance, and node admittance matrix; node operating parameters include slack node voltage, active and reactive power output of PV nodes, target voltage of PV nodes, and active and reactive loads of PQ nodes; distributed generation output parameters include active power and power source state of charge (SOC); and VSG steady-state operating parameters include... , , and .

[0077] The output of the power flow calculation includes: the steady-state operating point of the system, the steady-state voltage and phase angle of each node in the weak grid at the steady-state operating point, the steady-state power angle, angular velocity deviation, and dq-axis current of the VSG, and the steady-state active power output of the distributed generation.

[0078] Step 203: Solve for the partial derivatives at the steady-state operating point and assemble the system state matrix.

[0079] Based on the nonlinear mathematical model of three types of interconnecting elements, partial derivatives are solved at the steady-state operating point to obtain sub-matrices for each type of interconnecting element and coupling sub-matrices between interconnecting elements. The specific steps are as follows.

[0080] For the VSG state parameters, the virtual rocking equation of calculation formula (2) is transformed to obtain the calculation relationship of the rate of change of angular velocity deviation, which is denoted as calculation formula (6). The specific calculation relationship of the rate of change of angular velocity deviation is as follows:

[0081] At the steady-state operating point x0, partial derivatives are calculated sequentially for four types of VSG state parameters: power angle, angular velocity deviation, d-axis output current, and q-axis output current. These four partial derivative results together constitute the VSG submatrix. The first row of elements; the steady-state power angle value, the d-axis voltage differential term at the steady-state point, and the q-axis voltage differential term at the steady-state point are all calculated by power flow and output at the steady-state operating point.

[0082] Taking the partial derivative of equation (3) with respect to the angular velocity deviation, and setting the partial derivatives of the other state parameters to zero, we obtain the VSG submatrix. The second row of elements. VSG submatrix. For a square matrix, the VSG submatrix The number of rows equals the total number of VSGs deployed; VSG submatrix The number of columns is equal to the number of status parameters of a single VSG, and the row and column variables are sorted sequentially according to the status parameters of all VSGs.

[0083] Using the same linearization method, partial derivatives are calculated for the model of the second type of network element. For the two types of electrical variables—node voltage and phase angle—partial derivatives are solved at the steady-state point to generate the network submatrix. Network submatrix For a square matrix, the network submatrix The number of rows and columns is twice the number of nodes.

[0084] For the calculation formula (5) corresponding to the third type of interconnection element, the partial derivatives of the distributed source are solved at the steady-state operating point based on the state parameters of the distributed source to obtain the distributed source sub-matrix. Distributed power supply submatrix It is a second-order square matrix, and the two state parameters are the active power output and reactive power output of the distributed power source, respectively.

[0085] Furthermore, after solving the independent submatrices for the three types of components, partial derivatives are calculated for the electrical coupling relationships between different components to construct coupling submatrices for different types of interconnected components:

[0086] 1. Coupling submatrix of network state to VSG state The coupling submatrix of the network state to the VSG state is obtained by taking partial derivatives of the network element equations with respect to all VSG state parameters. The number of rows in the coupling submatrix of the network state to the VSG state is equal to the number of VSG state parameters, and the number of columns in the coupling submatrix of the network state to the VSG state is equal to twice the number of nodes. Since only the VSG access nodes have dq-axis current-voltage coupling components, the coupling submatrix of the network state to the VSG state is a sparse matrix, with non-zero values ​​only in the columns corresponding to the VSG deployment nodes.

[0087] 2. Coupled submatrix of VSG state to network state The VSG control equations are derived by taking the partial derivatives of the dq-axis current components with respect to the network voltage and phase angle parameters. The number of rows in the coupling submatrix of the VSG state to the network state is twice the number of nodes, and the number of columns is equal to the number of VSG state parameters. The node power balance equations only include VSG power injection at VSG access nodes, and only the corresponding node rows have non-zero values.

[0088] 3. Coupling sub-matrix of network state to distributed generation The value is obtained by taking the partial derivative of the grid voltage variable with respect to the output parameters of the distributed generation. The number of rows in the coupling submatrix of the VSG state to the network state is equal to twice the number of nodes, and the number of columns is equal to the number of VSG state parameters. The values ​​in the coupling submatrix of the VSG state to the network state represent the disturbance effect of the distributed generation output fluctuation on the voltage and phase angle of all nodes in the network.

[0089] 4. Coupling submatrix of distributed source state to network state The value is obtained by taking the partial derivative of the active power output of the distributed generation with respect to the node electrical parameters. The number of rows in the coupling sub-matrix of the distributed generation state to the network state is equal to the number of VSG state parameters, and the number of columns is equal to twice the number of nodes. The values ​​in the coupling sub-matrix of the distributed generation state to the network state represent the constraints of grid voltage and phase fluctuations on the output characteristics of the distributed generation.

[0090] VSG and distributed power sources have no direct electrical coupling loop. The cross-coupling positions of the two correspond to two sets of zero matrices, and the row and column dimensions of the zero matrices match the dimensions of the adjacent sub-matrices.

[0091] The system integrates sub-matrices containing all interconnected components with coupling sub-matrices containing different types of interconnected components, and assembles the complete system state matrix in a hierarchical and modular manner. Specifically:

[0092] Step 204: Perform system linearization processing and construct standard state-space equations.

[0093] At the steady-state operating point, the nonlinear dynamic equations of the system are linearized globally to obtain the state-space equations. The state-space equations are as follows:

[0094] In this process, the state vector is obtained by sequentially arranging the VSG state variables, grid state variables, and distributed generation state variables as the first, second, and third row elements, respectively. Represents the state vector; Represents the system state matrix; The matrix representing the partial derivatives of the system's nonlinear equations with respect to the input vector; This represents the perturbation input vector.

[0095] The system state matrix characterizes the coupling strength of the state variables within the system. The eigenvalues ​​of the system state matrix directly determine the system stability. Elements in the system state matrix represent the influence of the change in the state variable corresponding to the row of the element on the rate of change of the state variable corresponding to the column of the element. The partial derivative matrix of the system's nonlinear equations with respect to the input vector characterizes the relationship between external disturbance inputs and the rates of change of the system's state variables. Elements in the partial derivative matrix of the system's nonlinear equations with respect to the input vector represent the influence weight of the disturbance input corresponding to the row of the element on the system state variable corresponding to the column of the element.

[0096] This application employs a system of differential-algebraic equations to achieve overall linearization at the steady-state operating point, and completes the matrix solution through a three-step solution method that eliminates algebraic variables. The solution process is a well-known prior art and will not be elaborated further. The state variables of the differential-algebraic equation system include VSG state parameters, network state parameters, and distributed source state parameters. The algebraic variables are the voltage magnitudes and phase angles of each node, and the system input is the external disturbance input vector. The partial derivative matrix of the system's nonlinear equations with respect to the input vector exhibits a block-diagonal characteristic; the off-diagonal submatrices of cross-class coupling are all zero matrices, with non-zero values ​​existing only in the diagonal subblocks.

[0097] The disturbance input vector contains all typical disturbance components of the distribution substation, specifically including: VSG active power deviation disturbance, VSG reactive power deviation disturbance, network load active power disturbance, network load reactive power disturbance, photovoltaic power output disturbance, and wind power output disturbance, comprehensively covering internal and external disturbances during the operation of a weak power grid.

[0098] Step 205: Eigenvalue solving and key oscillation mode identification.

[0099] Solve for all complex eigenvalues ​​of the system state matrix, which include both real and imaginary parts. Calculate the modal damping ratio and oscillation frequency based on these complex eigenvalues.

[0100] Specifically, the square root of the sum of the squares of the real and imaginary parts of the complex eigenvalues ​​is used as the denominator, and the opposite of the real part of the complex eigenvalues ​​is used as the numerator. The value of the fraction formed by the numerator and denominator is denoted as the modal damping ratio of the complex eigenvalues; the imaginary part of the complex eigenvalues ​​and 2... The ratio of to is denoted as the oscillation frequency of the complex eigenvalue. Wherein, This represents pi, and for ease of calculation, it is taken as 3.14.

[0101] The smaller the damping ratio, the closer the corresponding complex eigenvalue is to the imaginary axis, the slower the oscillation decay rate, and the higher the risk of system instability; the larger the oscillation frequency, the more dangerous the oscillation, and the higher the risk of system instability.

[0102] Based on the left and right eigenvectors corresponding to the complex eigenvalues, the modal participation factor of each state variable within the left and right eigenvectors corresponding to the complex eigenvalues ​​is calculated to quantify the contribution of a single state variable to the oscillation mode. Specifically, the product of the absolute values ​​of the same state variable in the left and right eigenvectors of the complex eigenvalues ​​is denoted as the eigenproduct of the same state variable, and the ratio of the eigenproduct of the state variable to the sum of the eigenproducts of all state variables is denoted as the modal participation factor of the state variable.

[0103] The sum of the modal participation factors of the state variables corresponding to the complex eigenvalues ​​of the four state parameters of the VSG is denoted as the participation factor of the complex eigenvalues. Here, the four state parameters of the VSG are the four VSG state parameters.

[0104] Based on the modal damping ratio, oscillation frequency, and participation factor of complex eigenvalues, identification conditions are set, and complex eigenvalues ​​that satisfy any one of the identification conditions are recorded as key eigenvalues.

[0105] The specific identification criteria are as follows:

[0106] Identification condition 1: The modal damping ratio is less than a first preset threshold. The empirical range of the first preset threshold is 0.1 to 0.15, and in this embodiment, the first preset threshold is set to 0.1. The modal damping ratio of conventional power system oscillations is concentrated in the range of 0.02 to 0.1. Under weak grid conditions, the coupling effect between the VSG and the grid is aggravated, and the risk of oscillation instability is higher. Therefore, this embodiment selects the upper limit of the range as the screening threshold, which can cover the model uncertainty caused by low short-circuit ratio conditions.

[0107] Identification condition 2: Modal oscillation frequency is greater than or equal to 0.1Hz and less than or equal to 2Hz. The active power control bandwidth of the VSG is 2Hz to 10Hz, and the reactive power control bandwidth is 5Hz to 20Hz. Among them, the system interval oscillation frequency is concentrated in 0.1Hz to 0.7Hz, and the unit local oscillation frequency is concentrated in 0.7Hz to 2Hz. Oscillation in this frequency band will directly disturb the voltage and frequency at the grid connection point and is strongly coupled with the internal state parameters of the VSG, so it needs to be identified carefully.

[0108] Identification Condition 3: The participation factor is greater than the second preset threshold. Identification Condition 3 can filter oscillation modes that are strongly correlated with VSG state parameters. The empirical range of the second preset threshold is 0.3 to 0.5, and in this embodiment, the value of the second preset threshold is 0.3. The system includes 4 VSG state parameters, 4 network state parameters, and 4 distributed power source state parameters. If the elements of the feature vector are evenly weighted, the theoretical proportion of VSG state parameters is 4 divided by all 12 state parameters, i.e., 0.33. In this embodiment, a threshold slightly lower than the theoretical average of 0.3 is selected as the threshold, which can completely retain oscillation modes that are correlated with VSG and avoid missed detection.

[0109] Step 206: Construct the eigenvalue-VSG parameter sensitivity matrix.

[0110] The system state matrix changes with the VSG tuning parameters, leading to shifts in eigenvalues ​​and eigenvectors. For each key eigenvalue, its partial derivative with respect to the core VSG tuning parameters is calculated. Using the classical eigenvalue sensitivity calculation formula, the parameter sensitivity is obtained. The core VSG tuning parameters include virtual moment of inertia, virtual damping coefficient, active droop coefficient, and reactive droop coefficient.

[0111] The sign of the real part of the sensitivity calculation result determines the direction of parameter adjustment: a positive real part means that increasing the corresponding VSG parameter will shift the eigenvalue to the right half of the complex plane, reducing system stability; a negative real part means that increasing the corresponding VSG parameter will shift the eigenvalue to the left half of the complex plane, effectively improving system modal stability. The amplitude of the sensitivity characterizes the parameter's ability to regulate the oscillation mode; the larger the amplitude, the more significant the optimization effect of parameter tuning; if the amplitude approaches zero, the parameter has no regulating effect on the target mode, and the tuning priority can be reduced.

[0112] Based on the sensitivity corresponding to all key eigenvalues, a sensitivity matrix is ​​constructed. The number of rows in the sensitivity matrix is ​​equal to the number of key eigenvalues, and the number of columns of key eigenvalues ​​is equal to the number of system parameters of the VSG, i.e., the number of columns of key eigenvalues ​​is 4.

[0113] Among them, the sensitivity matrix can quantitatively guide the adjustment direction and adjustment range of multi-parameter collaborative tuning, solving the technical problems of strong blindness and lack of quantitative optimization basis in traditional empirical tuning.

[0114] At this point, the sensitivity matrix is ​​obtained.

[0115] Step S300: Search for the stable boundary along the sensitivity direction within the sensitivity matrix and apply damping ratio constraints to construct the stable feasible region of the combination of tuning parameters. Using the principle of maximizing safety margin, complete the tuning of the tuning parameters based on the damping ratio of all tuning parameters in the combination of tuning parameters within the stable feasible region.

[0116] The sensitivity matrix only indicates the direction of the effect of VSG parameter adjustments on system stability, but cannot determine the allowable range of parameter adjustments. Therefore, based on the direction of change provided by the sensitivity, the critical boundary of system instability is searched along each direction to establish the parameter stability domain.

[0117] First, a set of initial values ​​for tuning parameters are preset as the search starting point. The tuning parameters include virtual moment of inertia, virtual damping coefficient, active droop coefficient, and reactive droop coefficient. Specifically: the value of virtual moment of inertia is greater than or equal to 2s and less than or equal to 5s, and the initial value of virtual moment of inertia is taken as the empirical value of 4s; the value of virtual damping coefficient is greater than or equal to 10pu and less than or equal to 20pu, and the initial value of virtual damping coefficient is taken as the empirical value of 20pu; the initial values ​​of active droop coefficient and reactive droop coefficient are taken as empirical values ​​of 0.05 and 0.1, respectively; the initial values ​​of virtual moment of inertia, virtual damping coefficient, active droop coefficient, and reactive droop coefficient are used as the search starting point.

[0118] For each key eigenvalue, a one-dimensional scan is performed along the sensitivity direction of the key eigenvalue to each tuning parameter to find the instability critical parameter where the real part of the eigenvalue is equal to zero. The scanning iteration formula is as follows:

[0119] in, This indicates the value of the tuning parameters after scanning; This indicates the value of the tuning parameter at the starting point of the search; Indicates the scan step size increment; Indicates the scan step size; Represents a symbolic function; Indicates key feature values; This indicates the value of the tuning parameter.

[0120] The tuned parameter values ​​after the scan are the tuned parameter values ​​after a single iteration; the scan step size, i.e., the step index, is used to control the iteration variables during the scan; the scan step size controls the amount of change in the tuned parameters during each iteration; using a sign function ensures that the parameters follow the specified path. Increased direction search, that is, searching in the direction where the modes in the state space tend to be unstable, in order to find the boundary value.

[0121] Based on the preset upper and lower limits of four tuning parameters, the extreme endpoints of the parameter space are determined. The central sampling point is determined by combining the mean of the upper and lower limits, and multiple sets of random sampling points are added. A one-dimensional scan is performed starting from all sampling points, with each starting point corresponding to a connected stable interval. All critical boundary points obtained from the scans are merged to form a discrete critical point set. .

[0122] The number of additional random sampling points should be greater than or equal to 20 and less than or equal to 40. In this embodiment, the number of random sampling points is 30.

[0123] Existing techniques such as polynomial fitting and spline interpolation are used to perform surface fitting on all discrete critical points in the discrete critical point set to generate a continuous stable feasible region. The constraint condition for the stable feasible region is that the damping ratio of all key characteristic values ​​corresponding to any set of parameters within the stable feasible region is not lower than the minimum damping ratio threshold. In this embodiment, the minimum damping ratio threshold is set to 0.1.

[0124] It should be noted that when screening key feature values, the damping ratio threshold was set to 0.1 to screen dangerous modes that already have low-frequency oscillations under the initial parameters. Here, the minimum damping ratio threshold is set to 0.1, so that the constraint damping ratio is greater than 0.1. This sets a hard requirement on the stability standard after parameter tuning, ensuring that the system has no weakly damped oscillations after parameter adjustment.

[0125] All parameter combinations within the stable feasible region can meet the small-signal stability requirements of the system. The parameter tuning objective is transformed into finding the parameter combination that is furthest from the instability boundary within the feasible region. The parameter points are more resistant to parameter drift and model error, and have the best robustness.

[0126] For any combination of tuning parameters within the stable feasible region, the minimum damping ratio among all tuning parameters in the combination is denoted as the characteristic safety margin of the corresponding combination of tuning parameters. The absolute value of the difference between the characteristic safety margin of the combination of tuning parameters and the nearest boundary value within the stable feasible region is used as the numerator, and the range of the characteristic safety margins of all determinable combinations of tuning parameters within the stable feasible region is used as the denominator. The value of the fraction formed by the numerator and denominator is denoted as the safety margin of the combination of tuning parameters.

[0127] The minimum and maximum values ​​of the damping ratios of the combinations of all determinable tuning parameters within the stable feasible region are denoted as the minimum damping ratio and the maximum damping ratio, respectively. For any combination of tuning parameters within the stable feasible region, the difference between the minimum damping ratio of all tuning parameters within the combination and the minimum damping ratio is taken as the numerator, and the difference between the maximum damping ratio and the minimum damping ratio is taken as the denominator. The value of the fraction formed by the numerator and the denominator is denoted as the damping ratio margin of the combination of tuning parameters.

[0128] A uniform grid search method is used to traverse the stable feasible region: the feasible intervals of the four types of tuning parameters are equally divided and sampled, with the number of sampling points in a single dimension set to 30, generating four-dimensional grid nodes; each four-dimensional grid node is checked to see if it is within the stable feasible region; for nodes within the stable feasible region, the parameter boundary distance and damping ratio margin are calculated respectively, and the grid node corresponding to the maximum value of the safety margin is selected; the virtual moment of inertia, virtual damping coefficient, active droop coefficient, and reactive droop coefficient corresponding to the grid node corresponding to the maximum value of the safety margin are taken as the final values ​​of the tuning parameters, thus completing the tuning of the energy storage VSG parameters, which are the tuning parameters.

[0129] This completes the tuning of the VSG parameters for grid-type energy storage in weak power grids.

Claims

1. A method for tuning the parameters of a grid-connected energy storage (VSG) system suitable for weak power grids, characterized in that, The method includes the following steps: Collect line parameters of the target distribution area, calculate the short-circuit ratio (SCR) of each node, construct a voltage-power sensitivity matrix, and screen candidate nodes based on the short-circuit ratio; Three types of interconnecting elements are set up, and mathematical models of the three types of interconnecting elements are constructed differently. The steady-state operating point of the target distribution substation is solved. The partial derivatives of the mathematical models of the three types of interconnecting elements are solved at the steady-state operating point to obtain the sub-matrices of each type of interconnecting element and the coupling sub-matrices between interconnecting elements. Then, the system state matrix is ​​constructed. The distribution system containing VSG, power grid and distributed power source is linearized according to the steady-state operating point. The standard state space model is constructed and the complex eigenvalues ​​are solved. The modal damping ratio, oscillation frequency and participation factor of the complex eigenvalues ​​are calculated. Based on the modal damping ratio, oscillation frequency and participation factor, the key eigenvalues ​​in the complex eigenvalues ​​are selected. The sensitivity corresponding to the key eigenvalues ​​is calculated and the sensitivity matrix is ​​constructed. The stability boundary is searched along the sensitivity direction within the sensitivity matrix and damping ratio constraints are applied to construct the stable feasible region of the combination of tuning parameters. Using the principle of maximizing safety margin, the tuning parameters are tuned according to the damping ratio of all tuning parameters in the combination of tuning parameters within the stable feasible region. The method for constructing the system state matrix is ​​as follows: The sub-matrices of VSG, grid network elements, and distributed generation are respectively denoted as... , and ; The coupling sub-matrix between interconnected elements includes the coupling sub-matrix of network state to VSG state, the coupling sub-matrix of VSG state to network state, and the coupling sub-matrix of network state to distributed generation, which are denoted as follows: , and ; System state matrix , This represents the coupling submatrix of distributed power sources to network states.

2. The method for tuning VSG parameters of grid-type energy storage suitable for weak power grids according to claim 1, characterized in that, The method for selecting candidate nodes is as follows: The comprehensive score index of a node is calculated based on the short-circuit ratio (SCR) of the node and the sum of the column elements of the corresponding node in the voltage-power sensitivity matrix. The first preset number of nodes with the highest comprehensive score index are recorded as candidate nodes.

3. The method for tuning VSG parameters for grid-type energy storage suitable for weak power grids according to claim 2, characterized in that, The formula for calculating the comprehensive score index of the node is: Represents a node The comprehensive scoring indicators; Represents a node The short-circuit ratio of SCR; This represents the preset threshold for the short-circuit ratio; This represents the maximum value normalization function; , These represent the first weight and the second weight, respectively, and the sum of the first weight and the second weight is 1. Represents the voltage-power sensitivity matrix of the first... The sum of column elements; This represents the maximum value among the cumulative sums of the elements in each column of the voltage-power sensitivity matrix.

4. The method for tuning VSG parameters for grid-type energy storage suitable for weak power grids according to claim 1, characterized in that, The standard state-space model is as follows: In this process, the state vector is obtained by sequentially arranging the VSG state variables, grid state variables, and distributed generation state variables as the first, second, and third row elements, respectively. Represents the state vector; Represents the system state matrix; The matrix representing the partial derivatives of the system's nonlinear equations with respect to the input vector; Represents the perturbation input vector; Indicates time.

5. The method for tuning VSG parameters for grid-type energy storage suitable for weak power grids according to claim 1, characterized in that, The modal damping ratio of the complex eigenvalues ​​and the oscillation frequency are calculated as follows: The arithmetic square root of the sum of the squares of the real and imaginary parts of the complex eigenvalues ​​is used as the denominator, and the opposite of the real part of the complex eigenvalues ​​is used as the numerator. The value of the fraction formed by the numerator and denominator is denoted as the modal damping ratio of the complex eigenvalue. The imaginary part of the complex eigenvalue is combined with 2. The ratio of is denoted as the oscillation frequency of the complex eigenvalue.

6. The method for tuning VSG parameters for grid-type energy storage suitable for weak power grids according to claim 1, characterized in that, The calculation method for the participation factor is as follows: The product of the absolute values ​​of the same state variable in the left and right eigenvectors of the complex eigenvalues ​​is denoted as the eigenproduct of the same state variable. The ratio of the eigenproduct of the state variable to the sum of the eigenproducts of all state variables is denoted as the modal participation factor of the state variable. The sum of the modal participation factors of the state variables corresponding to the complex eigenvalues ​​of the four state parameters of the VSG is denoted as the participation factor of the complex eigenvalues. The state parameters of the VSG include power angle, angular velocity deviation, and dq axis output current.

7. The method for tuning VSG parameters of grid-type energy storage suitable for weak power grids according to claim 1, characterized in that, The method for selecting the key feature values ​​is as follows: Identification condition 1: Modal damping ratio is less than the first preset threshold; Identification condition 2: Modal oscillation frequency is greater than or equal to 0.1 Hz and less than or equal to 2 Hz; Identification condition 3: The participation factor is greater than the second preset threshold; The complex feature value that satisfies any one of the recognition conditions is recorded as the key feature value.

8. The method for tuning VSG parameters for grid-type energy storage suitable for weak power grids according to claim 1, characterized in that, The tuning parameters include virtual moment of inertia, virtual damping coefficient, active droop coefficient, and reactive droop coefficient.

Citation Information

Patent Citations

  • Multi-target parameter setting method for networking fan controller

    CN120999675A

  • Hybrid energy storage virtual synchronous machine adaptive control method based on power grid state

    CN121813492A