Frequency support method for multi-machine parallel operation system of grid-connected converter based on improved particle swarm

Abstract

This invention discloses a frequency support method for a multi-machine parallel grid converter system based on an improved particle swarm optimization (PSO) algorithm. The method includes: maximizing the damping ratio of the system's dominant oscillation mode as the optimization objective; using an improved PSO algorithm to solve for the optimal initial values ​​of virtual inertia and damping coefficients for each VSG unit; the improved PSO algorithm balances global exploration and local convergence through nonlinearly decreasing inertia weights, strengthens individual cognition in the early stages of iteration and group cognition in the later stages through dynamic switching of learning factors, and maintains population diversity through a hybrid mechanism of Gaussian and Cauchy mutations; when the system experiences disturbances, the system's operating status is monitored to obtain the change in system angular frequency and the rate of change of angular frequency, and the virtual inertia and damping coefficients of each VSG unit are adjusted collaboratively to suppress frequency overshoot and accelerate oscillation convergence. This addresses the problems of severe frequency overshoot, slow oscillation convergence, and stability issues caused by multi-machine interaction.

Description

Technical Field

[0001] This invention relates to the field of power system operation and control technology, specifically to a frequency support method for a multi-machine parallel grid converter system based on improved particle swarm optimization. Background Technology

[0002] Modern power systems are evolving rapidly, characterized by increasing penetration of new energy sources and deepening power electronics. However, the integration of numerous power electronic devices into the traditional power grid has raised new stability issues, particularly the inherent low inertia and low damping characteristics of converters, which challenge the grid's frequency and voltage support capabilities.

[0003] Grid-based control, due to its ability to independently support system frequency and voltage, has attracted extensive research from scholars both domestically and internationally. Among these studies, the Virtual Synchronous Generator (VSG) control within grid-based converter control has garnered significant attention. Its core idea is to incorporate a synchronous generator model into the inverter's control algorithm, enabling the inverter to simulate the rotor inertia, damping, and external characteristics of a synchronous generator, such as primary frequency regulation and reactive power regulation. This allows the inverter to "actively" provide the necessary inertia and damping support to the grid. This enables distributed generation to participate in grid frequency and voltage regulation like traditional synchronous generators, becoming an effective solution for achieving renewable energy integration and improving grid stability. However, limited by the physical characteristics of power electronic devices, the disturbance rejection capability and overload tolerance of VSG systems still lag significantly behind those of traditional synchronous generators. Currently, scholars both domestically and internationally have conducted relatively complete research on the small-disturbance stability of grid-based converter single-unit grid-connected systems. Existing studies have established state-space models of grid-based converter single-unit grid-connected systems, including power loops, voltage loops, delay elements, and circuit components. Some literature uses harmonic linearization to establish the sequence impedance model of the grid-type converter and compares and analyzes the differences in sequence impedance characteristics between the grid-type and line-connected converters. Other literature uses the impedance ratio matrix formed by the grid impedance and the converter impedance to derive a stability criterion and points out that the grid-type converter may cause low-frequency oscillations in the system when connected to a strong grid. Still other literature uses the phase diagram method to combine voltage amplitude dynamics with power control loop dynamics to perform transient stability analysis on grid-type converters using single-loop voltage amplitude control. Some literature uses the equal area rule to analyze the frequency oscillation process during faults. In summary, there is currently a considerable amount of research on controller design and modeling, as well as adaptive parameter optimization, for VSG single-machine systems.

[0004] When multiple grid-connected inverters are connected to the same grid point via connecting impedances, complex interactions arise between the multi-unit parallel grid-connected inverter systems, which can easily lead to new stability problems. Preliminary research has been conducted on the stability analysis of multi-unit grid-connected inverter systems. Some literature studies the small-signal stability of multi-unit parallel grid-connected inverter systems. Other literature analyzes the impact of virtual inertia and line impedance matching on system stability and proposes an adaptive compensation method, but its effect on dynamic performance optimization is poor. Some literature proposes an adaptive virtual impedance control strategy suitable for multiple VSG parallel connections, achieving dynamic adjustment of virtual impedance and the effects of equalizing output reactive power and reducing circulating current, but it also fails to consider dynamic characteristics. Still other literature derives a small-signal model for VSG dual-unit parallel systems, analyzes the parameter response to the system, and proposes inertia matching control strategies and parameter self-optimization control strategies to improve system frequency stability. However, its analysis assumes that the parameters of the two VSGs are identical and change according to the same pattern, thus failing to consider the interaction between the two VSGs under different parameter conditions. Summary of the Invention

[0005] To address the shortcomings and deficiencies of existing technologies, this invention provides a frequency support method and system for multi-machine parallel grid-connected converter systems based on an improved particle swarm optimization algorithm. It aims to solve the pain points of power system inertia and damping caused by large-scale integration of new energy sources, severe frequency overshoot, slow oscillation convergence, and stability issues arising from multi-machine interaction in multi-machine parallel grid-connected converter systems. Simultaneously, it compensates for the insufficient disturbance rejection and overload tolerance of traditional virtual synchronous generator (VSG) systems.

[0006] This invention first constructs a closed-loop small-signal model suitable for multi-unit parallel operation scenarios of grid-connected converters: the dq reference coordinate axis of one VSG unit is selected as the common synchronous coordinate system, and the state variables of the other VSG units in their own coordinate systems are unified to the common synchronous coordinate system through coordinate transformation. The load at the grid connection point is introduced to construct the coupling relationship between the multi-unit and the grid, forming a closed-loop state-space model covering the VSG state of the multi-unit, the grid current, and the angle difference between the grid and the common synchronous coordinate system, thereby clarifying the influence of virtual inertia and damping coefficient on system stability.

[0007] To optimize the steady-state control parameters of the system, this invention designs an improved particle swarm optimization algorithm that integrates multiple mechanisms: by adjusting the inertia weight nonlinearly as the iteration progresses, the algorithm balances the global exploration capability of virtual inertia and damping coefficients of multiple machines in the early stage and the local fine convergence capability in the later stage; by dynamically switching the learning factor, the algorithm strengthens the particles' cognition of their own historical optimal parameters in the early stage of iteration, and shifts to strengthening the learning of the global optimal parameters of the swarm in the later stage; for stagnant particles that have not updated their individual optimal solutions in consecutive iterations, Gaussian mutation (to achieve local fine search of parameters) or Cauchy mutation (to escape local optima to expand the search space) is triggered according to a preset probability, and finally, with the goal of maximizing the damping ratio of the dominant oscillation mode of the system, the optimal initial values ​​of virtual inertia and damping coefficients of each VSG unit are solved.

[0008] At the system transient control level, this invention, combined with the transient power frequency characteristics of VSG multi-machine parallel systems, proposes an adaptive coordinated adjustment strategy for virtual inertia and damping coefficient: When the system encounters load disturbances, the change in system angular frequency and the rate of change of angular frequency are monitored and acquired in real time. Based on the positive and negative combination relationship between the two, the control parameters are adjusted according to different scenarios. When both the change in angular frequency and the rate of change are negative, the virtual inertia is increased and the damping coefficient is increased in the later stage of adjustment. When the change in angular frequency is negative but the rate of change is positive, the virtual inertia is decreased and the damping coefficient is increased. When both the change in angular frequency and the rate of change are positive, both the virtual inertia and the damping coefficient are increased simultaneously. When the change in angular frequency is positive but the rate of change is negative, the virtual inertia is decreased and the damping coefficient is increased. At the same time, the parameter adjustment range is limited by a preset threshold to avoid frequent switching of virtual inertia and damping coefficient, thereby further optimizing the dynamic response of the system.

[0009] Through the above design, the present invention can effectively suppress frequency overshoot and accelerate oscillation convergence under both sudden load increases and decreases, significantly improving the transient stability of multi-machine parallel grid converter systems, and providing a reliable frequency support solution for the grid-friendly integration of new energy sources.

[0010] The present invention specifically adopts the following technical solution:

[0011] A frequency support method for a multi-machine parallel grid converter system based on improved particle swarm optimization includes:

[0012] With the optimization objective of maximizing the damping ratio of the system's dominant oscillation mode, an improved particle swarm optimization algorithm is used to solve for the optimal initial values ​​of virtual inertia and damping coefficients of each VSG unit. The improved particle swarm optimization algorithm balances global exploration and local convergence by nonlinearly decreasing the inertia weight, strengthens individual cognition in the early stage of iteration and strengthens group cognition in the later stage by dynamically switching the learning factor, and maintains population diversity through a mixed mechanism of Gaussian mutation and Cauchy mutation.

[0013] When the system is disturbed, the system operating status is monitored to obtain the change in system angular frequency Δω and the rate of change of angular frequency dω / dt. Based on the dynamic state of Δω and dω / dt, the virtual inertia and damping coefficient of each VSG unit are adjusted in a coordinated manner to suppress frequency overshoot and accelerate oscillation convergence.

[0014] Furthermore, in the improved particle swarm optimization algorithm: the inertia weight is adjusted from large to small according to a nonlinear law with the number of iterations, so as to balance the global exploration capability of virtual inertia and damping coefficient of multi-machine in the early stage of iteration and the local fine convergence capability in the later stage; the learning factor is dynamically switched with the iteration process, mainly to enhance the particle's cognition of its own historical optimal parameters in the early stage of iteration, and mainly to enhance the particle's learning of the global optimal parameters of the swarm in the later stage; for stagnant particles that have not updated their individual optimal solutions for a preset number of consecutive iterations, Gaussian mutation or Cauchy mutation is triggered according to a preset probability, wherein Gaussian mutation is used for local fine search of parameters, and Cauchy mutation is used to escape local optima to expand the parameter search space, and the position vector of each particle directly corresponds to the virtual inertia and damping coefficient of each VSG unit.

[0015] Furthermore, the step of collaboratively adjusting the virtual inertia and damping coefficient of each VSG unit based on the dynamic state of Δω and dω / dt specifically includes: determining the adjustment direction according to the positive and negative combinations of Δω and dω / dt.

[0016] When both Δω and dω / dt are negative, increase the virtual inertia and increase the damping coefficient in the later stage of adjustment;

[0017] When Δω is negative and dω / dt is positive, the virtual inertia decreases and the damping coefficient increases.

[0018] When both Δω and dω / dt are positive, the virtual inertia and damping coefficient are increased simultaneously.

[0019] When Δω is positive and dω / dt is negative, the virtual inertia decreases and the damping coefficient increases.

[0020] Furthermore, the adjustment of virtual inertia and damping coefficient is combined with preset threshold limits to avoid frequent parameter switching. Virtual inertia is adjusted in intervals based on the absolute value of dω / dt and the sign of the product of Δω and dω / dt, while damping coefficient is adjusted in intervals based on the absolute value of Δω.

[0021] Furthermore, before solving for the optimal initial values ​​of virtual inertia and optimal damping coefficient, a closed-loop small-signal model of the VSG multi-unit parallel system is constructed: the dq reference coordinate axis of one of the VSG units is selected as the common synchronous coordinate system, and the state variables of the other VSG units in their own coordinate systems are unified to the common synchronous coordinate system through coordinate transformation. The load at the grid connection point is introduced to construct the coupling relationship between the multi-unit system and the power grid, forming a closed-loop state-space model that includes the state variables of the multi-unit VSG, the current components of the power grid, and the angle difference between the power grid and the common synchronous coordinate system.

[0022] Furthermore, the range of values ​​for the virtual inertia and damping coefficient is determined based on the operating capacity, rated angular frequency, and maximum angular frequency change rate of the grid-connected converter multi-machine parallel system. The combination of virtual inertia and damping coefficient should ensure that the system damping ratio is within the ideal range of 0.7 to 1.0. The system damping ratio is calculated based on the eigenvalues ​​of the state matrix of the small-signal model of the VSG multi-machine parallel system and is used to quantitatively evaluate the decay rate of the system response oscillation.

[0023] Furthermore, the grid-connected converter multi-machine parallel system includes multiple VSG converter units, grid access components, and grid connection point loads; each VSG converter unit is connected to the grid connection point via the grid access components and then connected to the external power grid through transmission lines.

[0024] And, a frequency support system for a multi-machine parallel grid converter system based on improved particle swarm optimization, comprising:

[0025] Multiple VSG converter units are used to output electrical energy to the grid;

[0026] The monitoring module is used to collect and output the change in angular frequency Δω and the rate of change of angular frequency dω / dt when the system is disturbed.

[0027] The control module is communicatively connected to both the VSG converter unit and the monitoring module, and is configured as follows:

[0028] With the optimization objective of maximizing the damping ratio of the system's dominant oscillation mode, an improved particle swarm optimization algorithm is used to solve for the optimal initial values ​​of virtual inertia and optimal initial values ​​of damping coefficient for each VSG unit. The improved particle swarm optimization algorithm balances global exploration and local convergence by nonlinearly decreasing the inertia weight, strengthens individual cognition in the early stage of iteration and strengthens group cognition in the later stage by dynamically switching the learning factor, and maintains population diversity through a mixed mechanism of Gaussian mutation and Cauchy mutation.

[0029] The system receives Δω and dω / dt from the monitoring module and sends control signals to each VSG unit based on the dynamic state of Δω and dω / dt. This coordinates the adjustment of the virtual inertia and damping coefficient of each VSG unit, suppresses frequency overshoot, and accelerates oscillation convergence.

[0030] Furthermore, the control module is used to solve for the optimal initial value of virtual inertia and the optimal initial value of damping coefficient, and to coordinate the adjustment of virtual inertia and damping coefficient under disturbance; the system also includes a grid connection component and a grid connection point load. The grid connection component is used to realize the power transmission between the VSG converter unit and the grid connection point, and the grid connection point load is used to establish the coupling relationship between the multi-machine VSG and the grid, and to assist the control module in realizing the closed-loop control of the system.

[0031] And, a computer device, including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, it implements the method described above, including: using an improved particle swarm optimization algorithm to solve for the optimal initial values ​​of virtual inertia and optimal initial values ​​of damping coefficients for each VSG unit with the optimization objective of maximizing the damping ratio of the dominant oscillation mode of the system; and when the system is disturbed, coordinating the virtual inertia and damping coefficients according to the monitored Δω and dω / dt.

[0032] A non-transitory computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method described above, including adaptive parameter adjustment of an improved particle swarm optimization algorithm, execution of a hybrid mutation mechanism, and dynamic coordinated adjustment of virtual inertia and damping coefficient under disturbance.

[0033] Compared with the prior art, the present invention and its preferred embodiments have at least the following beneficial effects:

[0034] This invention significantly improves the transient stability of multi-machine parallel grid-connected converter systems, effectively solving the problems of severe frequency overshoot and slow oscillation convergence that easily occur in existing systems under load disturbances. By combining transient power frequency characteristics with adaptive and coordinated parameter adjustment, this invention can adjust the virtual inertia and damping coefficient according to the dynamic changes in the system's angular frequency in different scenarios. Under typical operating conditions such as sudden load increases and decreases, it can suppress the frequency deviation amplitude, accelerate the oscillation decay process, and compensate for the insufficient disturbance rejection and overload tolerance of traditional virtual synchronous generator (VSG) systems.

[0035] This invention optimizes the fit between steady-state and dynamic parameters of the system, ensuring the accuracy and reliability of control parameters. The improved particle swarm optimization algorithm designed in this invention, through nonlinear decreasing inertia weights, dynamic switching of learning factors, and a Gaussian-Cauchy hybrid mutation mechanism, avoids the problems of premature convergence and imbalance between global exploration and local convergence common in conventional optimization algorithms. Furthermore, it aims to maximize the damping ratio of the system's dominant oscillation mode, solving for the optimal initial values ​​of virtual inertia and damping coefficients suitable for multi-machine parallel scenarios, thus providing a stable parameter basis for the system's steady-state operation.

[0036] This invention enhances the coupled and coordinated control capabilities of multi-machine parallel systems, addressing the problem that existing modeling methods struggle to accurately reflect the interactive characteristics of multiple machines. The VSG multi-machine parallel closed-loop small-signal model constructed in this invention uses a common synchronous coordinate system to represent the state variables of each unit, and introduces the load at the grid connection point to establish the coupling relationship between the multiple machines and the power grid. This model can accurately analyze the impact of key parameters such as virtual inertia and damping coefficient on system stability, providing reliable theoretical support for subsequent parameter optimization and dynamic adjustment, and avoiding control deviations caused by neglecting multi-machine coupling in the modeling.

[0037] This invention enhances the flexibility and adaptability of system operation, meeting the evolving needs of the power system under high penetration of new energy sources. The control strategy of this invention does not rely on fixed parameters or a single adjustment logic. It can adapt to parallel scenarios with different numbers of VSG units through improved algorithms, and can cope with diverse load disturbances through adaptive adjustment. This enables multi-unit parallel grid converter systems to participate stably in grid frequency regulation like traditional synchronous generators, providing a feasible path for the friendly integration of new energy into the grid and improving the overall inertia and damping support capacity of the grid. Attached Figure Description

[0038] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0039] Figure 1 This is a topology diagram of a multi-machine parallel grid-connected converter system according to an embodiment of the present invention;

[0040] Figure 2 This is a coordinate transformation diagram of an embodiment of the present invention;

[0041] Figure 3 This is a characteristic root locus diagram of a dual-machine parallel system according to an embodiment of the present invention;

[0042] Figure 4 This is a graph showing the rotor angular frequency oscillation and power angle characteristics of a synchronous generator (SG) according to an embodiment of the present invention.

[0043] Figure 5 This is a diagram illustrating the overall control strategy of an embodiment of the present invention.

[0044] Figure 6 This is a graph showing the adaptive changes of relevant parameters during a sudden load increase in an embodiment of the present invention.

[0045] Figure 7 This is a comparison chart of system frequency changes during a sudden load increase according to an embodiment of the present invention;

[0046] Figure 8 This is a graph showing the adaptive changes of relevant parameters during a sudden load reduction in an embodiment of the present invention.

[0047] Figure 9 This is a comparison chart of system frequency changes during a sudden load reduction in an embodiment of the present invention. Detailed Implementation

[0048] In the following, specific embodiments of this application will be described in detail with reference to the accompanying drawings. Based on these detailed descriptions, those skilled in the art will be able to clearly understand and implement this application. Without departing from the principles of this application, features from various embodiments can be combined to obtain new implementations, or certain features from some embodiments can be substituted to obtain other preferred implementations.

[0049] To make the features and advantages of the present invention more apparent and understandable, specific embodiments are described below in conjunction with the accompanying drawings:

[0050] With the large-scale integration of new energy sources, the inertia and damping of the power system have decreased significantly. Grid-connected converters have been widely used due to their active support capabilities. However, complex interactions will occur between multiple grid-connected converters in parallel systems. Typical grid-connected control strategies with fixed control parameters are prone to problems such as severe power overshoot and long transition time when the system is subjected to large disturbances.

[0051] To address the aforementioned issues, this invention proposes a frequency support strategy for a multi-machine parallel VSG converter system based on an improved particle swarm optimization (PSO) algorithm. First, a small-signal model of the VSG multi-machine parallel system is established to analyze the influence of virtual inertia and damping coefficients on system stability. Second, an improved PSO algorithm integrating adaptive parameter adjustment and a hybrid mutation mechanism is designed to optimize the steady-state control parameters with the goal of maximizing the damping ratio of the system's dominant oscillation mode. Furthermore, considering the system's transient power-frequency characteristics, an adaptive coordinated adjustment strategy for inertia and damping parameters is proposed to dynamically adjust the control parameters during disturbances to improve the system's dynamic response performance. Finally, MATLAB / Simulink simulation experiments verify that the proposed strategy effectively suppresses frequency overshoot, accelerates oscillation convergence, and improves system transient stability under both sudden load increases and decreases.

[0052] This embodiment focuses on a VSG multi-machine parallel grid-connected system. By selecting appropriate state variables, a general small-signal model of the VSG multi-machine parallel grid-connected system is derived and established from a state-space perspective. The changes in the system's characteristic roots and their impact on small-signal stability are analyzed when the virtual moment of inertia and damping coefficient parameters change. Furthermore, an adaptive cooperative control strategy for the damping inertia of the VSG multi-machine parallel system based on an improved particle swarm optimization algorithm is proposed to improve the stability of the VSG multi-machine parallel system.

[0053] Specifically, it includes:

[0054] 1. Small-signal modeling of VSG multi-machine parallel operation

[0055] 1.1 VSG Multi-Machine Parallel System Topology

[0056] In practice, renewable energy power plants typically employ a multi-unit parallel structure. A VSG multi-unit parallel system consists of multiple converter units. Each converter connects to the grid point (PCC) via a filter, connecting lines, and a transformer, and then is connected to the power grid through transmission lines. This invention, considering transformer leakage inductance and local purely resistive loads, uniformly converts system parameters to the low-voltage side of the transformer, resulting in the following... Figure 1 The equivalent topology of the VSG multi-machine parallel grid-connected system is shown. Where, uidqi The dq-axis component of the internal potential of the i-th GFM in its own coordinate system; u 0dq i represents the dq-axis component of the terminal voltage of the i-th GFM in its own coordinate system; R fi L fi and C fi These are the filter resistor, filter inductor, and filter capacitor for the i-th GFM; i idqi Let i represent the dq-axis component of the output current of the i-th GFM in its own coordinate system; odqi The dq-axis component of the grid-connected current of the i-th GFM in its own coordinate system is represented; u pccDQ Represents the DQ-axis components of the grid connection point voltage in the common coordinate system; R load The resistor is the resistor of a parallel purely resistive load; U gdq It is the dq-axis component of the grid voltage in the grid's own coordinate system; R g L g These are the equivalent resistance and inductance of the transmission line between the grid connection point and the infinite power grid, respectively.

[0057] For ease of analysis, this embodiment ignores the influence of millisecond-scale time scale elements such as voltage and current double-loop circuits on the external characteristics of the system when performing small-signal modeling, which reduces the system order without compromising the accuracy of the analysis.

[0058] 1.2VSG Single-Machine Small-Signal Model

[0059] 1.2.1 Small-signal modeling of the power loop

[0060] The traditional VSG power outer loop consists of two parts: active-frequency control and reactive-voltage control. The power outer loop control expression is:

[0061] (1)

[0062] In the formula, P ref and P e These represent the active power reference value and the actual output value, respectively; ω is the actual angular frequency; ω n U is the rated angular frequency; E is the magnetomotive force; u n It is the voltage rating; k q It is the reactive power droop coefficient; Q ref and Q e These represent the reactive power reference value and the actual output value, respectively; J is the moment of inertia of the virtual synchronous machine; D p Where D is the damping coefficient, and D is the damping factor. p This enables the VSG to dampen power oscillations, and the inertia J gives the VSG inertia during power and frequency dynamic processes.

[0063] The active power P output by the VSG eand reactive power Q e From the output voltage u odq and output current i odq The instantaneous power calculation formula is as follows (the subscripts od and oq in formula (2) correspond to the d-axis and q-axis respectively):

[0064] (2)

[0065] Combining equations (1) and (2) and linearizing them, we obtain the small-signal model of the VSG power loop: (3)

[0066] Where θ is the rotor phase angle of the VSG. The value represents the rate of change of the VSG rotor phase deviation. The superscript "·" indicates the derivative sign, and the symbol Δ represents the deviation. and These are the small-signal deviations of the d-axis and q-axis components of the VSG internal potential, respectively. This represents the angular frequency deviation.

[0067] The above formula can be rearranged into its standard form as follows:

[0068] (4)

[0069] in, , , , , , , .

[0070] 1.2.2 Small-signal modeling of filters and transformers

[0071] In this embodiment, the transformer impedance and the line impedance from the filter capacitor to the grid connection point are uniformly equivalent to the transformer resistance R. T L T Meanwhile, ignoring inverter losses, the inverter output voltage is the reference voltage E generated by the VSG control loop. dq At this point, the equivalent mathematical model for the filter and transformer is:

[0072] (5)

[0073] Among them, R f L f and C f These represent the equivalent resistance, inductance, and capacitance of the filter, respectively; R T and L T These are the equivalent resistance and inductance of the transformer, respectively; i dq i is the current output from the VSG to the filter.qd u oqd i dq u odq The axis exchange component, u pdq This is the output voltage of the transformer.

[0074] Linearizing equation (4), we obtain the small-signal model of the filter and transformer as follows:

[0075] (6)

[0076] in, , , , ,Δu pdq =[Δu pd ,Δu pq ] T .

[0077] Combining equations (4) and (6), we obtain the single-machine small-signal model of VSG in its own coordinate system:

[0078] (7)

[0079] in, , , .

[0080] 1.3VSG Multi-Machine Parallel Small-Signal Model

[0081] 1.3.1 Common Coordinate System and Coordinate Transformation

[0082] When performing small-signal modeling for a multi-unit parallel grid-connected converter system, it's necessary to consider that the operating parameters of each converter reside under its own dq reference coordinate axis. Different dq reference coordinate axes have angular differences. Therefore, in the small-signal modeling of the multi-stage parallel system, a common dq reference coordinate axis needs to be chosen, and the data under this axis should be used to establish the connections between the units when building the network equations. In this embodiment, the dq reference coordinate axis of the GFM1 grid-connected converter is taken as the common dq reference axis. For example... Figure 2 As shown, assume the angle difference between the i-th grid-type converter's own dq reference coordinate axis and the common DQ reference coordinate axis is... .

[0083] Figure 2 This reflects the relationship between the state variables under their own dq reference coordinate axis and the state variables under the common dq reference coordinate axis. The mathematical relationship is as follows:

[0084] (8)

[0085] Linearizing the above equation yields:

[0086] (9)

[0087] in, Let be the steady-state angle difference between the i-th VSG's own coordinate system and the common coordinate system. , Let be the steady-state values ​​of the i-th VSG state variables in its own d and q coordinate systems.

[0088] The small-signal model in its own coordinate system is transformed to a common coordinate system. The transformed part includes the phase of each unit in its own coordinate system in the state variables. Transform into Linearization yields:

[0089] (10)

[0090] Input variable u Pdq It needs to be transformed into u through inverse transformation. PDQ After linearization, the small-signal model is obtained as follows:

[0091] (11)

[0092] For the i-th unit, the input variables of the state-space expression need to be increased. The transformed VSG single-machine small-signal model is as follows:

[0093] (12)

[0094] in, , .

[0095] When multiple VSGs are running in parallel, the system small-signal model can be represented as:

[0096] (13)

[0097] in, , , .

[0098] 1.3.2 Small-signal modeling of grid-side transmission lines and resistive loads

[0099] If the power grid is an infinite power grid, then the grid-side voltage amplitude E can be considered as... g Since the angular frequency is a constant value, when modeling the entire system using small signals, the grid-side voltage also needs to be transformed to a common coordinate system and defined. The subscript g represents the power grid, and the power grid voltage in the common coordinate system (the D-axis and Q-axis components in the common DQ coordinate system) can be obtained as follows:

[0100] (14)

[0101] From the system topology diagram, the mathematical model of the transmission line between the grid connection point and the power grid is as follows:

[0102] (15)

[0103] Among them, L g R is the equivalent inductance of the transmission line from the grid connection point PCC to the infinite power grid. g i is the equivalent resistance of the transmission line; gDQ Let i be the component of the transmission line current in the common DQ coordinate system. gQD For i gDQ The axis exchange component, u gDQ It is the vector of the grid voltage in the common DQ coordinate system.

[0104] Linearizing the above equation, the small-signal model of the transmission line can be expressed as:

[0105] (16)

[0106] in, .

[0107] The presence of a purely resistive load is mainly used to represent the grid connection point voltage and establish the coupling relationship between the VSG parallel units and the grid side, fully highlighting the coupling characteristics of multiple VSG parallel units compared to single-unit parallel units, enabling the small-signal modeling of the entire system to complete the closed loop, and the state-space expression will have no input or output variables:

[0108] (17)

[0109] Among them, u pD and u pQ These represent the D-axis and Q-axis components of the PCC voltage at the grid connection point in the common DQ coordinate system, respectively. load The purely resistive load resistor at the grid connection point PCC; i oD1 ~i oDn i oQ1 ~i oQn These represent the D-axis and Q-axis components of the grid-connected current of the 1st to nth VSGs in the common DQ coordinate system, i gD i gQ These are the D-axis and Q-axis components of the transmission line current in the common DQ coordinate system, respectively.

[0110] Linearizing the above equation yields:

[0111] (18)

[0112] in, , .

[0113] Combining the above equations, we can obtain the small-signal model of the entire system:

[0114] (19)

[0115] in, Let be the small-signal deviation vector of the overall system state variables. This is the overall small-signal state matrix of the system.

[0116] 1.4 Stability Analysis of VSG Multi-stage Parallel System

[0117] To investigate the frequency stability characteristics between VSG units and between VSG and the power grid, this embodiment takes a dual-unit VSG parallel grid-connected system as an example and uses eigenvalue analysis to analyze the stability of multi-stage parallel systems. The inverter main circuit and virtual synchronous machine control loop parameters of the dual-unit parallel system, converted to the low-voltage side, are shown in Table 1.

[0118] Table 1 Parameters of grid-connected system with two VSGs in parallel (subscripts 1 and 2 correspond to the two VSGs in the two-unit system)

[0119]

[0120] For a VSG dual-machine parallel system, its state-space expression is established as follows:

[0121] (20)

[0122] By obtaining the eigenvalues ​​of the 19th-order matrix A2 in the above equation, all eigenvalues ​​are distributed to the left of the imaginary axis, indicating that the system is stable under small disturbances. This embodiment only considers eigenvalues ​​that are close to the imaginary axis and have a significant impact on stability. The focus is on analyzing the influence of the virtual inertia J and the damping coefficient Dp on the system stability.

[0123] Existing research indicates that when the parameters of the active-frequency control loop change simultaneously, the frequency stability characteristics of a multi-machine parallel system are equivalent to those of a single grid-connected converter. This embodiment studies the impact of changes in the control parameters of a single converter on the system's frequency stability. Figure 3 As shown in (a), when the damping coefficient remains constant, as the virtual inertia J1 of GFM1 changes from 2 to 20, the system characteristic roots gradually shift towards the virtual axis, the system damping decreases, the active power overshoot increases during dynamic processes, the system's dynamic response gradually slows down, the system exhibits a low-frequency oscillation trend, and the stability deteriorates. Figure 3As shown in (b), when the virtual inertia remains constant, when the damping coefficient of GFM1 changes from 10 to 100, the characteristic roots of the system gradually move away from the virtual axis, the system response becomes faster, and the system damping increases, making it less prone to oscillation.

[0124] Therefore, a fixed virtual inertia and damping coefficient cannot ensure that the system has both large inertia and good dynamic and steady-state performance. The output characteristics of a network-type VSG multi-stage parallel system can be optimized by rationally optimizing and allocating the virtual inertia and damping coefficients of the multi-stage parallel system, and by sampling a certain amount of adaptive cooperative control technology.

[0125] 2. Stability Improvement and Optimization Strategies for VSG Dual-Machine Parallel Systems

[0126] 2.1 Analysis of System Transient Power Frequency Characteristics

[0127] When the system is disturbed, the oscillation process of the VSG can be considered similar to that of a synchronous generator, and the relationship between the power angle and rotor angular frequency is shown in the curve. Figure 4 As shown, when the load suddenly increases, the system frequency first decreases and then recovers, and the power angle change trajectory is as follows. Finally, the oscillations converged to Frequency modulation in section I plays a major role. ref If the electromagnetic power is less than the output power P, it can be seen from the rotor motion equation that the rotor will decelerate at this time, and the angular frequency of the system will decrease. Less than the rated angular frequency , and According to the VSG active-loop control equations, increasing J can reduce the amplitude of the rate of change of angular frequency. Increase the damping coefficient D at the end of interval I. p It can reduce the change in angular frequency. This improves system stability. Secondary frequency modulation in interval II plays a major role; the system angular frequency is restored. At this point, but At this point, decreasing the inertia J and increasing D can improve the angular frequency recovery speed. In interval III, due to system inertia, the angular frequency continues to rise for a period of time. and To reduce and At this point, J and D should be increased. P The frequency modulation effect of the IV-range system restores the angular frequency, at which point... and The parameter value pattern during the recovery phase is referenced within interval II. Based on the above analysis, the control parameters J and D can be adaptively changed. p To improve the dynamic characteristics of the system. The adaptation rules of the control parameters are summarized as follows:

[0128] Table 2 Selection Principles of J and Dp

[0129]

[0130] Therefore J and D p The value can be obtained according to the following formula:

[0131] (twenty one)

[0132] (twenty two)

[0133] In the formula, J0 is the virtual inertia parameter in steady state, x and y are the adaptive inertia adjustment coefficients, and T... J1 T J2 This is the inertia adaptive start threshold, mainly used to prevent frequent switching of virtual inertia. D0 is the damping coefficient in steady state, z is the damping adaptive coefficient, and T... D1 T D2 It is a damping adaptive threshold. (D) max This is the maximum value of the damping.

[0134] 2.2 Design of an Improved Particle Swarm Optimization Algorithm Based on Dynamic Optimization

[0135] To improve the transient stability of a dual-VSG parallel system, this embodiment designs an improved particle swarm optimization algorithm based on dynamic optimization to solve for the virtual inertia (J) and damping coefficient (D) of the two VSGs in real time. p This algorithm achieves online global optimization search by introducing adaptive parameter adjustment and hybrid mutation mechanisms into the standard PSO framework.

[0136] 2.2.1 Adaptive Parameter Adjustment

[0137] The core formula of the classic PSO algorithm is as follows:

[0138] (twenty three)

[0139] Let be the velocity of particle i in the kth iteration; The current position of particle i; This represents the historical best position of particle i; The optimal position for the entire group; c1 and c2 are inertia weights; c1 and c2 are learning factors. r1 and r2 are random numbers, uniformly distributed between 0 and 1. Inertia weights The learning factors c1 and c2 are key parameters affecting PSO performance. This algorithm dynamically changes them with the iteration process to balance exploration and exploitation capabilities, where the inertia weight... Using a nonlinear decreasing strategy, a larger initial value is beneficial for global exploration, while a smaller value in the later stages is beneficial for local convergence. Its value is calculated using the following formula:

[0140] (twenty four)

[0141] Where k is the current iteration number, K max The maximum number of iterations, and The upper and lower limits are preset for the inertia weight.

[0142] Learning factors c1 and c2 control the degree to which a particle is influenced by its own historical experience and the social experience of the group, respectively. The overall design of this invention emphasizes individual cognition in the early stages of iteration and shifts to social cognition in the later stages to accelerate convergence. The values ​​are as follows:

[0143] (25).

[0144] Where α is the learning factor decay coefficient, used to control the dynamic switching rate of learning factors c1 and c2, c 10 c 20 These are the initial values ​​for learning factors c1 and c2, respectively.

[0145] 2.2.2 Mixed Mutation Mechanism

[0146] In the later stages of standard particle swarm optimization (PSO), population diversity drops sharply, and particles tend to cluster around a local optimum, leading to premature convergence. Introducing a mutation mechanism is an effective strategy to overcome this drawback. However, traditional single mutation strategies have limitations: Gaussian mutation has a small step size, which is beneficial for fine-grained local search but has limited ability to escape local optima; Cauchy mutation, due to its heavy-tailed distribution, can generate a larger step size, thus more effectively helping particles escape local optima, but its excessive leaps may disrupt good search directions. This invention combines the strong local exploitation capability of Gaussian mutation with the strong global exploration capability of Cauchy mutation, adaptively triggering different types of mutations based on the particle's state, thereby achieving the optimal balance between maintaining population diversity and accelerating convergence.

[0147] First, it is necessary to identify particles that are stuck in a search stagnation state. Define a particle i such that its individual historical optimal solution is... It has not been updated in S consecutive iterations, which means it satisfies:

[0148] (26)

[0149] The particle is then identified as a "stagnant particle." Here, S is a preset stagnation algebra threshold, which is set to 3 in this embodiment. For the identified stagnant particles, mutation is performed according to the following rules: with probability P... m Perform Gaussian mutation with probability 1−P m Perform Cauchy mutation, where the mutation probability P mThe value is usually 0.5.

[0150] The Gaussian variant is as follows:

[0151] (27)

[0152] in, This indicates that the mean is 0 and the standard deviation is 0. The Gaussian distribution. Gaussian variation is achieved by applying the current position X. i Applying a small, zero-mean random perturbation to enable a fine search in the neighborhood is beneficial for local exploitation and mining near the current optimal solution.

[0153] The Cauchy variants are as follows:

[0154] (28)

[0155] in, This indicates that the position parameter is 0 and the scale parameter is... The Cauchy distribution has a flatter, longer tail, which allows it to generate larger steps away from the mean with a higher probability. This property enables Cauchy mutation to give stagnant particles a stronger jumping ability, thus effectively escaping the current local optimum and exploring a wider new region in the solution space.

[0156] 2.2.3 Determining the range of control parameters

[0157] Inertia J and damping D in VSG control p The calculation expression is as follows:

[0158] (29)

[0159] In the formula, This represents the maximum change in the system's active power. It is the rated angular frequency; It is the maximum rate of change of angular frequency. It is the maximum change in angular frequency.

[0160] The experimental system in this embodiment has an operating capacity of 10KW, a maximum frequency variation range of ±0.2Hz, and a maximum angular frequency change rate of 6.28rad / s. 2 The maximum range of variation in the system's active power is 100% of the system's operating capacity, from which we can derive:

[0161]

[0162] J max Determined by the system's maximum allowable adjustment time. Take J. max=20. Therefore, the inertia coefficient in this embodiment ranges from 5 to 20, P n This is the system's rated power.

[0163] The active power loop of VSG can be approximated as a second-order system: Its damping ratio To achieve the optimal transient response, the damping ratio... The design value is typically between 0.7 and 1.0. From this, we can deduce... The range of values ​​for .

[0164] Where K is a coefficient related to line impedance and operating point; for a typical system, its value is approximately 10. 3 ~10 4 Magnitude.

[0165] Through numerous such designs and simulations, it was discovered that D p Within the range of 50~200kW·s / rad, it can effectively coordinate with J to ensure that the system damping ratio falls within the ideal range.

[0166] 2.2.4 Fitness Function Design

[0167] The fitness function result serves as a crucial basis for determining whether the particle swarm optimization (PSO) iteration converges and terminates. In power system stability assessment, the system's damping characteristics are a key indicator for evaluating transient performance. To directly optimize the system's dynamic response quality, this embodiment uses the damping ratio of the system's dominant oscillation mode as the core component of the fitness function.

[0168] The damping ratio calculation is based on the small-signal model analysis of the VSG dual-machine parallel system. For a linearized dynamic system, its characteristics are determined by the state matrix. The eigenvalues ​​determine the eigenvalues. In the following formula, the eigenvalues ​​are... real part The damping ratio determines the stability of the system (ω corresponds to the imaginary part). This quantitatively describes the decay rate of the system response oscillations.

[0169] Damping ratio The calculation formula is as follows:

[0170] (30)

[0171] As can be seen from the small-signal model constructed in Section 1.4, the state matrix of the system is... VSG control parameters The function of . For matrix By solving for the eigenvalues, we can obtain the eigenvalue spectrum and find the conjugate complex eigenvalues ​​corresponding to the dominant oscillation modes. In the damping ratio fitness function, the damping characteristics of the dominant system mode are taken as the core objective for optimization. Therefore, the final expression of the fitness function is:

[0172] (31)

[0173] The fitness function aims to be maximized. When the algorithm converges iteratively, the fitness function reaches its maximum value, and the output VSG parameter combination at this point is the solution that maximizes system damping and achieves optimal transient stability.

[0174] Therefore, the overall control strategy flow of the improved particle swarm optimization algorithm is as follows: Figure 5 As shown, the initial particle information is first established. The position of each particle i is a four-dimensional vector, directly representing the parameters of the two VSGs: The constraints are J and D. p The range of parameter values. For each particle, calculate its fitness and update Pbest and Gbest, and update the inertia weights according to the formula. And learning factors c1 and c2. Update the particles according to the velocity-displacement formula, limit the amplitude of particles that go out of bounds, and ensure that they are always within the constraint range. Perform Gaussian or Cauchy mutation on stagnant particles to escape local optima. The algorithm terminates when the maximum number of iterations is reached or the quality of the solution no longer improves significantly. Output the global optimal solution to provide the optimal initial values ​​of the control parameters for the two VSGs, and then according to formulas (21) and (22) based on the change of the system's operating angular frequency. and rate of change of angular frequency The parameters are adaptively adjusted, and the overall control strategy based on the above algorithm is as follows: Figure 5 As shown.

[0175] 4. Simulation Analysis

[0176] To verify the effectiveness and accuracy of the proposed optimization strategy for improving the stability of the VSG dual-machine parallel system, a simulation experiment was conducted on the VSG dual-machine parallel grid-connected model on the MATLAB / Simulink simulation platform. The model system parameters are shown in Table X, and the algorithm parameter configuration is shown in Table 3. The frequency change before and after applying the control strategy of this invention was compared by designing two operating conditions: a sudden load increase and a sudden load decrease. The initial state of the system was a load of 50kW, with each VSG outputting 10kW, and the remaining load being supplied by the power grid.

[0177] Table 3 Algorithm Parameter Configuration

[0178]

[0179] The improved particle swarm optimization algorithm proposed in this invention was calculated using the MATLAB simulation platform, and the state matrix A constructed by the system parameters of this invention under different control parameters was calculated. totalThe eigenvalues ​​are calculated and the fitness is determined. Finally, the optimal combination of control parameters for system stability is obtained and rounded down as follows: .

[0180] Operating condition 1 is that the load suddenly increases to 100kW at t=5s. Figure 6 The changes in relevant parameters after applying the control strategy of the present invention are shown, where (a) is D. p1 Adaptive adjustment curve, (b) is D p2 Adaptive adjustment curves, (c) is the J1 adaptive adjustment curve, (d) is the J2 adaptive adjustment curve. As the load increases, the system frequency decreases, the inertia coefficient responds quickly, and J1 and J2 increase rapidly to suppress the rate of change of angular frequency. When the rate of change of angular frequency exceeds the threshold T... J1 When, the damping coefficient D p1 D p2 By rapidly increasing or decreasing the angular frequency overshoot, the system angular frequency decreases to its minimum value and recovers after approximately 5.1 seconds. At this point, the damping coefficient reaches its maximum value and then gradually recovers. The inertia coefficient decreases rapidly at this time, increasing the angular frequency recovery speed. The frequency changes before and after applying the control strategy of this invention are as follows: Figure 7 As shown in the comparison, after applying the strategy of the present invention, the minimum frequency of the system under the same disturbance is increased from 49.78Hz to 49.84Hz, which reduces the frequency overshoot and improves the oscillation convergence speed, thereby improving the system stability.

[0181] Operating condition two involves a sudden load reduction to 40kW at t=5s. Figure 8 The changes in relevant parameters after applying the control strategy of the present invention are shown, where (a) is D. p1 Adaptive adjustment curve, (b) is D p2 Adaptive adjustment curves, (c) is the J1 adaptive adjustment curve, (d) is the J2 adaptive adjustment curve. When the load decreases, the system frequency increases, and J1 and J2 also increase rapidly to suppress the rate of change of angular frequency. When the rate of change of angular frequency exceeds the threshold T... J1 When, the damping coefficient D p1 D p2 Rapidly increasing or decreasing the angular frequency overshoot, the system angular frequency reaches its maximum value and recovers in approximately 5.1 seconds. At this point, the damping coefficient reaches its maximum and then begins to decrease in line with the change in angular frequency. The inertia coefficient decreases rapidly at this time, increasing the angular frequency recovery speed. Figure 9 As shown, after applying the strategy of this invention, the maximum frequency of the system decreased from 50.052Hz to 50.047Hz, reducing the frequency overshoot and improving system stability.

[0182] This invention first establishes a small-signal model of a multi-unit parallel system of grid-connected converters based on VSG control. The system eigenvalue trajectories of the dual-unit parallel VSG grid-connected system are obtained and plotted as the virtual moment of inertia and damping coefficient change, and their impact on system stability is analyzed. An improved particle swarm optimization algorithm is used to allocate control parameters to the two VSGs with the goal of improving system stability. An adaptive control strategy for adjusting inertia and damping is designed by combining the control parameters with the system's transient power frequency characteristics. Finally, simulations verify the effectiveness of the strategy.

[0183] Based on the same inventive concept, this invention also provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the programs include program instructions, and the processor executes the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, used to implement one or more instructions, specifically for loading and executing one or more instructions stored in a computer storage medium to implement the above-described method.

[0184] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, performs the above-described method. This storage medium can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0185] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0186] The foregoing has shown and described the basic principles, main features, and advantages of this disclosure. Those skilled in the art should understand that this disclosure is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this disclosure. Various changes and modifications can be made to this disclosure without departing from its spirit and scope, and all such changes and modifications fall within the scope of this disclosure as claimed.

[0187] This invention is not limited to the preferred embodiment described above. Anyone inspired by this invention can derive other various forms of frequency support methods for multi-machine parallel systems of grid converters based on improved particle swarm optimization. All equivalent variations and modifications made within the scope of the claims of this invention should be included in the scope of this invention.

Claims

1. A frequency support method for a multi-machine parallel system of grid-connected converters based on improved particle swarm optimization, characterized in that, include: With the optimization objective of maximizing the damping ratio of the system's dominant oscillation mode, an improved particle swarm optimization algorithm is used to solve for the optimal initial values ​​of virtual inertia and damping coefficients of each VSG unit. The improved particle swarm optimization algorithm balances global exploration and local convergence by nonlinearly decreasing the inertia weight, strengthens individual cognition in the early stage of iteration and strengthens group cognition in the later stage by dynamically switching the learning factor, and maintains population diversity through a mixed mechanism of Gaussian mutation and Cauchy mutation. When the system is disturbed, the system operating status is monitored to obtain the change in system angular frequency Δω and the rate of change of angular frequency dω / dt. Based on the dynamic state of Δω and dω / dt, the virtual inertia and damping coefficient of each VSG unit are adjusted in coordination to suppress frequency overshoot and accelerate oscillation convergence. In the improved particle swarm optimization algorithm: the inertia weight is adjusted from large to small according to a nonlinear law with the number of iterations, so as to balance the global exploration capability of virtual inertia and damping coefficient of multi-machine in the early stage of iteration and the local fine convergence capability in the later stage; the learning factor is dynamically switched with the iteration process, mainly to enhance the particle's cognition of its own historical optimal parameters in the early stage of iteration, and mainly to enhance the particle's learning of the global optimal parameters of the swarm in the later stage. For stagnant particles that have not updated their individual optimal solutions for a set number of consecutive preset number of iterations, Gaussian mutation or Cauchy mutation is triggered with a preset probability. Gaussian mutation is used for local fine-tuning of parameters, while Cauchy mutation is used to escape local optima to expand the parameter search space. Furthermore, the position vector of each particle directly corresponds to the virtual inertia and damping coefficient of each VSG unit. Before solving for the optimal initial values ​​of virtual inertia and optimal damping coefficient, a closed-loop small-signal model of the VSG multi-unit parallel system is constructed: the dq reference coordinate axis of one of the VSG units is selected as the common synchronous coordinate system, and the state variables of the other VSG units in their own coordinate systems are unified to the common synchronous coordinate system through coordinate transformation. The load at the grid connection point is introduced to construct the coupling relationship between the multi-unit system and the power grid, forming a closed-loop state-space model that includes the state variables of the multi-unit VSG, the current components of the power grid, and the angle difference between the power grid and the common synchronous coordinate system.

2. The frequency support method for a multi-machine parallel system of grid converters based on improved particle swarm optimization as described in claim 1, characterized in that: The specific steps of coordinating the adjustment of the virtual inertia and damping coefficient of each VSG unit based on the dynamic state of Δω and dω / dt include: determining the adjustment direction according to the positive and negative combinations of Δω and dω / dt. When both Δω and dω / dt are negative, increase the virtual inertia and increase the damping coefficient in the later stage of adjustment; When Δω is negative and dω / dt is positive, the virtual inertia decreases and the damping coefficient increases. When both Δω and dω / dt are positive, the virtual inertia and damping coefficient are increased simultaneously. When Δω is positive and dω / dt is negative, the virtual inertia decreases and the damping coefficient increases. Furthermore, the adjustment of virtual inertia and damping coefficient is combined with preset threshold limits to avoid frequent parameter switching. Virtual inertia is adjusted in intervals based on the absolute value of dω / dt and the sign of the product of Δω and dω / dt, while damping coefficient is adjusted in intervals based on the absolute value of Δω.

3. The frequency support method for a multi-machine parallel system of grid converters based on improved particle swarm optimization as described in claim 1, characterized in that: The range of values ​​for the virtual inertia and damping coefficient is determined based on the operating capacity, rated angular frequency, and maximum angular frequency change rate of the grid-connected converter multi-machine parallel system. The combination of virtual inertia and damping coefficient should ensure that the system damping ratio is within the ideal range of 0.7 to 1.

0. The system damping ratio is calculated based on the eigenvalues ​​of the state matrix of the small-signal model of the VSG multi-machine parallel system and is used to quantitatively evaluate the decay rate of the system response oscillation.

4. The frequency support method for a multi-machine parallel system of grid converters based on improved particle swarm optimization as described in claim 1, characterized in that: The grid-connected converter multi-machine parallel system includes multiple VSG converter units, grid access components, and grid connection point loads; each VSG converter unit is connected to the grid connection point via the grid access components and then connected to the external power grid through transmission lines.

5. A frequency support system for a multi-machine parallel grid converter system based on improved particle swarm optimization, used to implement the method as described in claim 1, characterized in that, include: Multiple VSG converter units are used to output electrical energy to the grid; The monitoring module is used to collect and output the change in angular frequency Δω and the rate of change of angular frequency dω / dt when the system is disturbed. The control module is communicatively connected to both the VSG converter unit and the monitoring module, and is configured as follows: With the optimization objective of maximizing the damping ratio of the system's dominant oscillation mode, an improved particle swarm optimization algorithm is used to solve for the optimal initial values ​​of virtual inertia and optimal initial values ​​of damping coefficient for each VSG unit. The improved particle swarm optimization algorithm balances global exploration and local convergence by nonlinearly decreasing the inertia weight, strengthens individual cognition in the early stage of iteration and strengthens group cognition in the later stage by dynamically switching the learning factor, and maintains population diversity through a mixed mechanism of Gaussian mutation and Cauchy mutation. The system receives Δω and dω / dt from the monitoring module and sends control signals to each VSG unit based on the dynamic state of Δω and dω / dt. This coordinates the adjustment of the virtual inertia and damping coefficient of each VSG unit, suppresses frequency overshoot, and accelerates oscillation convergence.

6. The frequency support system for a multi-machine parallel system of grid-connected converters based on improved particle swarm optimization as described in claim 5, characterized in that: The control module is used to solve for the optimal initial value of virtual inertia and the optimal initial value of damping coefficient, and to coordinate the adjustment of virtual inertia and damping coefficient under disturbance. The system also includes a grid connection component and a grid connection point load. The grid connection component is used to realize the power transmission between the VSG converter unit and the grid connection point. The grid connection point load is used to establish the coupling relationship between the multi-machine VSG and the grid, and to assist the control module in realizing the closed-loop control of the system.

7. A computer device, characterized in that, The system includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When executed by the processor, the computer program implements the method described in any one of claims 1-4, comprising: using an improved particle swarm optimization algorithm to solve for the optimal initial values ​​of virtual inertia and optimal damping coefficients of each VSG unit with the optimization objective of maximizing the damping ratio of the dominant oscillation mode of the system; and coordinating the adjustment of virtual inertia and damping coefficients based on the monitored Δω and dω / dt when the system experiences a disturbance.

8. A non-transitory computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method described in any one of claims 1-4, including adaptive parameter adjustment of the improved particle swarm optimization algorithm, execution of the hybrid mutation mechanism, and dynamic coordinated adjustment of virtual inertia and damping coefficient under disturbance.

Images