A method, system and device for coordinated control of grid-forming and grid-following photovoltaic converters

Abstract

The present application relates to the field of new energy grid-connected control technology, and more particularly to a grid-constructing and grid-following photovoltaic converter collaborative control method, system and device. The present application adopts an integrated architecture of offline pre-optimization combined with online fast and slow layer grading and linkage of two types of converters: offline construction of a parameter grid indexed by voltage drop depth and interaction intensity, screening by generalized operating short-circuit ratio safety domain and model predictive optimization with safety margin constraints to generate a multi-dimensional lookup table containing current limiting threshold and virtual impedance value; online fast layer table lookup interpolation with control period as the beat, and safety synthesis of segmented current limiting threshold to output dynamic current limiting parameters; slow layer long period identification of system characteristics and grid impedance, iterative update of current limiting curve parameters. At the same time, grid-constructing virtual flux adaptive synchronization and grid-following phase-locked loop bandwidth linkage adjustment are realized, effectively improving system operation stability and reliability under weak grid, fault ride-through and multi-machine parallel scenarios.

Description

Technical Field

[0001] This invention relates to the field of new energy grid-connected control technology, and in particular to a method, system, and equipment for coordinated control of grid-connected and grid-following photovoltaic converters. Background Technology

[0002] With the advancement of dual-carbon goals, the penetration rate of new energy power generation such as photovoltaic power generation continues to increase, and the proportion of power electronic equipment in the power system is constantly increasing. This leads to a decrease in the system's equivalent inertia and short-circuit capacity, making the characteristics of a weak power grid increasingly prominent. Grid-mounted photovoltaic converters have external voltage source characteristics and can actively provide voltage and frequency support to the grid, making them a core technical path to improve the grid-connected stability of new energy sources. Grid-following photovoltaic converters adopt the traditional current source control mode, relying on phase-locked loops to track the grid phase, and are currently the mainstream equipment form for existing photovoltaic power plants. The mixed parallel operation of these two types of converters has become a typical grid configuration mode for high-proportion new energy power plants.

[0003] To address the stability issues of parallel operation of grid-connected and grid-connected converters, existing technologies have conducted research in multiple directions: First, based on sequence impedance modeling theory, aggregated impedance models of the two types of converters are constructed. The risk of parallel oscillation is analyzed through impedance ratio criteria, and oscillation is suppressed through impedance shaping strategies. However, these methods are mostly based on steady-state parameter tuning and are difficult to dynamically adapt to complex operating conditions such as voltage dips and grid impedance fluctuations. Second, grid impedance parameters are identified through active disturbance injection or by utilizing natural grid disturbances. The system stability margin is evaluated online, and control parameters are adjusted. However, the impedance identification and margin evaluation calculations are highly complex and difficult to match the real-time requirements of the converter's microsecond-level control cycle. Third, the synchronous mode or current-limiting mode of the grid-connected converter is switched according to the system short-circuit ratio, voltage dip depth, or frequency deviation to achieve mode switching between steady-state operation and fault ride-through. However, the mode switching process is prone to phase jumps and power surges, and the coordinated adjustment of control parameters of the grid-connected converter is not considered simultaneously. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a collaborative control method, system, and electronic equipment for grid-connected and grid-linked photovoltaic converters. This is achieved through an integrated architecture that integrates offline pre-optimization to generate multi-dimensional lookup tables, online fast / slow layer hierarchical execution, and virtual flux linkage synchronization in the grid-connected type with adaptive bandwidth linkage in the grid-linked type. This solves the problems of balancing real-time performance and optimization accuracy, and impedance coupling conflicts easily caused by independent tuning of multiple control links in existing technologies. It improves the stability and reliability of parallel operation of the two types of converters in scenarios such as weak grids, fault ride-through, and multi-converter parallel operation. The specific technical solution is as follows:

[0005] A coordinated control method for grid-connected and grid-linked photovoltaic converters, applied to a photovoltaic power station with at least one grid-connected photovoltaic converter and at least one grid-linked photovoltaic converter connected to the same grid connection point, includes the following steps:

[0006] S1: Construct a parametric grid indexed by voltage sag depth, interaction strength between grid-connected and grid-connected photovoltaic converters, and filter grid points based on the generalized operating short-circuit ratio safety domain. Perform model prediction optimization with safety margin constraints on the effective grid points within the safety domain to solve for the current limiting threshold, virtual impedance value and active power loop correction coefficient corresponding to each grid point, forming a multidimensional lookup table.

[0007] S2: Establish a fast-layer real-time response mechanism: Using the converter control cycle as the execution cycle, collect the grid connection point voltage in real time and calculate the voltage drop depth. Combine the current effective interaction strength to interpolate the multi-dimensional lookup table mentioned in S1 to obtain the lookup table control parameters. At the same time, calculate the fast current limiting threshold based on the piecewise linear current limiting curve, and safely synthesize the lookup table current limiting threshold and the fast current limiting threshold to output the final dynamic current limiting parameters.

[0008] S3: Establish a slow layer periodic update mechanism: Using a period longer than the converter control cycle as the execution cycle, identify the system interaction strength, coupling level and grid impedance parameters. Based on the current safety margin, interaction strength and the state of the multidimensional lookup table in S1, iteratively update the parameters of the piecewise linear current limiting curve. After safety verification, send it to the fast layer. At the same time, output the identified impedance and damping parameters to the grid-connected photovoltaic converter.

[0009] S4: Based on the dynamic current limiting parameters output by the fast layer of S2, current reference limiting, virtual impedance compensation and active power loop correction are performed respectively; then, the control reference phase is obtained by virtual flux orientation, and the inertial damping factor is adaptively adjusted according to the virtual flux disturbance rate to suppress phase disturbances caused by voltage drop and parameter switching.

[0010] S5: Based on the grid equivalent impedance ratio and interactive damping disturbance amount identified by the slow layer in S3, the cutoff frequency of the phase-locked loop filter is adaptively adjusted. Under disturbance conditions, the bandwidth of the phase-locked loop is reduced to weaken coupled oscillations, and under steady-state conditions, the bandwidth is restored to ensure phase tracking accuracy.

[0011] Furthermore, step S1 specifically includes the following sub-steps:

[0012] S11: Read the grid connection point rated parameters, converter capacity and control parameters, and construct a two-dimensional parameter grid with voltage drop depth as the first index and interaction strength as the second index; the voltage drop depth is calculated based on the ratio of the difference between the grid connection point rated voltage and the real-time grid connection point voltage to the rated voltage.

[0013] S12: Based on the parameter grid constructed by S11, the generalized operating short-circuit ratio is used to determine the boundary of the system safety domain. The generalized operating short-circuit ratio safety margin of each grid point is calculated. Grid points with non-negative safety margins are determined as valid grid points, and grid points with negative safety margins are marked as forbidden boundaries. No usable control parameters are generated.

[0014] S13: Perform model prediction optimization with safety margin constraints on each valid grid point obtained from S12. The optimization objective is to minimize the comprehensive objective consisting of current tracking error, control increment smoothness and safety margin penalty. Solve for the current limiting threshold, virtual impedance value and active power loop correction coefficient corresponding to each grid point to form a multidimensional lookup table. Invalid grid points are filled with parameters of the adjacent more conservative interaction strength level or marked as forbidden.

[0015] Furthermore, step S2 specifically includes the following sub-steps:

[0016] S21: Using the converter control cycle as the execution beat, the grid connection point voltage is collected in real time and the voltage drop depth is calculated. Combined with the current effective interaction strength, the multidimensional lookup table generated in S1 is interpolated to obtain the lookup table current limiting threshold, virtual impedance value and active power loop correction coefficient under the corresponding operating conditions.

[0017] S22: Calculate the fast current limiting threshold based on the piecewise linear current limiting curve, combine the lookup current limiting threshold obtained in S21 with the fast current limiting threshold by taking the smaller value for safety, and output the final dynamic current limiting parameter; when the deviation between the two exceeds the set threshold, maintain the previous effective virtual impedance value and send a parameter verification flag to the slow layer.

[0018] Furthermore, step S3 specifically includes the following sub-steps:

[0019] S31: By injecting small-amplitude disturbances or utilizing the voltage and current responses under natural disturbances, the interaction strength, coupling level, grid equivalent impedance ratio, and interaction damping disturbance of the system are identified; S32: Using a period longer than the converter control cycle as the execution cycle, combined with the generalized operating short-circuit ratio safety margin of S1, the interaction strength identified by S31, and the multidimensional lookup table state of S1, the inflection point, slope, and intercept parameters of the updated piecewise linear current limiting curve are iteratively generated.

[0020] S33: Perform safety margin, version number and parameter change rate verification on the updated curve parameters generated by S32. If the verification passes, it is sent to the fast layer. If the verification fails, the fast layer maintains the valid curve parameters of the previous version and calls the conservative level control parameters in the multidimensional lookup table of S1.

[0021] Furthermore, step S4 specifically includes the following sub-steps:

[0022] S41: Receives the dynamic current limiting parameters output by S2, limits the current reference value, performs virtual impedance compensation on the output current, and corrects the active power loop deviation; at the same time, it collects the grid connection point voltage and output current and converts them to a stationary two-phase coordinate system, and outputs the voltage α / β axis components and current α / β axis components to the extended state observer.

[0023] S42: Receive the voltage and current components of the stationary two-phase coordinate system output by S41 through the extended state observer, estimate the α and β axis components of the virtual flux linkage vector, calculate the virtual flux linkage angle, and use it as the unified control reference phase for each control loop of the grid-type photovoltaic converter.

[0024] S43: Calculate the virtual flux disturbance rate based on the virtual flux vector of the adjacent control cycle obtained in S42, and adaptively adjust the inertial damping factor based on the disturbance rate; when the disturbance rate increases, increase the damping factor to suppress phase disturbance, and when the disturbance rate decreases, gradually restore the damping factor to the basic value.

[0025] Furthermore, in step S42, the virtual flux linkage angle is determined by the arctangent of the virtual flux linkage β-axis component and α-axis component. This angle is synchronously output to the voltage outer loop, the active power loop, and the virtual impedance loop as a common phase reference for each loop.

[0026] Furthermore, step S5 specifically includes the following sub-steps:

[0027] S51: Based on the grid equivalent impedance ratio and interactive damping disturbance obtained from S3, calculate the target cutoff frequency of the phase-locked loop filter, and impose upper and lower limits on the target value through preset minimum and maximum cutoff frequencies.

[0028] S52: Dynamically adjust the PLL bandwidth based on the target cutoff frequency calculated in S51: When the grid equivalent impedance ratio increases and the amount of interactive damping disturbance increases, reduce the PLL cutoff frequency to weaken the coupled oscillation; when the system returns to steady state, gradually increase the cutoff frequency to ensure phase tracking accuracy; when the identification parameters are invalid, maintain the previous effective cutoff frequency or switch to a conservative low bandwidth value.

[0029] Furthermore, in step S51, the cutoff frequency of the phase-locked loop filter is calculated by superimposing the grid equivalent impedance ratio adjustment term and the interactive damping disturbance adjustment term from S3, based on the steady-state reference cutoff frequency. The final output value is limited to the range between the minimum allowable cutoff frequency and the maximum allowable cutoff frequency.

[0030] A collaborative control system for grid-connected and grid-following photovoltaic converters includes an offline lookup table generation module, a dynamic interactive identification module, a slow-layer parameter update module, a fast-layer current limiting control module, a virtual flux linkage phase-locked loop synchronization module, and a grid-following phase-locked loop bandwidth adaptive module.

[0031] The offline lookup table generation module is used to construct a parameter grid, perform security domain filtering and constrained model prediction optimization, generate a multidimensional lookup table indexed by voltage drop depth and interaction strength, and output the multidimensional lookup table to the fast layer current limiting control module and the slow layer parameter update module respectively.

[0032] The dynamic interaction identification module is used to identify the system interaction strength, coupling level, grid equivalent impedance ratio and interaction damping disturbance amount, and outputs the interaction strength and coupling level data to the fast layer current limiting control module and the slow layer parameter update module respectively, and outputs the grid equivalent impedance ratio and interaction damping disturbance amount to the grid-type phase-locked loop bandwidth adaptive module.

[0033] The slow layer parameter update module is used to receive the multidimensional lookup table output by the offline lookup table generation module, the interaction strength and coupling level data output by the dynamic interaction identification module, and the parameter verification flag reported by the fast layer rate limiting control module, and iteratively generate the updated piecewise linear rate limiting curve parameters, which are then output to the fast layer rate limiting control module after security verification.

[0034] The fast-layer current limiting control module is used to receive multi-dimensional lookup tables, interaction intensity data and piecewise linear current limiting curve parameters with the converter control cycle as the execution beat. After lookup table interpolation and safe synthesis, it outputs dynamic current limiting parameters to the virtual magnetic flux phase-locked loop synchronization module.

[0035] The virtual flux linkage phase-locked loop synchronization module is used to receive the dynamic current limiting parameters output by the fast layer current limiting control module, estimate the virtual flux linkage vector and generate the control reference phase by combining the converter voltage and current sampling signals, and adaptively adjust the inertial damping factor according to the virtual flux linkage disturbance rate, and output it to each control loop of the grid-type photovoltaic converter.

[0036] The grid-connected phase-locked loop bandwidth adaptive module is used to receive the grid equivalent impedance ratio and interactive damping disturbance output by the dynamic interactive identification module, and adaptively adjust the loop filter cutoff frequency of the grid-connected photovoltaic converter phase-locked loop.

[0037] An electronic device includes a processor and a memory, wherein the memory stores a program that can implement the above-described coordinated control method for grid-connected and grid-linked photovoltaic converters, for execution by the processor.

[0038] The beneficial effects of this invention are as follows:

[0039] 1. This invention divides the time scale into two layers: fast-layer real-time response and slow-layer periodic update. The fast layer uses the microsecond-level converter control cycle as the beat, and realizes the rapid output of current limiting parameters through table lookup interpolation and simple logic operations to meet the high-speed response requirements under fault ride-through conditions. The slow layer performs system identification and parameter optimization iteration with a millisecond-level long cycle, which solves the technical contradiction that high-complexity calculations such as safety domain assessment and impedance identification cannot be adapted to the real-time control cycle, and achieves a balance between response speed and optimization accuracy.

[0040] 2. In the offline stage, this invention completes the operating condition screening based on the generalized operating short-circuit ratio safety domain. It pre-generates the optimal control parameters for all operating conditions through model prediction optimization with safety margin constraints, ensuring the stability of the parameters from the source. In the online stage, the optimal value from the lookup table and the piecewise linear current limiting value are combined by taking the smaller safe value. At the same time, a parameter verification and conservative parameter fallback mechanism are set to avoid control over-limit under extreme operating conditions, thus achieving a balance between optimality and safety.

[0041] 3. This invention adopts a virtual flux-oriented phase-locked loop (PLL)-free synchronization method for grid-type converters, which eliminates the dependence of traditional PLLs on grid voltage. Combined with the virtual flux disturbance rate adaptive adjustment of the inertial damping factor, it effectively suppresses phase jumps and oscillations caused by voltage drops and parameter switching, thereby improving the voltage support capability and synchronization stability of grid-type converters under weak grid fault scenarios.

[0042] 4. Based on the unified identification results of grid impedance and interactive damping, this invention synchronously adjusts the current limiting and damping parameters on the grid side and the phase-locked loop bandwidth on the grid side, avoiding impedance coupling conflicts caused by independent setting of the control parameters of the two types of converters; the phase-locked loop bandwidth is dynamically adjusted according to the operating conditions, weakening coupling oscillations under disturbed operating conditions and ensuring tracking accuracy under steady-state operating conditions, thus achieving a balance between stability and dynamic performance. Attached Figure Description

[0043] Figure 1 This is a schematic diagram illustrating the steps of a coordinated control method for grid-connected and grid-following photovoltaic converters according to the present invention. Detailed Implementation

[0044] A coordinated control method for grid-connected and grid-linked photovoltaic converters, applied to a photovoltaic power station with at least one grid-connected photovoltaic converter and at least one grid-linked photovoltaic converter connected to the same grid connection point, includes the following steps:

[0045] S1: Construct a parametric grid indexed by voltage sag depth, interaction strength between grid-connected and grid-connected photovoltaic converters, and filter grid points based on the generalized operating short-circuit ratio safety domain. Perform model prediction optimization with safety margin constraints on the effective grid points within the safety domain to solve for the current limiting threshold, virtual impedance value and active power loop correction coefficient corresponding to each grid point, forming a multidimensional lookup table.

[0046] S2: Establish a fast-layer real-time response mechanism: Using the converter control cycle as the execution beat, the grid connection point voltage is collected in real time and the voltage drop depth is calculated. The multi-dimensional lookup table of S1 is interpolated to obtain the lookup table control parameters in combination with the current effective interaction strength. At the same time, the fast current limiting threshold is calculated based on the piecewise linear current limiting curve. The lookup table current limiting threshold and the fast current limiting threshold are safely synthesized to output the final dynamic current limiting parameters.

[0047] S3: Establish a slow layer periodic update mechanism: Using a period longer than the converter control cycle as the execution cycle, identify the system interaction strength, coupling level and grid impedance parameters. Based on the current safety margin, interaction strength and the state of the multidimensional lookup table in S1, iteratively update the parameters of the piecewise linear current limiting curve. After safety verification, send it to the fast layer. At the same time, output the identified impedance and damping parameters to the grid-connected photovoltaic converter.

[0048] S4: Based on the dynamic current limiting parameters output by the fast layer of S2, current reference limiting, virtual impedance compensation and active power loop correction are performed respectively; then, the control reference phase is obtained by virtual flux orientation, and the inertial damping factor is adaptively adjusted according to the virtual flux disturbance rate to suppress phase disturbances caused by voltage drop and parameter switching.

[0049] S5: Based on the grid equivalent impedance ratio and interactive damping disturbance amount identified by the slow layer in S3, the cutoff frequency of the phase-locked loop filter is adaptively adjusted. Under disturbance conditions, the bandwidth of the phase-locked loop is reduced to weaken coupled oscillations, and under steady-state conditions, the bandwidth is restored to ensure phase tracking accuracy.

[0050] Step S1 specifically includes the following sub-steps:

[0051] S11: Read the grid connection point rated parameters, converter capacity and control parameters, and construct a two-dimensional parameter grid with voltage sag depth as the first index and interaction strength as the second index; the voltage sag depth is calculated based on the ratio of the difference between the grid connection point rated voltage and the real-time grid connection point voltage to the rated voltage.

[0052] S12: Based on the parameter grid constructed by S11, the generalized operating short-circuit ratio is used to determine the boundary of the system safety domain. The generalized operating short-circuit ratio safety margin of each grid point is calculated. Grid points with non-negative safety margins are determined as valid grid points, and grid points with negative safety margins are marked as forbidden boundaries. No usable control parameters are generated.

[0053] S13: Perform model prediction optimization with safety margin constraints on each valid grid point obtained from S12. The optimization objective is to minimize the comprehensive objective consisting of current tracking error, control increment smoothness and safety margin penalty. Solve for the current limiting threshold, virtual impedance value and active power loop correction coefficient corresponding to each grid point to form a multidimensional lookup table. Invalid grid points are filled with parameters of the adjacent more conservative interaction strength level or marked as forbidden.

[0054] Step S2 specifically includes the following sub-steps:

[0055] S21: Using the converter control cycle as the execution beat, the grid connection point voltage is collected in real time and the voltage drop depth is calculated. Combined with the current effective interaction strength, the multidimensional lookup table generated by S1 is interpolated to obtain the lookup table current limiting threshold, virtual impedance value and active power loop correction coefficient under the corresponding operating conditions.

[0056] S22: Calculate the fast current limiting threshold based on the piecewise linear current limiting curve, combine the lookup current limiting threshold obtained in S21 with the fast current limiting threshold by taking the smaller value for safety, and output the final dynamic current limiting parameter; when the deviation between the two exceeds the set threshold, maintain the previous effective virtual impedance value and send a parameter verification flag to the slow layer.

[0057] Step S3 specifically includes the following sub-steps:

[0058] S31: By injecting small-amplitude disturbances or utilizing the voltage and current responses under natural disturbances, the interaction strength, coupling level, grid equivalent impedance ratio, and interaction damping disturbance of the system are identified; S32: Using a period longer than the converter control cycle as the execution cycle, combined with the generalized operating short-circuit ratio safety margin of S1, the interaction strength identified by S31, and the multidimensional lookup table state of S1, the inflection point, slope, and intercept parameters of the updated piecewise linear current limiting curve are iteratively generated.

[0059] S33: Perform safety margin, version number and parameter change rate verification on the updated curve parameters generated by S32. If the verification passes, it is sent to the fast layer. If the verification fails, the fast layer maintains the valid curve parameters of the previous version and calls the conservative level control parameters in the multidimensional lookup table of S1.

[0060] Step S4 specifically includes the following sub-steps:

[0061] S41: Receives the dynamic current limiting parameters output by S2, limits the current reference value, performs virtual impedance compensation on the output current, and corrects the active power loop deviation; at the same time, it collects the grid connection point voltage and output current and converts them to a stationary two-phase coordinate system, and outputs the voltage α / β axis components and current α / β axis components to the extended state observer.

[0062] S42: Receive the voltage and current components of the stationary two-phase coordinate system output by S41 through the extended state observer, estimate the α and β axis components of the virtual flux linkage vector, calculate the virtual flux linkage angle, and use it as the unified control reference phase for each control loop of the grid-type photovoltaic converter.

[0063] S43: Calculate the virtual flux disturbance rate based on the virtual flux vector of the adjacent control cycle obtained in S42, and adaptively adjust the inertial damping factor based on the disturbance rate; when the disturbance rate increases, increase the damping factor to suppress phase disturbance, and when the disturbance rate decreases, gradually restore the damping factor to the basic value.

[0064] In step S42, the virtual flux linkage angle is determined by the arctangent of the virtual flux linkage β-axis component and α-axis component. This angle is synchronously output to the voltage outer loop, active power loop and virtual impedance loop as a common phase reference for each loop.

[0065] Step S5 specifically includes the following sub-steps:

[0066] S51: Based on the grid equivalent impedance ratio and interactive damping disturbance obtained from S3, calculate the target cutoff frequency of the phase-locked loop filter, and impose upper and lower limits on the target value through preset minimum and maximum cutoff frequencies.

[0067] S52: Dynamically adjust the PLL bandwidth based on the target cutoff frequency calculated in S51: When the grid equivalent impedance ratio increases and the amount of interactive damping disturbance increases, reduce the PLL cutoff frequency to weaken the coupled oscillation; when the system returns to steady state, gradually increase the cutoff frequency to ensure phase tracking accuracy; when the identification parameters are invalid, maintain the previous effective cutoff frequency or switch to a conservative low bandwidth value.

[0068] In step S51, the cutoff frequency of the phase-locked loop filter is calculated by superimposing the grid equivalent impedance ratio adjustment term and the interactive damping disturbance adjustment term from S3, based on the steady-state reference cutoff frequency. The final output value is limited to the range between the minimum allowable cutoff frequency and the maximum allowable cutoff frequency.

[0069] A collaborative control system for grid-connected and grid-following photovoltaic converters includes an offline lookup table generation module, a dynamic interactive identification module, a slow-layer parameter update module, a fast-layer current limiting control module, a virtual flux linkage phase-locked loop synchronization module, and a grid-following phase-locked loop bandwidth adaptive module.

[0070] The offline lookup table generation module is used to construct parameter grids, perform safety domain filtering and constrained model prediction optimization, generate multidimensional lookup tables indexed by voltage drop depth and interaction strength, and output the multidimensional lookup tables to the fast layer current limiting control module and the slow layer parameter update module respectively.

[0071] The dynamic interaction identification module is used to identify the system interaction strength, coupling level, grid equivalent impedance ratio and interaction damping disturbance amount. It outputs the interaction strength and coupling level data to the fast layer current limiting control module and the slow layer parameter update module, respectively, and outputs the grid equivalent impedance ratio and interaction damping disturbance amount to the grid-type phase-locked loop bandwidth adaptive module.

[0072] The slow layer parameter update module receives the multidimensional lookup table output by the offline lookup table generation module, the interaction strength and coupling level data output by the dynamic interaction identification module, and the parameter verification flag reported by the fast layer rate limiting control module. Iteratively generates the updated piecewise linear rate limiting curve parameters, and outputs them to the fast layer rate limiting control module after security verification.

[0073] The fast-layer current limiting control module is used to receive multi-dimensional lookup tables, interaction intensity data and piecewise linear current limiting curve parameters with the converter control cycle as the execution time. After lookup table interpolation and safe synthesis, it outputs dynamic current limiting parameters to the virtual magnetic flux phase-locked loop synchronization module.

[0074] The virtual flux linkage phase-locked loop synchronization module is used to receive the dynamic current limiting parameters output by the fast layer current limiting control module, estimate the virtual flux linkage vector and generate the control reference phase by combining the voltage and current sampling signals of the converter, and adaptively adjust the inertial damping factor according to the virtual flux linkage disturbance rate, and output it to each control loop of the grid-type photovoltaic converter.

[0075] The grid-type phase-locked loop bandwidth adaptive module is used to receive the grid equivalent impedance ratio and interactive damping disturbance output by the dynamic interactive identification module, and adaptively adjust the loop filter cutoff frequency of the grid-type photovoltaic converter phase-locked loop.

[0076] An electronic device includes a processor and a memory, the memory storing a program that can implement the above-described coordinated control method for grid-connected and grid-linked photovoltaic converters, for execution by the processor.

[0077] Example

[0078] The technical solutions of the present invention will now be described in detail with reference to the accompanying drawings.

[0079] like Figure 1 As shown, this embodiment provides a method for coordinated control of grid-connected and grid-connected photovoltaic converters, which is applied to a photovoltaic power station with at least one grid-connected photovoltaic converter and at least one grid-connected photovoltaic converter connected to the same grid connection point. The method includes steps S1 to S5, as detailed below.

[0080] S1, offline parameter mesh, security region constraints and multidimensional lookup table generation

[0081] S11, Define the grid for voltage drop depth and interaction strength parameters.

[0082] The offline lookup table generation module reads the grid-connected rated voltage, fault ride-through voltage curve, capacity, control cycle, rated current, virtual impedance range, active power loop correction range, and amplitude limit of injectable disturbances for grid-connected and grid-connected photovoltaic converters. The module first defines the parameter grid of voltage sag depth and interaction strength. The voltage sag depth is calculated according to Equation (1) and serves as the first index of the offline parameter grid and the real-time input for online fast layer current limiting.

[0083] Voltage drop depth is calculated according to formula (1):

[0084] ;

[0085] in, This refers to the voltage drop depth. The rated voltage at the grid connection point, expressed in V; This is the real-time voltage at the grid connection point, expressed in V.

[0086] The sources of the above variables are: The voltage is calculated from the rated voltage and the real-time voltage at the grid connection point according to formula (1); Obtained from grid connection point design parameters or grid connection protocol configuration; It is obtained from the grid connection point voltage sampling unit.

[0087] The usage and boundaries of the above variables are as follows: Used for offline grid indexing, fast layer current limiting and table lookup interpolation, updated every converter control cycle, and frozen or switched to exception handling when the voltage sampling is invalid or the rated voltage configuration is missing; Used for voltage drop depth normalization, updating after equipment parameter changes or grid connection point switching, and preventing the startup of offline table generation when configuration is missing or out of bounds, or freezing or switching to exception handling. Used to calculate voltage drop depth and trigger fault current limiting, updated every sampling period, and frozen or switched to exception handling when sampling channel is abnormal or timestamp is out of sync.

[0088] S12. Determine the boundary of the generalized operating short-circuit ratio safety domain.

[0089] The dynamic interaction identification module provides the interaction intensity range during the offline tuning or trial operation phase and writes the minimum safe short-circuit ratio under different interaction intensities into the safety domain boundary. The generalized operating short-circuit ratio safety margin is calculated according to formula (2). Only grid points with non-negative safety margins are allowed to enter the offline model prediction optimization; grid points with negative safety margins are only retained as forbidden boundaries and no usable lookup table parameters are generated.

[0090] The generalized operating short-circuit ratio safety margin is calculated according to formula (2):

[0091] ;

[0092] in, This refers to the generalized operating short-circuit ratio safety margin. This refers to the generalized operating short-circuit ratio; The minimum safe short-circuit ratio varies with interaction strength; The interaction strength.

[0093] The sources of the above variables are: It is obtained by calculating the generalized operating short-circuit ratio and the minimum safe short-circuit ratio according to formula (2); It is estimated from the grid strength at the grid connection point, converter capacity, and equivalent impedance identification results; Obtained from the offline small-disturbance stable security domain calibration table; It is obtained by the dynamic interactive identification module based on the impedance ratio and damping coefficient.

[0094] The usage and boundaries of the above variables are as follows: Used for offline security domain filtering and slow layer update constraints, updated when recalculated after the interaction strength or power grid equivalent parameters are updated, and frozen or switched to anomaly handling when writing aggressive parameters is prohibited when the security margin is negative. Used for safety margin calculation, grid impedance identification, or updating after capacity configuration changes; when the identification is invalid, the conservative level is used and the system is frozen or switched to anomaly handling. Used for safety margin calculation and forbidden boundary judgment, updated after interactive strength mesh or stable boundary calibration, and frozen or switched to anomaly handling when slow layer updates are prohibited when the safety domain table is missing. Used for security domain indexing, table lookup interpolation, and slow-level updates, updated for each valid identification window, and frozen or switched to exception handling when identification fails and the conservative high level is not maintained.

[0095] S13. Perform model prediction optimization and generate a multidimensional lookup table.

[0096] The offline lookup table generation module performs model prediction optimization at each valid grid point in S12. The optimization objective is represented by Equation (3), and its output is not directly written to the real-time control terminal. Instead, the current limiting threshold, virtual impedance value, and active power loop correction coefficient are pre-calculated and stored in the multidimensional lookup table according to the voltage drop depth and interaction strength in S11. If a grid point becomes invalid due to insufficient safety margin in S12, non-convergence of optimization, or output exceeding the hardware boundary, the module fills it with parameters of the adjacent more conservative interaction strength level or marks it as forbidden.

[0097] The offline model prediction optimization objective is expressed by equation (3):

[0098] ;

[0099] in, The target value is the prediction objective for the offline model, and its unit or dimension is the normalized target value; This refers to the prediction step number; To predict the step size; For the first Step reference current, in units or dimensions of A.

[0100] Among the other symbols in the formula, For the first Predict the current step, in units or dimensions of A; This is the current tracking weight matrix, with units or dimensions of normalized weights; For the first Step control increment, the unit or dimension is the control quantity unit; To control the incremental weight matrix, the units or dimensions are normalized weights.

[0101] Among the other symbols in the formula, This is the safety margin penalty coefficient; This is the generalized operating short-circuit ratio safety margin for S12.

[0102] The sources of the above variables are: It consists of the tracking term, the control increment term, and the safety margin penalty term in equation (3); Obtained from the model prediction optimization window definition; Obtained from the controller design phase configuration; It is generated by grid connection control objectives and fault ride-through current limiting strategies.

[0103] The sources of the remaining variables are: Calculated from the converter prediction model; Obtained through calibration during the system design phase; The step difference between adjacent variables is obtained by optimizing the solution from model prediction; Obtained through calibration during the system design phase.

[0104] The sources of the remaining variables are: Obtained by the priority specification of security domain constraints; The generalized operating short-circuit ratio and the minimum safe short-circuit ratio of S12 are calculated according to formula (2).

[0105] The usage and boundaries of the above variables are as follows: Used to solve the lookup parameters of each grid point offline; offline grid or weight matrix relabeling is updated periodically; when the optimization fails to converge, the grid point is marked as forbidden and frozen or transferred to anomaly handling. Used for offline objective function summation index, updated after prediction step size configuration changes, and frozen or switched to exception handling when offline optimization is not performed when window configuration is invalid; Used to limit the offline optimization prediction window, control the update after the cycle or changes in computing resource configuration, freeze or switch to anomaly handling when the offline grid density is reduced due to insufficient computing resources; Used for offline current tracking, updated when the offline optimization sample is updated, and frozen or switched to abnormal handling when used after the reference current exceeds the rated boundary and is limited.

[0106] The usage and boundaries of the remaining variables are as follows: Used for offline current tracking error calculation, updated when recalculated after changes in model parameters or grid points, and frozen or switched to anomaly handling when the grid point is forbidden when the model is unstable. Used to balance the current tracking term, control the target recalculation time update, and freeze or switch to anomaly handling when the weight matrix refuses optimization due to non-positive timing. Used to constrain abrupt changes in control variables, updated during offline optimization, and the solution becomes unavailable when the control increment boundary is exceeded; Used to balance the smoothness of control actions, update after changes in actuator boundaries or control cycles, and freeze or switch to exception handling when optimization is rejected due to non-positive timing of the weight matrix.

[0107] The usage and boundaries of the remaining variables are as follows: The parameter combination used to penalize insufficient safety margin is updated after the stability margin strategy changes. When the penalty coefficient is not calibrated, the output of aggressive parameters is prohibited or the process is frozen or switched to exception handling. Used for offline security domain filtering and slow layer update constraints, updated when recalculated after the interaction strength or power grid equivalent parameters are updated, and frozen or switched to exception handling when writing aggressive parameters is prohibited when the security margin is negative.

[0108] S2, Online Fast Layer Dynamic Rate Limiting and Lookup Table Control

[0109] S21. Interpolation to obtain lookup table parameters

[0110] A fast-layer real-time response mechanism is established. The online fast layer executes within each converter control cycle without waiting for slow-layer optimization. The fast layer reads the real-time grid-connected point voltage from the voltage sampling unit, obtains the voltage drop depth according to equation (1), and reads the strength of the most recent effective interaction from the dynamic interaction identification module. The fast layer obtains the lookup table parameters from the multidimensional lookup table of S1 according to equation (4). The lookup table parameters include the current limiting threshold, virtual impedance value, and active power loop correction coefficient, which are then directly fed into the current inner loop, virtual impedance loop, and active power loop of the grid-type photovoltaic converter.

[0111] The fast layer lookup table interpolation relationship is expressed by equation (4):

[0112] ;

[0113] in, The set of parameters for table lookup has units or dimensions that are a mixed set of dimensions. S1 is a multidimensional lookup table, with units or dimensions as parameter tables; This refers to the voltage drop depth. The interaction strength.

[0114] The sources of the above variables are: The multidimensional lookup table of S1 is obtained by interpolation according to equation (4); Generated by offline model prediction optimization of S1; The voltage is calculated from the rated voltage and the real-time voltage at the grid connection point according to formula (1); It is obtained by the dynamic interactive identification module based on the impedance ratio and damping coefficient.

[0115] The usage and boundaries of the above variables are as follows: Used for fast-layer dynamic current limiting, virtual impedance control and active power loop correction. Updated after each control cycle or interaction strength update. When the lookup point is invalid, the conservative adjacent point is selected and the process is frozen or switched to anomaly handling. Provides fast layer interpolation reading, offline optimization is completed and updated after verification, and freezes or switches to exception handling when the table version number is not continuous and keeps the previous version. Used for offline grid indexing, fast layer current limiting and table lookup interpolation, updated every converter control cycle, and frozen or switched to exception handling when the voltage sampling is invalid or the rated voltage configuration is missing; Used for security domain indexing, table lookup interpolation, and slow-level updates, updated for each valid identification window, and frozen or switched to exception handling when identification fails and the conservative high level is not maintained.

[0116] S22. Calculate the fast current limiting threshold and perform secure synthesis.

[0117] The fast layer also calculates the fast current limiting threshold according to the piecewise linear current limiting curve, which is represented by equation (5). The fast layer combines the fast current limiting threshold of equation (5) with the lookup current limiting threshold of S21 for safety, and preferentially adopts the smaller current limiting value that does not exceed the boundary of the safety domain; when the difference between the two exceeds the set deviation, the fast layer maintains the previous effective virtual impedance value and issues a slow layer verification flag.

[0118] The piecewise linear current limiting curve is expressed by equation (5):

[0119] ;

[0120] in, The fast current limiting threshold is expressed in units or dimensions as A. For the first The slope coefficient of the segment, with units or dimensions in A; This refers to the voltage drop depth. For the first The segment intercept coefficient, with units or dimensions in A.

[0121] Among the other symbols in the formula, For the first Voltage drop inflection point; For the first The inflection point of voltage drop.

[0122] The sources of the above variables are: It is obtained by calculating the piecewise linear current limiting curve according to equation (5); Obtained from slow-layer curve parameter updates or offline initial calibration; The voltage is calculated from the rated voltage and the real-time voltage at the grid connection point according to formula (1); Obtained from slow layer curve parameter updates or offline initial calibration.

[0123] The sources of the remaining variables are: Obtained from slow-layer curve parameter updates or offline initial calibration; Obtained from slow layer curve parameter updates or offline initial calibration.

[0124] The usage and boundaries of the above variables are as follows: Used for current inner loop current limiting and virtual impedance control, each control cycle is updated according to the real-time voltage drop depth, and when the calculation result exceeds the hardware boundary, a safe boundary value is used to freeze or switch to abnormal handling. Used to calculate the fast rate limiting threshold, updated after receiving the valid curve parameters output by the slow layer, and frozen when the version verification fails or when the previous valid value is frozen or transferred to exception handling. Used for offline grid indexing, fast layer current limiting and table lookup interpolation, updated every converter control cycle, and frozen or switched to exception handling when the voltage sampling is invalid or the rated voltage configuration is missing; Used to calculate the fast rate limiting threshold, updated after receiving valid curve parameters from the slow layer output, and frozen when version verification fails, or switched to exception handling.

[0125] The usage and boundaries of the remaining variables are as follows: Used to determine segmented intervals, updated after receiving valid curve parameters from the slow layer output, and frozen or switched to exception handling when rejecting the use when the inflection point sequence is abnormal. Used to define the upper boundary of the current segment, and updated after receiving valid curve parameters from the slow layer output.

[0126] S3, Slow Layer Curve Parameter Update and Dynamic Interaction Identification

[0127] S31. Identify interaction strength, coupling level, and impedance parameters.

[0128] A slow-layer periodic update mechanism is established, with the slow layer executing within a communication cycle longer than the converter control cycle. Voltage and current responses of grid-connected and grid-linked photovoltaic converters are obtained through small-amplitude disturbance injection. Equivalent output impedance, equivalent grid impedance ratio, and cross-damping coefficient are calculated, resulting in cross-interaction strength and coupling level. The cross-interaction strength is fed into the fast-layer lookup table index and the slow-layer curve update, while the coupling level is fed into the slow-layer update trigger logic, outputting the equivalent grid impedance ratio and cross-interaction damping disturbance amount. If disturbance injection is restricted by grid connection regulations, the voltage and current responses under natural disturbances are used for recursive identification, and the slow-layer update frequency is reduced.

[0129] S32. Determine the curve parameters for the next version.

[0130] The slow layer reads the interaction strength and coupling level of S31, the safety margin and multidimensional lookup table version number of S1, and the fast layer verification flag of S2, and determines the next version of the curve parameters through the curve parameter update mapping shown in Equation (6). Equation (6) represents the engineering calibration mapping rather than the new empirical weight score: its input is the safety margin of S1, the interaction strength of S31, and the lookup table status of S1, and the output is a set of curve parameters consisting of inflection point parameters, slope coefficients, and intercept coefficients.

[0131] The slow-layer curve parameter update mapping is expressed by equation (6):

[0132] ;

[0133] in, The updated set of curve parameters, with units or dimensions being a mixed-dimensional set; S1 represents the generalized operating short-circuit ratio safety margin. The interaction strength of S31; This is a multidimensional lookup table for S1, with units or dimensions represented by a parameter table.

[0134] The sources of the above variables are: Obtained by mapping the slow layer according to equation (6); The generalized operating short-circuit ratio and the minimum safe short-circuit ratio of S1 are calculated according to equation (2); The dynamic interactive identification module of S31 obtains the results based on the impedance ratio and damping coefficient. It is generated by offline model prediction optimization of S1.

[0135] The usage and boundaries of the above variables are as follows: The segmented current limiting curve is used for the next control cycle and is updated every effective slow layer communication cycle. When the safety margin is negative or the version check fails, it is not output and the previous effective parameter is frozen or the exception handling is switched. Used for offline security domain filtering and slow layer update constraints, updated when recalculated after the interaction strength or power grid equivalent parameters are updated, and frozen or switched to anomaly handling when the output of aggressive parameters is prohibited when the security margin is negative. Used for security domain indexing, table lookup interpolation and slow layer updates, updated for each valid identification window, and frozen or switched to exception handling when identification fails and the conservative high level is not maintained. Provides fast layer interpolation reading, updates after offline optimization and verification, and freezes or switches to exception handling when the table version number is not continuous and the previous version is retained.

[0136] S33. Verify and output the updated curve parameters.

[0137] The slow layer outputs updated curve parameters only if the updated curve parameters in S32 satisfy the following conditions: non-negative safety margin, continuous version number, correct checksum, and parameter change rate not exceeding the limit. The fast layer only receives curve parameters with a complete version number and a valid flag. If the slow layer fails to converge within the communication cycle, outputs parameter verification fails, or has a negative safety margin, the fast layer freezes the valid curve parameters of the previous version and selects a conservative parameter corresponding to a higher interaction strength from the multidimensional lookup table in S1.

[0138] S4. Dynamic current limiting and virtual flux linkage directional synchronization control for grid-type photovoltaic converters

[0139] S41. Receive the output parameters of S2 and execute the control of the grid-type photovoltaic converter.

[0140] The grid-type photovoltaic converter receives the final dynamic current limiting command, virtual impedance value, and active power loop correction coefficient from S2.

[0141] The current reference value of the grid-type photovoltaic converter is limited according to the final dynamic current limiting command of S2, and the limited current reference value is input into the inner current loop to ensure that the output current of the grid-type photovoltaic converter does not exceed the allowable current limiting range.

[0142] The output current of the grid-type photovoltaic converter is compensated for virtual impedance based on the virtual impedance value of S2 to obtain the virtual impedance compensation voltage. The voltage control command is then corrected using the virtual impedance compensation voltage to suppress current surges and interactive oscillations between the grid-type photovoltaic converter and the power grid.

[0143] The deviation between the active power reference value and the actual output active power is corrected based on the active power loop correction coefficient of S2, and the corrected active power deviation is input into the active power loop to adjust the active power control quantity and phase change rate of the grid-type photovoltaic converter.

[0144] During the execution of dynamic current limiting, virtual impedance compensation, and active power loop correction, the grid-connected voltage and output current of the grid-connected photovoltaic converter are collected, and the voltage and current state quantities required for virtual flux linkage observation are generated in combination with the virtual impedance compensation results.

[0145] Transforming the voltage and current state variables to a stationary two-phase coordinate system yields the voltage... Axial components and Shaft components and output current Axial components and The axis component is calculated and output to the extended state observer as input for S42 to estimate the virtual flux linkage vector.

[0146] Meanwhile, based on the voltage state variables, current state variables, and virtual flux estimation results in adjacent control cycles of the grid-type photovoltaic converter, basic data is provided for S43 to calculate the virtual flux disturbance rate and adjust the adaptive inertial damping factor.

[0147] When the final dynamic current limiting command, virtual impedance value, or active power loop correction coefficient of S2 is invalid or exceeds the allowable range, the control is executed using the previous valid parameter or the preset conservative parameter; when the grid connection point voltage or output current sampling is abnormal, the previous valid observation state is maintained, and the phase freeze or abnormal handling is switched to.

[0148] S42. Estimate the virtual flux linkage vector and determine the control reference phase.

[0149] Grid-type photovoltaic converters employ virtual flux-oriented phase-locked loop (PLL) synchronization. An extended state observer receives the voltage from S41. Axial components and Shaft components and output current Axial components and Axis components, estimating the virtual flux linkage vector Axial components and Axial components.

[0150] The virtual flux linkage angle is determined according to equation (7):

[0151] ;

[0152] in, This is a virtual flux linkage angle, with units or dimensions in rad. For virtual magnet links Axial components; For virtual magnet links Axial components.

[0153] virtual magnet link angle As the control reference phase for grid-type photovoltaic converters, and input voltage outer loop, active power loop and virtual impedance loop, the dynamic current limiting control, virtual impedance control and active power loop correction of S41 are all performed based on the same virtual flux linkage phase reference.

[0154] The sources of the above variables are: The virtual flux linkage is calculated from the two-phase components according to equation (7); The extended state observer determines the voltage of S41. Shaft components and output current Obtained by shaft component estimation; The extended state observer determines the voltage of S41. Shaft components and output current The shaft components were estimated.

[0155] The usage and boundaries of the above variables are as follows: As the control reference phase of the grid-type photovoltaic converter, it is updated by the extended state observer in each control cycle. When the flux estimation fails, the previous effective phase is maintained and the control bandwidth is reduced or anomaly handling is initiated. Used to calculate the virtual flux linkage angle, updated every control cycle, and enters phase freeze when estimating divergence; Used to calculate the virtual flux linkage angle, updated every control cycle, and enters phase freeze when estimating divergence.

[0156] S43. Calculate the virtual flux disturbance rate and adjust the inertial damping factor.

[0157] To suppress flux disturbances caused by voltage dips, harmonics, dynamic current limiting switching, virtual impedance parameter changes, and sampling noise, the virtual flux disturbance rate is calculated based on the virtual flux vector of the current control cycle and the virtual flux vector of the previous control cycle in S42.

[0158] The virtual flux perturbation rate is determined by the following formula:

[0159] ;

[0160] in, For the first Virtual flux disturbance rate per control cycle; For S42's The virtual flux vector estimated by the extended state observer in each control cycle; The virtual flux linkage vector estimated for the previous control cycle of S42; This refers to the control cycle of a grid-type photovoltaic converter.

[0161] The adaptive inertial damping factor is determined according to equation (8):

[0162] ;

[0163] in, This is the adaptive inertial damping factor; The basic inertial damping factor; This is the flux linkage perturbation gain; The virtual flux perturbation rate of S42 is expressed in units or dimensions as Wb / s or normalized perturbation rate.

[0164] When the virtual flux disturbance rate increases, the adaptive inertial damping factor is increased to reduce the impact of voltage drop, current limiting parameter switching, virtual impedance changes and high-frequency noise on the control reference phase. After the virtual flux disturbance rate recovers to the steady-state range, the adaptive inertial damping factor is gradually restored to near the basic inertial damping factor according to the preset decline slope to maintain the phase response speed of the grid-type photovoltaic converter.

[0165] The sources of the above variables are: The virtual flux linkage vector and control cycle of adjacent control cycles of S42 Calculated; The virtual flux disturbance rate of S42 is calculated by equation (8) from the basic inertial damping factor and the virtual flux disturbance rate of S42. Obtained from the control parameter calibration of a grid-type photovoltaic converter; The calibration results are obtained from the disturbance rejection performance of the extended state observer and the stability requirements of the network control.

[0166] The usage and boundaries of the above variables are as follows: It is used to reflect the degree of change of the virtual flux vector in adjacent control cycles. It is updated in each control cycle and a preset conservative disturbance level is adopted when the state of the extended state observer is abnormal. It is used to suppress virtual flux phase disturbances. It is updated according to the virtual flux disturbance rate in each control cycle. When the calculation result exceeds the allowable range, it is limited to the preset damping upper and lower limits.

[0167] The base value used to form the adaptive inertial damping factor is updated after the network control parameters are recalibrated, and the safe default value is used when calibration is not completed. Used to map virtual flux disturbance rate to damping increment, updated after disturbance suppression strategy changes, and uses conservative default value when gain goes out of bounds.

[0168] S5, Adaptive control of phase-locked loop bandwidth for grid-connected photovoltaic converters

[0169] S51. Identify impedance parameters and determine the phase-locked loop cutoff frequency.

[0170] The grid-connected photovoltaic converter adaptively adjusts the cutoff frequency of the phase-locked loop (PLL) filter based on the equivalent grid impedance ratio and interactive damping disturbance of S31. The cutoff frequency of the PLL filter is not fixed. During the phase detection process, the PLL uses the phase detection error signal and the voltage and current response at the grid connection point to identify the equivalent grid impedance ratio and interactive damping disturbance. The cutoff frequency is determined according to equation (9), and the minimum and maximum cutoff frequencies form upper and lower limits.

[0171] The cutoff frequency of the phase-locked loop filter is determined according to equation (9):

[0172] ;

[0173] in, This is the cutoff frequency of the phase-locked loop filter; This is the steady-state reference cutoff frequency; This is the impedance ratio adjustment coefficient; The equivalent impedance ratio of the power grid for S31.

[0174] Among the other symbols in the formula, This is the interactive damping adjustment coefficient; For the interactive damping disturbance of S31; The minimum allowable cutoff frequency, expressed in rad / s or Hz; The maximum permissible cutoff frequency, expressed in rad / s or Hz.

[0175] The sources of the above variables are: The impedance ratio and damping disturbance of S31 are calculated according to equation (9) and obtained through boundary constraints; Obtained from the design parameters of the mesh-type phase-locked loop; Obtained from weak grid impedance sensitivity calibration; It is obtained by phase detection signal and voltage and current response.

[0176] The sources of the remaining variables are: Obtained from the interactive damping sensitivity calibration; It is obtained by the dynamic interactive identification module of S31 based on the change of damping coefficient; Obtained from the stability and lower limit calibration of the phase-locked loop; It is obtained from the stability of the phase-locked loop and the noise amplification boundary calibration.

[0177] The usage and boundaries of the above variables are as follows: Used in a mesh-type phase-locked loop filter, it updates each valid identification window, and when identification is invalid, it maintains the previous valid frequency or freezes or switches to abnormal handling when conservatively using low bandwidth. For the molecule of equation (9), the phase-locked loop parameters are recalibrated and updated at the timing; if not calibrated, adaptive bandwidth output is prohibited. Used to reduce bandwidth based on the equivalent impedance ratio of the power grid, updated after changes in the weak grid adaptation strategy, and frozen or switched to abnormal handling when the coefficient exceeds the limit and a conservative default value is used. Used for phase-locked loop bandwidth adaptation, updated for each valid identification window, and frozen or switched to exception handling when identification fails, maintaining the previous valid value or conservatively high value.

[0178] The usage and boundaries of the remaining variables are as follows: Used to reduce bandwidth based on damping disturbance amount, updated after damping identification strategy changes, and frozen or switched to exception handling when conservative default value is used when coefficients go out of bounds. Used for phase-locked loop bandwidth adaptation, updated for each valid identification window, and frozen or switched to exception handling when identification fails and set to a conservative perturbation value. Used for the lower limit constraint of equation (9), updated after the grid connection specification or controller parameters are changed, and frozen or transferred to abnormal handling when the configuration is missing, a conservative lower limit is adopted. For the upper limit constraint of equation (9), after the grid connection specification or controller parameters are changed and updated, when the configuration is missing, the conservative upper limit is used to freeze or switch to abnormal handling.

[0179] S52. Dynamically adjust the phase-locked loop bandwidth according to the disturbance conditions.

[0180] When the equivalent impedance ratio of the grid in S51 increases or the amount of cross-damped disturbance increases, the denominator of equation (9) of S51 increases, causing the cutoff frequency to decrease, thereby weakening the coupling of the phase-locked loop to weak grid disturbances. When the system returns to steady state, the cutoff frequency of S51 gradually recovers to a higher value within the allowable range to improve phase tracking accuracy. If the equivalent impedance ratio or cross-damped disturbance of S51 is invalid, the bandwidth adaptive module of the grid-type phase-locked loop does not use the identification result of this identification, but maintains the previous effective cutoff frequency of S51 or switches to a conservative low bandwidth value.

Claims

1. A method for coordinated control of grid-connected and grid-following photovoltaic converters, applied to a photovoltaic power station with at least one grid-connected photovoltaic converter and at least one grid-following photovoltaic converter connected to the same grid connection point, characterized in that, Includes the following steps: S1: Construct a parametric grid indexed by voltage sag depth, interaction strength between grid-connected and grid-connected photovoltaic converters, and filter grid points based on the generalized operating short-circuit ratio safety domain. Perform model prediction optimization with safety margin constraints on the effective grid points within the safety domain to solve for the current limiting threshold, virtual impedance value and active power loop correction coefficient corresponding to each grid point, forming a multidimensional lookup table. S2: Establish a fast-layer real-time response mechanism: Using the converter control cycle as the execution cycle, collect the grid connection point voltage in real time and calculate the voltage drop depth. Combine the current effective interaction strength to interpolate the multi-dimensional lookup table mentioned in S1 to obtain the lookup table control parameters. At the same time, calculate the fast current limiting threshold based on the piecewise linear current limiting curve, and safely synthesize the lookup table current limiting threshold and the fast current limiting threshold to output the final dynamic current limiting parameters. S3: Establish a slow layer periodic update mechanism: Using a period longer than the converter control cycle as the execution cycle, identify the system interaction strength, coupling level and grid impedance parameters. Based on the current safety margin, interaction strength and the state of the multidimensional lookup table in S1, iteratively update the parameters of the piecewise linear current limiting curve. After safety verification, send it to the fast layer. At the same time, output the identified impedance and damping parameters to the grid-connected photovoltaic converter. S4: Based on the dynamic current limiting parameters output by the fast layer of S2, current reference limiting, virtual impedance compensation and active power loop correction are performed respectively; then, the control reference phase is obtained by virtual flux orientation, and the inertial damping factor is adaptively adjusted according to the virtual flux disturbance rate to suppress phase disturbances caused by voltage drop and parameter switching. S5: Based on the grid equivalent impedance ratio and interactive damping disturbance amount identified by the slow layer in S3, the cutoff frequency of the phase-locked loop filter is adaptively adjusted. Under disturbance conditions, the bandwidth of the phase-locked loop is reduced to weaken coupled oscillations, and under steady-state conditions, the bandwidth is restored to ensure phase tracking accuracy.

2. The method for coordinated control of grid-connected and grid-following photovoltaic converters according to claim 1, characterized in that, Step S1 specifically includes the following sub-steps: S11: Read the grid connection point rated parameters, converter capacity and control parameters, and construct a two-dimensional parameter grid with voltage drop depth as the first index and interaction strength as the second index; the voltage drop depth is calculated based on the ratio of the difference between the grid connection point rated voltage and the real-time grid connection point voltage to the rated voltage. S12: Based on the parameter grid constructed by S11, the generalized operating short-circuit ratio is used to determine the boundary of the system safety domain. The generalized operating short-circuit ratio safety margin of each grid point is calculated. Grid points with non-negative safety margins are determined as valid grid points, and grid points with negative safety margins are marked as forbidden boundaries. No usable control parameters are generated. S13: Perform model prediction optimization with safety margin constraints on each valid grid point obtained from S12. The optimization objective is to minimize the comprehensive objective consisting of current tracking error, control increment smoothness and safety margin penalty. Solve for the current limiting threshold, virtual impedance value and active power loop correction coefficient corresponding to each grid point to form a multidimensional lookup table. Invalid grid points are filled with parameters of the adjacent more conservative interaction strength level or marked as forbidden.

3. The method for coordinated control of grid-connected and grid-following photovoltaic converters according to claim 1, characterized in that, Step S2 specifically includes the following sub-steps: S21: Using the converter control cycle as the execution beat, the grid connection point voltage is collected in real time and the voltage drop depth is calculated. Combined with the current effective interaction strength, the multidimensional lookup table generated in S1 is interpolated to obtain the lookup table current limiting threshold, virtual impedance value and active power loop correction coefficient under the corresponding operating conditions. S22: Calculate the fast current limiting threshold based on the piecewise linear current limiting curve, combine the lookup current limiting threshold obtained in S21 with the fast current limiting threshold by taking the smaller value for safety, and output the final dynamic current limiting parameter; when the deviation between the two exceeds the set threshold, maintain the previous effective virtual impedance value and send a parameter verification flag to the slow layer.

4. The method for coordinated control of grid-connected and grid-following photovoltaic converters according to claim 1, characterized in that, Step S3 specifically includes the following sub-steps: S31: By injecting small-amplitude disturbances or utilizing the voltage and current responses under natural disturbances, the interaction strength, coupling level, grid equivalent impedance ratio, and interaction damping disturbance amount of the system can be identified. S32: Using a period longer than the converter control cycle as the execution cycle, combined with the generalized operating short-circuit ratio safety margin of S1, the interaction strength identified by S31 and the multidimensional lookup table state of S1, the inflection point, slope and intercept parameters of the updated piecewise linear current limiting curve are iteratively generated. S33: Perform safety margin, version number and parameter change rate verification on the updated curve parameters generated by S32. If the verification passes, it is sent to the fast layer. If the verification fails, the fast layer maintains the valid curve parameters of the previous version and calls the conservative level control parameters in the multidimensional lookup table of S1.

5. The method for coordinated control of grid-connected and grid-following photovoltaic converters according to claim 1, characterized in that, Step S4 specifically includes the following sub-steps: S41: Receives the dynamic current limiting parameters output by S2, limits the current reference value, performs virtual impedance compensation on the output current, and corrects the active power loop deviation; at the same time, it collects the grid connection point voltage and output current and converts them to a stationary two-phase coordinate system, and outputs the voltage α / β axis components and current α / β axis components to the extended state observer. S42: Receive the voltage and current components of the stationary two-phase coordinate system output by S41 through the extended state observer, estimate the α and β axis components of the virtual flux linkage vector, calculate the virtual flux linkage angle, and use it as the unified control reference phase for each control loop of the grid-type photovoltaic converter. S43: Calculate the virtual flux disturbance rate based on the virtual flux vector of the adjacent control cycle obtained in S42, and adaptively adjust the inertial damping factor based on the disturbance rate; When the disturbance rate increases, the damping factor is increased to suppress phase disturbances; when the disturbance rate decreases, the damping factor is gradually restored to the baseline value.

6. The method for coordinated control of grid-connected and grid-following photovoltaic converters according to claim 5, characterized in that, In step S42, the virtual flux linkage angle is determined by the arctangent of the virtual flux linkage β-axis component and α-axis component. This angle is synchronously output to the voltage outer loop, active power loop and virtual impedance loop as a common phase reference for each loop.

7. The method for coordinated control of grid-connected and grid-following photovoltaic converters according to claim 1, characterized in that, Step S5 specifically includes the following sub-steps: S51: Based on the grid equivalent impedance ratio and interactive damping disturbance obtained from S3, calculate the target cutoff frequency of the phase-locked loop filter, and impose upper and lower limits on the target value through preset minimum and maximum cutoff frequencies. S52: Dynamically adjust the PLL bandwidth based on the target cutoff frequency calculated in S51: When the grid equivalent impedance ratio increases and the amount of interactive damping disturbance increases, reduce the PLL cutoff frequency to weaken the coupled oscillation; when the system returns to steady state, gradually increase the cutoff frequency to ensure phase tracking accuracy; when the identification parameters are invalid, maintain the previous effective cutoff frequency or switch to a conservative low bandwidth value.

8. The method for coordinated control of grid-connected and grid-following photovoltaic converters according to claim 7, characterized in that, In step S51, the cutoff frequency of the phase-locked loop filter is calculated by superimposing the grid equivalent impedance ratio adjustment term and the interactive damping disturbance adjustment term from S3, based on the steady-state reference cutoff frequency. The final output value is limited to the range between the minimum allowable cutoff frequency and the maximum allowable cutoff frequency.

9. A coordinated control system for grid-connected and grid-following photovoltaic converters, characterized in that, It includes an offline lookup table generation module, a dynamic interactive identification module, a slow layer parameter update module, a fast layer rate limiting control module, a virtual magnetic flux non-locked loop synchronization module, and a network-type phase-locked loop bandwidth adaptive module. The offline lookup table generation module is used to construct a parameter grid, perform security domain filtering and constrained model prediction optimization, generate a multidimensional lookup table indexed by voltage drop depth and interaction strength, and output the multidimensional lookup table to the fast layer current limiting control module and the slow layer parameter update module respectively. The dynamic interaction identification module is used to identify the system interaction strength, coupling level, grid equivalent impedance ratio and interaction damping disturbance amount, and outputs the interaction strength and coupling level data to the fast layer current limiting control module and the slow layer parameter update module respectively, and outputs the grid equivalent impedance ratio and interaction damping disturbance amount to the grid-type phase-locked loop bandwidth adaptive module. The slow layer parameter update module is used to receive the multidimensional lookup table output by the offline lookup table generation module, the interaction strength and coupling level data output by the dynamic interaction identification module, and the parameter verification flag reported by the fast layer rate limiting control module, and iteratively generate the updated piecewise linear rate limiting curve parameters, which are then output to the fast layer rate limiting control module after security verification. The fast-layer current limiting control module is used to receive multi-dimensional lookup tables, interaction intensity data and piecewise linear current limiting curve parameters with the converter control cycle as the execution beat. After lookup table interpolation and safe synthesis, it outputs dynamic current limiting parameters to the virtual magnetic flux phase-locked loop synchronization module. The virtual flux linkage phase-locked loop synchronization module is used to receive the dynamic current limiting parameters output by the fast layer current limiting control module, estimate the virtual flux linkage vector and generate the control reference phase by combining the converter voltage and current sampling signals, and adaptively adjust the inertial damping factor according to the virtual flux linkage disturbance rate, and output it to each control loop of the grid-type photovoltaic converter. The grid-connected phase-locked loop bandwidth adaptive module is used to receive the grid equivalent impedance ratio and interactive damping disturbance output by the dynamic interactive identification module, and adaptively adjust the loop filter cutoff frequency of the grid-connected photovoltaic converter phase-locked loop.

10. An electronic device, comprising a processor and a memory, characterized in that, The memory stores a program that can implement the coordinated control method for grid-connected and grid-following photovoltaic converters as described in any one of claims 1 to 8, for execution by a processor.

Images