Network-constructed converter adaptive virtual resistance fault overcurrent suppression method and system

Abstract

The embodiment of the application relates to the technical field of smart grid, and particularly relates to a network-constructing type converter adaptive virtual resistance fault overcurrent inhibition method and system, the method comprising: establishing a basic control model based on a virtual synchronous machine control framework; establishing a transient current analysis model during power grid fault, and analyzing the influence mechanism of virtual resistance on fault current; combining low voltage ride through requirements and converter overcurrent capacity to determine the safe regulation range of virtual resistance; constructing a multi-scenario weighted algorithm within the range to calculate the optimal starting value of virtual resistance under different scenarios; taking the optimal starting value as a starting point, constructing a three-layer game model and coupling a dynamic programming algorithm to form a distributed collaborative optimization framework, and solving the optimal control value of virtual resistance; embedding the optimal control value into a voltage-current double-loop control to replace the original virtual resistance setting to participate in closed-loop regulation. The application can effectively inhibit fault overcurrent, and give consideration to single-machine protection and overall system stability.

Description

Technical Field

[0001] This application relates to the field of smart grid technology, and in particular to a method and system for suppressing overcurrent in grid-type converters under adaptive virtual resistance faults. Background Technology

[0002] With the continuous increase in the proportion of renewable energy grid connection, grid-connected converters, due to their independent voltage and frequency support capabilities, have become the core equipment and key technology direction for future high-proportion renewable energy power systems. Grid-connected virtual synchronous generator (VSG) control can achieve autonomous control of active power-frequency and reactive power-voltage by simulating the rotor motion characteristics and excitation regulation mechanism of a synchronous generator. Under normal operating conditions, it can provide effective inertia and damping support for the power grid, exhibiting excellent stability performance.

[0003] However, in actual power grid operation, three-phase symmetrical short-circuit faults occur frequently. Under these conditions, grid-connected VSGs generate significant transient current surges, threatening not only the safety of converter power devices but also potentially leading to power distribution imbalances in multi-unit parallel systems. Particularly in multi-unit cluster grid-connected scenarios, traditional fixed-parameter virtual resistance control strategies struggle to simultaneously meet the dual requirements of single-unit overcurrent protection and multi-unit collaborative stability. Once parameters are set, they cannot adaptively adjust according to fault depth, easily leading to overcurrent tripping or even severe consequences like system oscillation and instability. Furthermore, existing virtual resistance optimization methods are mostly designed for single-unit systems, lacking effective multi-unit collaborative optimization mechanisms and failing to address system stability issues caused by inconsistent multi-unit regulation. Therefore, establishing an adaptive virtual resistance control method with multi-unit collaborative capabilities to achieve collaborative overcurrent suppression under three-phase symmetrical faults has become a critical technical challenge urgently needing to be solved in current high-proportion renewable energy power systems. Summary of the Invention

[0004] To address this issue, the present invention provides an adaptive virtual resistance fault overcurrent suppression method and system for grid-type converters, which overcomes the problem in the prior art where the virtual resistance parameters are fixed and there is a lack of multi-machine coordination, making it difficult to balance fault overcurrent suppression and system stability.

[0005] To achieve the above objectives, this invention provides a method for suppressing overcurrent in a grid-type converter under adaptive virtual resistance faults. This includes:

[0006] Step S1: Based on the virtual synchronous machine control framework, establish the basic control model of the grid-type converter;

[0007] Step S2: Based on the basic control model, establish an analytical model of transient current during power grid faults, and analyze the influence mechanism of virtual resistance on fault current peak value, current change rate, DC decay rate and system damping.

[0008] Step S3: Based on the aforementioned influence mechanism, and in conjunction with the low voltage ride-through requirement and the converter's overcurrent capability, derive the adaptive upper and lower limits of the virtual resistance, and determine the safe control range of the virtual resistance.

[0009] Step S4: Within the safety control range, construct a multi-scenario weighted algorithm, comprehensively consider system inertia, line impedance and transient power requirements, and calculate the optimal initial value of virtual resistance under different scenarios.

[0010] Step S5: Starting from the optimal initial value, construct a three-layer game model and couple it with a dynamic programming algorithm to form a distributed game-dynamic programming collaborative optimization framework, and solve for the optimal control value of the output virtual resistor.

[0011] Step S6: Embed the optimal control value into the voltage and current dual-loop control of the converter, replace the original virtual resistance setpoint, and use the updated virtual resistance value to participate in the dual-loop closed-loop regulation calculation.

[0012] Furthermore, the basic control model in step S1 comprises three parallel sub-models:

[0013] Step S11: By simulating the rotor motion characteristics and excitation regulation mechanism of the synchronous generator, a rotor motion equation is established, which includes the inertia time coefficient and the damping coefficient.

[0014] Step S12: Reactive voltage control is achieved by using a droop method. A reactive voltage droop control equation is established, which includes a droop coefficient.

[0015] Step S13: Embed the virtual resistor in series between the output side of the virtual synchronous machine and the dual-loop control, and establish the core control equation of the virtual resistor. The virtual resistor value in the core control equation is the variable to be optimized.

[0016] Further, step S2 includes:

[0017] Step S21: Based on the basic control model, for a three-phase symmetrical short-circuit fault, establish a transient expression for the fault current according to the step response of the RL circuit.

[0018] Step S22: Based on the transient expression of the fault current and the virtual resistance value, establish a calculation formula for the correlation between the peak value of the fault current and the peak value of the virtual resistance, as well as a calculation formula for the correlation between the rate of change of the fault current and the rate of change of the virtual resistance in the initial stage of the fault.

[0019] Step S23: Based on the peak correlation calculation formula and the rate of change correlation calculation formula, it is determined that increasing the virtual resistance can simultaneously reduce the peak fault current and the rate of change of the initial fault current, and accelerate the attenuation speed of the DC component.

[0020] Furthermore, step S2 also includes:

[0021] Step S24: Perform small-signal modeling on the multi-machine parallel system with virtual resistance and establish a linearized expression for the electromagnetic power of a single machine: The linearized expression for the electromagnetic power of a single machine includes the power angle synchronization coefficient, the voltage synchronization coefficient, and the influence coefficient of neighboring units;

[0022] Step S25: Based on the small signal modeling results, establish an equivalent synchronization coefficient expression for a single machine. The equivalent synchronization coefficient expression for a single machine is determined by the output voltage amplitude before the fault, the voltage amplitude at the grid connection point, the total output impedance amplitude, the power angle before the fault, and the total impedance angle.

[0023] Step S26: Based on the equivalent synchronization coefficient expression, establish the expression for the undamped natural oscillation frequency and the equivalent damping ratio of the single machine, and determine that increasing the virtual resistance can reduce the undamped natural oscillation frequency and increase the equivalent damping ratio, so as to characterize the influence mechanism of the virtual resistance on the system damping characteristics.

[0024] Further, step S3 includes:

[0025] Step S31: Based on the low voltage ride-through requirements, establish the expression for the minimum reactive current that the virtual synchronous machine needs to output when the grid connection point voltage drops.

[0026] Step S32: In the dq synchronous rotating coordinate system, combine the minimum reactive current expression with the reactive voltage droop control equation to derive the virtual resistance upper limit expression.

[0027] Step S33: Based on the converter's overcurrent capacity requirements, establish the maximum fault current constraint condition;

[0028] Step S34: Substitute the peak correlation calculation formula into the maximum fault current constraint condition to deduce the virtual resistance lower limit expression.

[0029] Step S35: Combine the upper limit expression and the lower limit expression of the virtual resistance to determine the safe control range of the virtual resistance.

[0030] Further, step S4 includes:

[0031] Step S41: Construct a multi-scenario weighted algorithm using the aforementioned safety control range as the constraint boundary;

[0032] Step S42: Under the multi-scenario weighted algorithm, the system inertia parameter, line impedance parameter and transient power parameter are obtained as input variables respectively;

[0033] Step S43: Based on the input variables, establish a calculation formula for the optimal initial value of the virtual resistance; the calculation formula includes a first weighting coefficient, a second weighting coefficient, and a third weighting coefficient, which correspond to the weights of system inertia, line impedance, and transient power, respectively;

[0034] Step S44: Adjust the values ​​of the first weight coefficient, the second weight coefficient, and the third weight coefficient according to the current operating scenario, and substitute them into the calculation formula for the optimal starting value of the virtual resistance to obtain the optimal starting value of the virtual resistance under the current operating scenario.

[0035] Furthermore, the construction of the three-layer game model in step S5 includes:

[0036] Step S51: Using the optimal initial value as the initial value, establish a state space model of the multi-machine system and define the state vector of a single machine;

[0037] Step S52: Based on the state space model, construct the cooperative game layer, the non-cooperative game layer, and the master-slave game layer respectively;

[0038] Step S53: Couple the cooperative game layer, the non-cooperative game layer, and the master-slave game layer to form a three-layer hybrid game model.

[0039] Furthermore, the coupled dynamic programming algorithm in step S5 includes:

[0040] Using the aforementioned three-layer hybrid game model as a framework, a local evaluation network and a global evaluation network are designed for each virtual synchronizer;

[0041] In the local evaluation network and the global evaluation network, transient energy term and impedance matching degree term are introduced respectively, and transient energy function and impedance matching degree function are defined;

[0042] An action network is designed in parallel with the local evaluation network and the global evaluation network. The action network outputs a virtual resistance adjustment amount based on the average of the local state and the neighboring states.

[0043] An experience replay buffer is designed for each virtual synchronizer to store the state-action-reward-next state samples collected during the fault period, and a priority experience replay mechanism is used to weight the samples according to the time difference error.

[0044] Using the weighted samples, the weights of the local evaluation network and the global evaluation network are updated using distributed temporal difference learning, and the weights of the action network are updated using the deterministic policy gradient method.

[0045] The virtual resistance adjustment value output by the updated action network is used as the optimal control value.

[0046] Further, step S6 includes:

[0047] The optimal control value is used as the updated virtual resistance value and substituted into the core control equation of the virtual resistance to obtain the updated reference voltage.

[0048] The updated reference voltage is injected into the voltage loop control equation and current loop control equation of the converter, and participates in the dual-loop closed-loop regulation calculation.

[0049] On the other hand, the present invention provides an adaptive virtual resistance fault overcurrent suppression system for grid-type converters. It includes:

[0050] The basic modeling module is used to build the basic control model of the grid-type converter;

[0051] The mechanism analysis module, which is connected to the basic modeling module, is used to establish an analytical model of transient current during power grid faults, analyze the influence mechanism of virtual resistance on fault current peak value, current change rate, DC decay rate and system damping, and output the analysis results of virtual resistance influence mechanism.

[0052] The safety boundary determination module is connected to the basic modeling module and the mechanism analysis module respectively. It is used to receive the influence mechanism analysis results, combine the low voltage ride-through requirements and the converter overcurrent capacity, derive the adaptive upper and lower limits of the virtual resistance, and determine the safe control range of the virtual resistance.

[0053] The optimal initial value calculation module is connected to the mechanism analysis module and the safety boundary determination module respectively. It is used to receive the safety control range and obtain the system inertia and line impedance. By constructing a multi-scenario weighted algorithm and comprehensively considering the system inertia, line impedance and transient power requirements, it calculates the optimal initial value of the virtual resistance under different scenarios.

[0054] The collaborative optimization solution module is connected to the basic modeling module, the mechanism analysis module, and the optimal starting value calculation module, respectively. It is used to receive the optimal starting value, obtain the system state parameters in the basic control model, obtain the coupling characteristics of the multi-machine parallel system, construct a three-layer game model starting from the optimal starting value, and couple the dynamic programming algorithm to form a distributed game-dynamic programming collaborative optimization framework, and solve the optimal control value of the virtual resistance.

[0055] The dual-loop embedded execution module is connected to the collaborative optimization solution module and the basic modeling module respectively. It receives the optimal control value and the virtual resistance core control equation, replaces the original virtual resistance setpoint with the optimal control value, substitutes it into the virtual resistance core control equation to obtain the updated reference voltage, and injects the reference voltage into the voltage loop control equation and current loop control equation of the converter to participate in the dual-loop closed-loop regulation calculation.

[0056] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention establishes a rotor motion equation including inertia time coefficient and damping coefficient by simulating the rotor motion characteristics of a synchronous generator, and establishes a reactive voltage droop control equation by adopting droop control. The virtual resistor is embedded in series between the VSG output side and the voltage and current dual-loop control structure. By introducing a virtual voltage drop, the output impedance of the converter is increased to an equivalent extent. This provides a unified and accurate basic control model for the subsequent adaptive optimization of the virtual resistor and the suppression of fault overcurrent.

[0057] Furthermore, by establishing an analytical model of transient current under symmetrical faults, this invention clarifies that increasing the virtual resistance can simultaneously reduce the peak fault current, decrease the rate of change of current, and accelerate the decay rate of the DC component, providing a quantitative theoretical basis for parameter tuning of the virtual resistance in fault overcurrent suppression. Simultaneously, by performing small-signal modeling on a multi-machine parallel system containing virtual resistance, this invention establishes linearized expressions for single-machine electromagnetic power, equivalent synchronization coefficients, undamped natural oscillation frequency, and equivalent damping ratio, revealing the damping enhancement mechanism by increasing the virtual resistance to reduce the undamped natural oscillation frequency and increase the equivalent damping ratio. This lays a theoretical foundation for the stability analysis and virtual resistance synergistic optimization of multi-machine parallel systems under fault conditions.

[0058] Furthermore, by combining the low-voltage ride-through standard and the converter overcurrent capacity constraint, this invention derives the upper and lower limits of the virtual resistance under symmetrical faults, thereby determining the full-condition safety control boundary of the virtual resistance. This satisfies the low-voltage ride-through requirement of the minimum reactive current output of the converter during grid faults, while ensuring that the peak fault current does not exceed a set multiple of the converter's rated current. This provides a clear safety constraint range for the adaptive adjustment of the virtual resistance, ensuring the safety feasibility and engineering practicality of the control strategy under all operating conditions.

[0059] Furthermore, this invention constructs a multi-scenario weighted algorithm, taking the safe control range as the constraint boundary, and comprehensively considers the system inertia damping characteristics, line impedance matching degree and transient power support capability requirements. It establishes a calculation formula for the optimal initial value of virtual resistance containing three adjustable weight coefficients, and flexibly adjusts the value of the weight coefficients according to different operating scenarios. This realizes the adaptive calculation of the optimal initial value of virtual resistance under various operating scenarios, providing an initial adjustment benchmark that is closer to the actual working conditions for subsequent multi-machine collaborative optimization.

[0060] Furthermore, this invention establishes a state-space model of a multi-machine system with optimal initial values ​​and constructs a three-layer hybrid game model coupled with cooperative game layer, non-cooperative game layer, and master-slave game layer. This model achieves multi-objective collaborative optimization of virtual resistance from three dimensions: overall system stability, single-machine overcurrent protection, and multi-machine adjustment consistency. It further couples an adaptive dynamic programming algorithm, designing local and global evaluation networks, action networks, and experience replay buffers, and introducing transient energy and impedance matching terms. Distributed time-series differential learning and deterministic policy gradient methods are used to update network weights, achieving online adaptive adjustment and distributed solution of virtual resistance. This overcomes the shortcomings of existing methods that lack multi-machine collaborative mechanisms and significantly improves the collaborative optimization capability and transient stability of multi-machine parallel systems under fault conditions.

[0061] Furthermore, this invention uses the optimal control value obtained by the adaptive dynamic programming algorithm as the updated virtual resistance value, substitutes it into the core control equation of the virtual resistance to obtain the updated reference voltage, and injects this reference voltage into the voltage loop control equation and the current loop control equation to participate in the dual-loop closed-loop regulation calculation. This achieves the accurate embedding of the optimized virtual resistance value into the voltage and current dual-loop control, completes the coordinated suppression of fault overcurrent and the stable support of system voltage, and forms a complete technical closed loop from parameter optimization to control execution.

[0062] Furthermore, this invention constructs a multi-machine electromagnetic transient model of a grid-connected converter based on the Simulink simulation platform, and conducts parallel simulation verification of two different operating conditions: power disturbance under normal operation and three-phase symmetrical short-circuit fault. Simulation results show that under normal operation, the strategy of this invention can ensure that the multi-machine system has excellent power decoupling characteristics and steady-state operation stability, and will not blindly adjust the virtual resistance due to power disturbance. Under the three-phase symmetrical short-circuit fault condition, the strategy of this invention can reduce the peak fault current by 57.3% compared with the traditional fixed virtual resistance strategy, while accelerating the attenuation rate of the DC component, and the virtual resistance adjustment curves of the multi-machines almost completely overlap, effectively verifying the effectiveness and superiority of the control strategy proposed in this invention in terms of fault overcurrent suppression and multi-machine coordination consistency. Attached Figure Description

[0063] Figure 1 This is a system topology diagram of the adaptive virtual resistance fault overcurrent suppression method for grid-type converters of the present invention;

[0064] Figure 2 This is a block diagram of the multi-machine adaptive virtual resistance control of the grid-type converter of the present invention;

[0065] Figure 3 Flowchart of a three-layer hybrid game-distributed adaptive dynamic programming multi-machine collaborative optimization algorithm;

[0066] Figure 4 The output current waveform of VSG1 under normal operating conditions when the load suddenly decreases;

[0067] Figure 5 The output current waveform of VSG2 under normal operating conditions when the load suddenly decreases;

[0068] Figure 6 This is a diagram showing the dynamic changes in the virtual resistance of the two VSG units when the load suddenly decreases under operating conditions.

[0069] Figure 7 The voltage waveform at the grid connection point is shown when the three-phase symmetrical voltage drops to 0.5 pu.

[0070] Figure 8 The waveform of the dual-machine output current under the traditional fixed virtual resistance strategy when the three-phase symmetrical voltage drops to 0.5pu;

[0071] Figure 9 The waveform of the dual-machine output current of the method of the present invention is shown when the three-phase symmetrical voltage drops to 0.5pu.

[0072] Figure 10 The diagram shows the dynamic changes of the virtual resistors of the two VSGs under the strategy of this invention when the three-phase symmetrical voltage drops to 0.5 pu. Detailed Implementation

[0073] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0074] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0075] The technical solution provided in this application includes the following steps:

[0076] Step S1: Based on the virtual synchronous machine control framework, establish the basic control model of the grid-type converter;

[0077] Step S2: Based on the basic control model, establish an analytical model of transient current during power grid faults, and analyze the influence mechanism of virtual resistance on fault current peak value, current change rate, DC decay rate and system damping.

[0078] Step S3: Based on the aforementioned influence mechanism, and in conjunction with the low voltage ride-through requirement and the converter's overcurrent capability, derive the adaptive upper and lower limits of the virtual resistance, and determine the safe control range of the virtual resistance.

[0079] Step S4: Within the safety control range, construct a multi-scenario weighted algorithm, comprehensively consider system inertia, line impedance and transient power requirements, and calculate the optimal initial value of virtual resistance under different scenarios.

[0080] Step S5: Starting from the optimal initial value, construct a three-layer game model and couple it with a dynamic programming algorithm to form a distributed game-dynamic programming collaborative optimization framework, and solve for the optimal control value of the output virtual resistor.

[0081] Step S6: Embed the optimal control value into the voltage and current dual-loop control of the converter, replace the original virtual resistance setpoint, and use the updated virtual resistance value to participate in the dual-loop closed-loop regulation calculation.

[0082] Specifically, the adaptive virtual resistance fault overcurrent suppression system for grid-type converters provided in this application includes:

[0083] The basic modeling module is used to build the basic control model of the grid-type converter;

[0084] The mechanism analysis module, which is connected to the basic modeling module, is used to establish an analytical model of transient current during power grid faults, analyze the influence mechanism of virtual resistance on fault current peak value, current change rate, DC decay rate and system damping, and output the analysis results of virtual resistance influence mechanism.

[0085] The safety boundary determination module is connected to the basic modeling module and the mechanism analysis module respectively. It is used to receive the influence mechanism analysis results, combine the low voltage ride-through requirements and the converter overcurrent capacity, derive the adaptive upper and lower limits of the virtual resistance, and determine the safe control range of the virtual resistance.

[0086] The optimal initial value calculation module is connected to the mechanism analysis module and the safety boundary determination module respectively. It is used to receive the safety control range and obtain the system inertia and line impedance. By constructing a multi-scenario weighted algorithm and comprehensively considering the system inertia, line impedance and transient power requirements, it calculates the optimal initial value of the virtual resistance under different scenarios.

[0087] The collaborative optimization solution module is connected to the basic modeling module, the mechanism analysis module, and the optimal starting value calculation module, respectively. It is used to receive the optimal starting value, obtain the system state parameters in the basic control model, obtain the coupling characteristics of the multi-machine parallel system, construct a three-layer game model starting from the optimal starting value, and couple the dynamic programming algorithm to form a distributed game-dynamic programming collaborative optimization framework, and solve the optimal control value of the virtual resistance.

[0088] The dual-loop embedded execution module is connected to the collaborative optimization solution module and the basic modeling module respectively. It receives the optimal control value and the virtual resistance core control equation, replaces the original virtual resistance setpoint with the optimal control value, substitutes it into the virtual resistance core control equation to obtain the updated reference voltage, and injects the reference voltage into the voltage loop control equation and current loop control equation of the converter to participate in the dual-loop closed-loop regulation calculation.

[0089] Specifically, the basic control model in step S1 includes three parallel sub-models:

[0090] Step S11: By simulating the rotor motion characteristics and excitation regulation mechanism of the synchronous generator, a rotor motion equation is established, which includes the inertia time coefficient and the damping coefficient.

[0091] Step S12: Reactive voltage control is achieved by using a droop method. A reactive voltage droop control equation is established, which includes a droop coefficient.

[0092] Step S13: Embed the virtual resistor in series between the output side of the virtual synchronous machine and the dual-loop control, and establish the core control equation of the virtual resistor. The virtual resistor value in the core control equation is the variable to be optimized.

[0093] In this embodiment of the invention, the specific process of establishing the basic control model of the grid-type converter VSG in S1 is as follows: The grid-type converter achieves autonomous control of active power-frequency and reactive power-voltage by simulating the rotor motion characteristics and excitation regulation mechanism of a synchronous generator. Its rotor motion equation is:

[0094] ;

[0095] In the formula, J is the inertial time coefficient, and the unit is... It characterizes the system's inertial response capability to active power imbalance; This is the damping coefficient, in units of... This is used to suppress system frequency oscillations; , These represent the mechanical and electromagnetic power of the analog synchronous machine, respectively, and the active power command given and the actual active power output of the converter; ω is the output angular frequency, in units of... .

[0096] Reactive power voltage control employs droop regulation, achieving autonomous reactive power allocation through the linear relationship between reactive power and output voltage amplitude. The expression is:

[0097] ;

[0098] In the formula, E is the output voltage amplitude of VSG, in V; The droop factor, measured in V / Var, determines the accuracy of reactive power distribution. The reactive power setpoint is given by the reactive power command. The reactive power output value is the actual reactive power output by the converter.

[0099] A virtual resistor is embedded in series between the VSG output side and the voltage-current dual-loop control structure. By introducing a virtual voltage drop at the control level, the output impedance of the converter is effectively increased. Its core control equation is:

[0100] ;

[0101] In the formula, and These are the virtual internal potentials of the VSG control output in the dq axis coordinate system, respectively. and These are the converter-side output currents in the dq-axis coordinate system, respectively. and These are the reference voltage values ​​for the injected voltage and current loops in the dq-axis coordinate system. Here, is the virtual resistance value, and is the control variable to be optimized.

[0102] This invention establishes a rotor motion equation that includes inertia time coefficient and damping coefficient by simulating the rotor motion characteristics of a synchronous generator. At the same time, it establishes a reactive voltage droop control equation by using droop control. The virtual resistor is embedded in series between the VSG output side and the voltage and current dual-loop control structure. By introducing a virtual voltage drop, the output impedance of the converter is effectively increased, providing a unified and accurate basic control model for subsequent adaptive optimization of the virtual resistor and fault overcurrent suppression.

[0103] Specifically, step S2 includes:

[0104] Step S21: Based on the basic control model, for a three-phase symmetrical short-circuit fault, establish a transient expression for the fault current according to the step response of the RL circuit.

[0105] Step S22: Based on the transient expression of the fault current and the virtual resistance value, establish a calculation formula for the correlation between the peak value of the fault current and the peak value of the virtual resistance, as well as a calculation formula for the correlation between the rate of change of the fault current and the rate of change of the virtual resistance in the initial stage of the fault.

[0106] Step S23: Based on the peak correlation calculation formula and the rate of change correlation calculation formula, it is determined that increasing the virtual resistance can simultaneously reduce the peak fault current and the rate of change of the initial fault current, and accelerate the attenuation speed of the DC component.

[0107] Specifically, step S2 further includes:

[0108] Step S24: Perform small-signal modeling on the multi-machine parallel system with virtual resistance and establish a linearized expression for the electromagnetic power of a single machine: The linearized expression for the electromagnetic power of a single machine includes the power angle synchronization coefficient, the voltage synchronization coefficient, and the influence coefficient of neighboring units;

[0109] Step S25: Based on the small signal modeling results, establish an equivalent synchronization coefficient expression for a single machine. The equivalent synchronization coefficient expression for a single machine is determined by the output voltage amplitude before the fault, the voltage amplitude at the grid connection point, the total output impedance amplitude, the power angle before the fault, and the total impedance angle.

[0110] Step S26: Based on the equivalent synchronization coefficient expression, establish the expression for the undamped natural oscillation frequency and the equivalent damping ratio of the single machine, and determine that increasing the virtual resistance can reduce the undamped natural oscillation frequency and increase the equivalent damping ratio, so as to characterize the influence mechanism of the virtual resistance on the system damping characteristics.

[0111] In this embodiment of the invention, the transient current analytical model under symmetrical faults is as follows: When a three-phase symmetrical short-circuit fault occurs in the power grid, according to the step response of the RL circuit in circuit theory, the transient expression of the fault current can be obtained as follows:

[0112] ;

[0113] In the formula, This refers to the steady-state current after the fault. For the angle of attack; This represents the phase of the steady-state current after the fault. The steady-state current amplitude before the fault, in amperes (A). The steady-state current phase before the fault is expressed in rad. The total resistance is denoted as , where Line resistance; For the total inductance, where For virtual inductance, For line inductance;

[0114] The peak fault current occurs when t≈π / ω, and its value is:

[0115] ;

[0116] The rate of change of current in the initial stage of the fault is:

[0117] ;

[0118] In the formula, The value of the internal potential of the VSG before the fault is expressed in V.

[0119] From the above formula, it can be seen that increasing the virtual resistance... It can simultaneously reduce the peak fault current and the rate of change of current, and accelerate the decay rate of the DC component;

[0120] The effect of virtual resistance on the system damping characteristics:

[0121] By performing small-signal modeling on a multi-VSG parallel system with virtual resistance, the linearized expression for the electromagnetic power of the i-th VSG can be obtained as follows:

[0122] ;

[0123] In the formula, Let be the power angle synchronization coefficient of the i-th VSG; Let be the voltage synchronization coefficient of the i-th VSG; Let be the influence coefficient of the power angle change of the j-th VSG on the output power of the ith VSG;

[0124] The equivalent synchronization coefficient of the i-th VSG is:

[0125] ;

[0126] In the formula, The value of the output voltage before the i-th VSG fails is expressed in V. This refers to the voltage amplitude at the grid connection point, in volts (V). The total output impedance amplitude of the i-th VSG is expressed in Ω. The power angle before the failure of the i-th VSG is expressed in rad. This is the total impedance angle, in rad.

[0127] The undamped natural oscillation frequency and equivalent damping ratio of the i-th VSG are respectively:

[0128] ;

[0129] In the formula, Let be the inertial time coefficient of the i-th VSG; Let be the damping coefficient of the i-th VSG;

[0130] Increase virtual resistance It will reduce the equivalent synchronization coefficient This reduces the frequency of undamped natural oscillation. At the same time, increase the equivalent damping ratio This enhances the system's damping effect.

[0131] This invention establishes an analytical model of transient current under symmetrical faults, clarifying that increasing the virtual resistance can simultaneously reduce the peak fault current, decrease the rate of change of current, and accelerate the decay rate of the DC component, providing a quantitative theoretical basis for parameter tuning of the virtual resistance in fault overcurrent suppression. Simultaneously, by performing small-signal modeling on a multi-machine parallel system containing virtual resistance, it establishes linearized expressions for single-machine electromagnetic power, equivalent synchronization coefficients, undamped natural oscillation frequency, and equivalent damping ratio, revealing the damping enhancement mechanism that increasing the virtual resistance can reduce the undamped natural oscillation frequency and increase the equivalent damping ratio. This lays a theoretical foundation for the stability analysis and virtual resistance synergistic optimization of multi-machine parallel systems under fault conditions.

[0132] Specifically, step S3 includes:

[0133] Step S31: Based on the low voltage ride-through requirements, establish the expression for the minimum reactive current that the virtual synchronous machine needs to output when the grid connection point voltage drops.

[0134] Step S32: In the dq synchronous rotating coordinate system, combine the minimum reactive current expression with the reactive voltage droop control equation to derive the virtual resistance upper limit expression.

[0135] Step S33: Based on the converter's overcurrent capacity requirements, establish the maximum fault current constraint condition;

[0136] Step S34: Substitute the peak correlation calculation formula into the maximum fault current constraint condition to deduce the virtual resistance lower limit expression.

[0137] Step S35: Combine the upper limit expression and the lower limit expression of the virtual resistance to determine the safe control range of the virtual resistance.

[0138] In this embodiment of the invention, the upper limit of virtual resistance under symmetrical faults Derivation: According to the power system low voltage ride-through standard, when the grid voltage drops, the minimum reactive current that the VSG needs to output is:

[0139] ;

[0140] In the formula, This is the minimum reactive current reference value that the i-th VSG needs to output, in A; Let be the per-unit value of the voltage drop at the grid connection point of the i-th VSG; The rated current of the i-th VSG is expressed in amperes (A).

[0141] In the dq synchronous rotating coordinate system, the reactive power output of VSG is:

[0142] ;

[0143] In the formula, , These are the d-axis and q-axis components of the VSG output electromotive force, respectively, in V. , These are the d-axis and q-axis components of the VSG output current, respectively, in amperes (A).

[0144] By combining the reactive voltage control equations and the reactive power calculation formula, we can obtain:

[0145] ;

[0146] In the formula, Let be the reactive power droop coefficient of the i-th VSG; The reactive power setting value for the i-th VSG, in Var;

[0147] The upper limit of virtual resistance under symmetrical faults is obtained as follows:

[0148] ;

[0149] Virtual resistance lower limit Derivation:

[0150] According to the converter's overcurrent capacity requirements, the maximum fault current must not exceed k times the converter's rated current (usually k = 1.5~2), that is:

[0151] ;

[0152] In the formula, The output electromotive force amplitude of the i-th VSG before the failure is expressed in V; The voltage amplitude at the grid connection point after the fault, in volts (V). The equivalent total resistance of the system's external transmission line for the i-th VSG is expressed in Ω. The equivalent total inductance of the system's external transmission line for the i-th VSG is expressed in H.

[0153] Solving for the lower limit of the virtual resistance:

[0154] ;

[0155] Therefore, the full-condition safety control boundary of the virtual resistor is:

[0156] .

[0157] This invention, by combining the low-voltage ride-through standard and the converter overcurrent capacity constraint, derives the upper and lower limits of the virtual resistance under symmetrical faults, thereby determining the full-condition safety control boundary of the virtual resistance. This satisfies the low-voltage ride-through requirement of the minimum reactive current output of the converter during grid faults, while ensuring that the peak fault current does not exceed a set multiple of the converter's rated current. This provides a clear safety constraint range for the adaptive adjustment of the virtual resistance, ensuring the safety feasibility and engineering practicality of the control strategy under all operating conditions.

[0158] Specifically, step S4 includes:

[0159] Step S41: Construct a multi-scenario weighted algorithm using the aforementioned safety control range as the constraint boundary;

[0160] Step S42: Under the multi-scenario weighted algorithm, the system inertia parameter, line impedance parameter and transient power parameter are obtained as input variables respectively;

[0161] Step S43: Based on the input variables, establish a calculation formula for the optimal initial value of the virtual resistance; the calculation formula includes a first weighting coefficient, a second weighting coefficient, and a third weighting coefficient, which correspond to the weights of system inertia, line impedance, and transient power, respectively;

[0162] Step S44: Adjust the values ​​of the first weight coefficient, the second weight coefficient, and the third weight coefficient according to the current operating scenario, and substitute them into the calculation formula for the optimal starting value of the virtual resistance to obtain the optimal starting value of the virtual resistance under the current operating scenario.

[0163] In this embodiment of the invention, considering the system's inertia damping characteristics, line impedance matching degree, and transient power support capability requirements, a multi-scenario weighted algorithm is constructed, and the calculation formula is as follows:

[0164] ;

[0165] In the formula, Let be the initial virtual resistance value of the i-th VSG, in Ω. The rated voltage of the grid for the i-th VSG is given in V. Let f be the equivalent total reactance of the system's external transmission line for the i-th VSG, in Ω, where f is the standard frequency of the power grid, in Hz; The system rated power of the i-th VSG is expressed in VA. The maximum transient power fluctuation of the i-th VSG is expressed in W. , , This is a weighting coefficient, which is adjusted according to different operating scenarios.

[0166] This invention constructs a multi-scenario weighted algorithm, using the safe control range as the constraint boundary, and comprehensively considers the system's inertia damping characteristics, line impedance matching degree, and transient power support capability requirements. It establishes a calculation formula for the optimal initial value of virtual resistance, which includes three adjustable weight coefficients. The value of the weight coefficients can be flexibly adjusted according to different operating scenarios, realizing the adaptive calculation of the optimal initial value of virtual resistance under various operating scenarios. This provides an initial adjustment benchmark that is closer to the actual working conditions for subsequent multi-machine collaborative optimization.

[0167] Specifically, the construction of the three-layer game model in step S5 includes:

[0168] Step S51: Using the optimal initial value as the initial value, establish a state space model of the multi-machine system and define the state vector of a single machine;

[0169] Step S52: Based on the state space model, construct the cooperative game layer, the non-cooperative game layer, and the master-slave game layer respectively;

[0170] Step S53: Couple the cooperative game layer, the non-cooperative game layer, and the master-slave game layer to form a three-layer hybrid game model.

[0171] Specifically, the coupled dynamic programming algorithm in step S5 includes:

[0172] Using the aforementioned three-layer hybrid game model as a framework, a local evaluation network and a global evaluation network are designed for each virtual synchronizer;

[0173] In the local evaluation network and the global evaluation network, transient energy term and impedance matching degree term are introduced respectively, and transient energy function and impedance matching degree function are defined;

[0174] An action network is designed in parallel with the local evaluation network and the global evaluation network. The action network outputs a virtual resistance adjustment amount based on the average of the local state and the neighboring states.

[0175] An experience replay buffer is designed for each virtual synchronizer to store the state-action-reward-next state samples collected during the fault period, and a priority experience replay mechanism is used to weight the samples according to the time difference error.

[0176] Using the weighted samples, the weights of the local evaluation network and the global evaluation network are updated using distributed temporal difference learning, and the weights of the action network are updated using the deterministic policy gradient method.

[0177] The virtual resistance adjustment value output by the updated action network is used as the optimal control value.

[0178] In this embodiment of the invention, a state-space model of a multi-machine system is established: the state vector of the i-th VSG is defined as follows:

[0179] ;

[0180] In the formula, The power angle of the i-th VSG is expressed in rad. Let be the angular frequency of the i-th VSG, in rad / s; , These are the negative-sequence d-axis and q-axis components of the output current of the i-th VSG, respectively, in A;

[0181] System total state vector:

[0182] ;

[0183] Control input vector:

[0184] ;

[0185] The nonlinear dynamic equation of the i-th VSG is:

[0186] ;

[0187] In the formula, It is a local nonlinear dynamic function; Let be the coupling function with neighbor j; Let i be the set of electrically connected neighbors of the i-th VSG; To control the input coefficients; The noise is the process noise, which follows a mean of 0 and a variance of . Gaussian distribution;

[0188] Construct a three-layer hybrid game model:

[0189] With the goal of achieving overall system stability, the objective function is:

[0190] ;

[0191] In the formula, This is the upper limit of the time for integrating the cost function, in seconds. , , , Let be the weighting coefficient, satisfying , >0, >0, >0; Let be the output current amplitude of the i-th VSG, in A; This is a voltage reference value, in volts (V). The average value of all VSG control inputs; The negative sequence component amplitude of the output current of the i-th VSG is given in A.

[0192] Constraints:

[0193] ;

[0194] In the formula, The maximum rate of change of the virtual resistance of the i-th VSG is expressed in Ω / s. The maximum permissible frequency deviation, in rad / s; The maximum permissible voltage deviation, in volts (V).

[0195] (2) Non-cooperative game layer (individual optimization):

[0196] With single-machine overcurrent protection as the core, the cost function of the i-th VSG is:

[0197] ;

[0198] In the formula, , , , Let be the weighting coefficient, satisfying , , , ; The terminal cost function, used to ensure the stability of the system terminal state, is defined as follows:

[0199] ;

[0200] In the formula, , This refers to the terminal weighting coefficient;

[0201] (3) Master-follower game layer (leader-follower):

[0202] Selecting the high-voltage grid-side VSG as the leader, a virtual resistance regulation reference trajectory is established. Other VSGs, acting as followers, track the leader's adjustment strategy to ensure consistency in multi-machine adjustment; the leader's objective function is:

[0203] ;

[0204] In the formula, the subscript l represents the leader, VSG;

[0205] The objective function for followers is:

[0206] ;

[0207] In the formula, To track the weighting coefficients, satisfy ;

[0208] Design a distributed adaptive dynamic programming algorithm:

[0209] (1) Critic Neural Network Design:

[0210] Design a local Critic network for each VSG and global Critic network The local policy and overall system performance were evaluated separately; the network adopted a deep residual network structure.

[0211] ;

[0212] In the formula, , , Here is the weight matrix of the Critic neural network; , , σ is the bias term; σ is the ReLU activation function; ϕ is the residual feature extraction function; n is the input layer dimension; , For the hidden layer dimension;

[0213] Introducing transient energy and impedance matching terms into the Critic network enables the algorithm to automatically learn transient stability constraints and impedance matching constraints:

[0214] ;

[0215] In the formula, Let be the transient energy function. Impedance matching degree; , These are the weighting coefficients;

[0216] The transient energy function is defined as:

[0217] ;

[0218] Impedance matching degree is defined as:

[0219] ;

[0220] (2) Actor Neural Network Design:

[0221] The Actor network outputs a virtual resistance adjustment based on the average of the local and neighbor states, expressed as:

[0222] ;

[0223] In the formula, The mean of the neighbor states; The weight matrix of the Actor neural network For bias terms; For the hidden layer dimension;

[0224] (3) Experience playback buffer design:

[0225] Design an experience replay buffer for each VSG. Used to store state-action-reward-next state samples collected during a failure:

[0226] ;

[0227] In the formula, M is the buffer capacity;

[0228] A priority experience replay mechanism is adopted to weight the samples according to the magnitude of the TD error, thereby improving the convergence speed of the algorithm;

[0229] (4) Distributed solution and network update:

[0230] Local Bellman equation:

[0231] ;

[0232] In the formula, R(k) represents the global instant reward;

[0233] Distributed temporal difference (TD) learning is used to update the Critic network weights. The local Critic update formula is as follows:

[0234] ;

[0235] In the formula, This is a local TD error; The learning rate for the Critic network;

[0236] Global Critic update formula:

[0237] ;

[0238] The Actor network weights are updated using a deterministic policy gradient method:

[0239] ;

[0240] In the formula, The action value function;

[0241] Updated formula:

[0242] ;

[0243] In the formula, The learning rate of the Actor network;

[0244] Output the final virtual resistance control value and implement fault classification control:

[0245] The final virtual resistance control strategy is

[0246] ;

[0247] In the formula, β∈[0.6,0.8] is the local adjustment weight; The local adjustment value output by the Actor network; The reference adjustment value for the leader's VSG output;

[0248] Fault recovery strategy:

[0249] When UPCC ≥ 0.9UN and remains so for more than 0.5s, the virtual resistance gradually returns to its initial value.

[0250] ;

[0251] In the formula, α∈(0,0.1) is the regression coefficient.

[0252] This invention establishes a state-space model of a multi-machine system with optimal initial values ​​and constructs a three-layer hybrid game model consisting of a cooperative game layer, a non-cooperative game layer, and a master-slave game layer. This model achieves multi-objective collaborative optimization of virtual resistance from three dimensions: overall system stability, single-machine overcurrent protection, and multi-machine adjustment consistency. Furthermore, it couples an adaptive dynamic programming algorithm, designing local and global evaluation networks, action networks, and an experience replay buffer. It also introduces transient energy and impedance matching terms and uses distributed time-series differential learning and deterministic policy gradient methods to update network weights. This enables online adaptive adjustment and distributed solution of virtual resistance, overcoming the shortcomings of existing methods that lack multi-machine collaborative mechanisms and significantly improving the collaborative optimization capability and transient stability of multi-machine parallel systems under fault conditions.

[0253] Specifically, step S6 includes:

[0254] The optimal control value is used as the updated virtual resistance value and substituted into the core control equation of the virtual resistance to obtain the updated reference voltage.

[0255] The updated reference voltage is injected into the voltage loop control equation and current loop control equation of the converter, and participates in the dual-loop closed-loop regulation calculation.

[0256] In this embodiment of the invention, the voltage loop control equation of the grid converter is:

[0257] ;

[0258] In the formula, , The dq-axis component of the grid connection point voltage is expressed in V. This is a filter capacitor, measured in F (amperes). , These are the parameters for the voltage loop PI controller;

[0259] The current loop control equation for a grid-type converter is:

[0260] ;

[0261] In the formula, This is a filter inductance, measured in watts (H). , These are the parameters for the current loop PI controller;

[0262] The optimized virtual resistance Substituting into the core control equation of the virtual resistor, we obtain the updated reference voltage:

[0263] ;

[0264] This enables dual-loop control of voltage and current, achieving coordinated suppression of fault overcurrent and stable support of system voltage.

[0265] This invention uses the optimal control value obtained by the adaptive dynamic programming algorithm as the updated virtual resistance value, substitutes it into the core control equation of the virtual resistance to obtain the updated reference voltage, and injects this reference voltage into the voltage loop control equation and the current loop control equation to participate in the dual-loop closed-loop regulation calculation. This achieves the accurate embedding of the optimized virtual resistance value into the voltage and current dual-loop control, completes the coordinated suppression of fault overcurrent and the stable support of system voltage, and forms a complete technical closed loop from parameter optimization to control execution.

[0266] Specifically, it also includes simulation verification steps:

[0267] A multi-machine electromagnetic transient model of a grid converter was built based on a simulation platform.

[0268] Based on the control methods constructed in steps S1 to S6, multi-machine parallel simulations were carried out under different operating conditions;

[0269] The effectiveness of the control method was verified through simulation results.

[0270] In this embodiment of the invention, a multi-machine electromagnetic transient model of a grid converter is built based on the Simulink simulation platform, and parallel experiments of two machines under different operating conditions are carried out to verify the effectiveness and superiority of the proposed control strategy.

[0271] Figure 1 This paper demonstrates the electrical topology of the grid-connected multi-unit parallel grid-connected experimental system based on this invention. Multiple grid-connected converters with identical parameters are connected to a common point of connection (PCC) via their respective LC filter circuits, and then connected to an infinite power grid. Both converters employ a virtual synchronous generator control architecture, with virtual resistors embedded in series between the VSG output side and the voltage-current dual-loop control structure. This topology strictly corresponds to the electromagnetic transient simulation model construction scheme. By standardizing the converter capacity, control parameters, and initial virtual resistance values, the interference of parameter differences on the experimental results is eliminated, ensuring the fairness and effectiveness of the experimental comparison.

[0272] Figure 2 This is a block diagram illustrating the control principle of the method of the present invention. The block diagram shows the optimal virtual resistance obtained through online adaptive adjustment using an adaptive dynamic programming algorithm, which is then embedded into the voltage and current dual-loop control loop of the grid-type converter to achieve coordinated suppression of fault overcurrent and stable support of system voltage.

[0273] Figure 3 The execution flow of the core algorithm of this invention is demonstrated. The algorithm first initializes the system and loads the offline-derived virtual resistance safety control boundary under all operating conditions. Upon detecting a grid fault trigger condition, it sequentially enters a three-layer hybrid game decision-making process: the cooperative game layer calculates the global optimal solution with the goal of overall system stability; the non-cooperative game layer calculates the individual optimal solution with the goal of single-machine overcurrent protection; and the master-slave game layer selects the strong grid-side VSG as the leader to output the reference control trajectory. Subsequently, online parameter optimization is performed using a distributed ADP algorithm, the Critic network is used to evaluate strategy performance, the Actor network outputs the virtual resistance adjustment amount, and a priority experience replay mechanism accelerates algorithm convergence. Finally, a virtual resistance control value that satisfies safety constraints is output.

[0274] The simulation model of this invention is built on the Matlab / Simulink simulation platform, and... Figure 1 A simulation model was built to establish the theoretical framework. The rated DC voltage was set to 600V. After passing through a grid-connected converter and filter circuit, the system was connected to the grid at the grid connection point. The grid voltage was 690V, forming a grid-connected system. The effectiveness and consistency of the proposed method in a multi-machine parallel system were tested. The simulation experiment used two grid-connected converters, both controlled by VSG. Two simulation conditions were used: one was power disturbance verification under normal operating conditions: at t=5s, a sudden load reduction was simulated, and no voltage drop occurred; the second was a symmetrical three-phase voltage drop to 0.5pu.: at t=4s, a symmetrical three-phase voltage drop was applied at the PCC, and the voltage amplitude suddenly dropped to 0.5pu., simulating a typical symmetrical short-circuit fault condition.

[0275] Figure 4 and Figure 5 The output current waveforms of VSG1 and VSG2 at t=5s under normal operating conditions (no grid voltage drop) are shown. The waveforms show that before the load drop at t=5s, the output current of VSG1 maintains a stable sinusoidal waveform without any distortion. At the instant the load command changes at t=5s, the output current responds quickly and tracks the new power command. The entire adjustment process is smooth and shock-free, with no significant overshoot or oscillation, and the adjustment time is less than 0.2s. This waveform verifies that the proposed strategy has excellent dynamic power regulation performance at the single-unit level. The current waveform of VSG2 is completely consistent with that of VSG1, also exhibiting fast and smooth dynamic response characteristics during the load drop at t=5s. More importantly, the power disturbance of VSG1 has almost no impact on the output current of VSG2, and the output currents of the two converters are independent. This result indicates that the proposed strategy can ensure excellent power decoupling characteristics in a multi-unit system, and changes in the power of a single unit will not interfere with the normal operation of other units, ensuring the stability and independence of the system's steady-state operation.

[0276] like Figure 6 As shown in the figure, this diagram illustrates the changes in the virtual resistance values ​​of VSG1 and VSG2 over time during the entire load reduction disturbance. It is clearly visible from the figure that throughout the entire experiment from t=0 to t=20s, the virtual resistance values ​​of the two converters remained essentially constant near their initial setpoints, with the maximum fluctuation not exceeding 5% of the initial value, and the virtual resistance change curves of the two converters completely overlapped. This result strongly demonstrates that the strategy proposed in this invention has strong operating condition specificity. Under normal operating conditions, it will not blindly adjust the virtual resistance due to power disturbances, avoiding additional power coupling or current fluctuations, and ensuring good compatibility with traditional VSG control strategies.

[0277] Figure 7 The voltage excitation waveform of a three-phase symmetrical short-circuit fault experiment is shown. The horizontal axis represents time (in seconds), and the vertical axis represents the voltage per unit value. At t=4s, a three-phase symmetrical voltage drop is applied at the common point of connection (PCC), with the voltage amplitude dropping sharply from the rated value of 1.0 pu to 0.5 pu. The fault duration is 2s, and the voltage recovers to the rated value at t=6s. This voltage waveform rigorously simulates a typical three-phase symmetrical short-circuit fault condition in a power system, providing a unified fault excitation condition for subsequent comparative experiments of two control strategies.

[0278] like Figure 8The output current waveform of a dual-VSG parallel system under a three-phase symmetrical voltage dip fault, using a traditional fixed virtual resistance control strategy, is shown. The waveform clearly shows that at the voltage dip of t=4s, the system experiences an extremely severe transient current surge, with a peak fault current of 5.83 pu, far exceeding the 1.5 to 2 times rated current limit that the converter's power devices can typically withstand. Such a high current surge can easily cause the converter's overcurrent protection to trip, and may even cause permanent damage to the power devices. Furthermore, the fault current contains a large DC component with a slow decay rate, further exacerbating the overcurrent risk.

[0279] like Figure 9 The output current waveforms under identical experimental conditions using the multi-machine collaborative adaptive virtual resistance control strategy proposed in this invention are shown. The comparison reveals that the strategy of this invention has an extremely significant suppression effect on fault transient currents: at the voltage drop instant of t=4s, the peak fault current is only 2.49 pu, a reduction of 57.3% compared to the traditional fixed virtual resistance strategy. Simultaneously, the DC component of the fault current decays at a significantly faster rate, and the current waveform is smooth and distortion-free throughout the fault process. These experimental results directly verify that the strategy proposed in this invention can effectively suppress transient overcurrents under three-phase symmetrical faults, ensuring the safe operation of the converter equipment.

[0280] Figure 10 The diagram illustrates the dynamic adjustment curves of the virtual resistance values ​​of VSG1 and VSG2 over time during a fault, demonstrating the strategy of this invention. As can be seen from the graph: at the moment the fault occurs at t=4s, the virtual resistance values ​​of the two converters increase synchronously and smoothly to quickly suppress the fault current surge; during the fault duration from t=4s to t=6s, the virtual resistance values ​​remain near the optimal adjustment value, ensuring overcurrent suppression; after voltage recovery at t=6s, the virtual resistance values ​​gradually and smoothly return to their initial set values. Crucially, the virtual resistance adjustment curves of VSG1 and VSG2 almost completely overlap, demonstrating the superior performance of the multi-machine collaborative optimization framework proposed in this invention and solving the problem of inconsistent multi-machine adjustment in traditional methods.

[0281] This invention constructs a multi-machine electromagnetic transient model of a grid-connected converter based on the Simulink simulation platform. Simulations are conducted to verify the effectiveness of the dual-machine parallel operation under two different conditions: power disturbance during normal operation and three-phase symmetrical short-circuit fault. Simulation results show that under normal operation, the proposed strategy ensures excellent power decoupling characteristics and steady-state stability of the multi-machine system, without blindly adjusting the virtual resistance due to power disturbances. Under three-phase symmetrical short-circuit fault conditions, compared with the traditional fixed virtual resistance strategy, the proposed strategy reduces the peak fault current by 57.3% and accelerates the DC component decay rate. Furthermore, the virtual resistance adjustment curves of the multi-machine systems almost completely overlap, effectively verifying the effectiveness and superiority of the proposed control strategy in fault overcurrent suppression and multi-machine coordination consistency.

[0282] The above embodiments are merely illustrative examples and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this application.

Claims

1. A method for suppressing overcurrent in a grid-type converter under adaptive virtual resistance faults, characterized in that, include: Step S1: Based on the virtual synchronous machine control framework, establish the basic control model of the grid-type converter; Step S2: Based on the basic control model, establish a transient current analytical model during power grid faults, and analyze the influence mechanism of virtual resistance on fault current peak value, current change rate, DC attenuation rate and system damping based on the transient current analytical model. Step S3: Based on the aforementioned influence mechanism, and in conjunction with the low voltage ride-through requirement and the converter's overcurrent capability, derive the adaptive upper and lower limits of the virtual resistance, and determine the safe control range of the virtual resistance. Step S4: Within the safety control range, construct a multi-scenario weighted algorithm, comprehensively consider system inertia, line impedance and transient power requirements, and calculate the optimal initial value of virtual resistance under different scenarios through the multi-scenario weighted algorithm; Step S5: Starting from the optimal initial value, construct a three-layer game model and couple it with a dynamic programming algorithm to form a distributed game-dynamic programming collaborative optimization framework, and solve for the optimal control value of the output virtual resistor. Step S6: Embed the optimal control value into the voltage and current dual-loop control of the converter, replace the original virtual resistance setpoint, and use the updated virtual resistance value to participate in the dual-loop closed-loop regulation calculation.

2. The method according to claim 1, characterized in that, The basic control model in step S1 includes three parallel sub-models: Step S11: By simulating the rotor motion characteristics and excitation regulation mechanism of the synchronous generator, a rotor motion equation is established, which includes the inertia time coefficient and the damping coefficient. Step S12: Reactive voltage control is achieved by using a droop method. A reactive voltage droop control equation is established, which includes a droop coefficient. Step S13: Embed the virtual resistor in series between the output side of the virtual synchronous machine and the dual-loop control, and establish the core control equation of the virtual resistor. The virtual resistor value in the core control equation is the variable to be optimized.

3. The method according to claim 1, characterized in that, Step S2 includes: Step S21: Based on the basic control model, for a three-phase symmetrical short-circuit fault, establish a transient expression for the fault current according to the step response of the RL circuit. Step S22: Based on the transient expression of the fault current and the virtual resistance value, establish a calculation formula for the correlation between the peak value of the fault current and the peak value of the virtual resistance, as well as a calculation formula for the correlation between the rate of change of the fault current and the rate of change of the virtual resistance in the initial stage of the fault. Step S23: Based on the peak correlation calculation formula and the rate of change correlation calculation formula, it is determined that increasing the virtual resistance can simultaneously reduce the peak fault current and the rate of change of the initial fault current, and accelerate the attenuation speed of the DC component.

4. The method according to claim 1, characterized in that, Step S2 further includes: Step S24: Perform small-signal modeling on the multi-unit parallel system with virtual resistance and establish a linearized expression for the electromagnetic power of a single unit; the linearized expression for the electromagnetic power of a single unit includes the power angle synchronization coefficient, the voltage synchronization coefficient, and the influence coefficient of neighboring units; Step S25: Based on the small signal modeling results, establish an equivalent synchronization coefficient expression for a single machine. The equivalent synchronization coefficient expression for a single machine is determined by the output voltage amplitude before the fault, the voltage amplitude at the grid connection point, the total output impedance amplitude, the power angle before the fault, and the total impedance angle. Step S26: Based on the equivalent synchronization coefficient expression, establish the expression for the undamped natural oscillation frequency and the equivalent damping ratio of the single machine, and determine that increasing the virtual resistance can reduce the undamped natural oscillation frequency and increase the equivalent damping ratio, so as to characterize the influence mechanism of the virtual resistance on the system damping characteristics.

5. The method according to claim 1, characterized in that, Step S3 includes: Step S31: Based on the low voltage ride-through requirements, establish the expression for the minimum reactive current that the virtual synchronous machine needs to output when the grid connection point voltage drops. Step S32: In the dq synchronous rotating coordinate system, combine the minimum reactive current expression with the reactive voltage droop control equation to derive the virtual resistance upper limit expression. Step S33: Based on the converter's overcurrent capacity requirements, establish the maximum fault current constraint condition; Step S34: Substitute the peak correlation calculation formula into the maximum fault current constraint condition to deduce the virtual resistance lower limit expression. Step S35: Combine the upper limit expression and the lower limit expression of the virtual resistance to determine the safe control range of the virtual resistance.

6. The method according to claim 1, characterized in that, Step S4 includes: Step S41: Construct a multi-scenario weighted algorithm using the aforementioned safety control range as the constraint boundary; Step S42: Under the multi-scenario weighted algorithm, the system inertia parameter, line impedance parameter and transient power parameter are obtained as input variables respectively; Step S43: Based on the input variables, establish a calculation formula for the optimal initial value of the virtual resistance; the calculation formula includes a first weighting coefficient, a second weighting coefficient, and a third weighting coefficient, which correspond to the weights of system inertia, line impedance, and transient power, respectively; Step S44: Adjust the values ​​of the first weight coefficient, the second weight coefficient, and the third weight coefficient according to the current operating scenario, and substitute them into the calculation formula for the optimal starting value of the virtual resistance to obtain the optimal starting value of the virtual resistance under the current operating scenario.

7. The method according to claim 1, characterized in that, The construction of the three-layer game model in step S5 includes: Step S51: Using the optimal initial value as the initial value, establish a state space model of the multi-machine system and define the state vector of a single machine; Step S52: Based on the state space model, construct the cooperative game layer, the non-cooperative game layer, and the master-slave game layer respectively; Step S53: Couple the cooperative game layer, the non-cooperative game layer, and the master-slave game layer to form a three-layer hybrid game model.

8. The method according to claim 1, characterized in that, The coupled dynamic programming algorithm in step S5 includes: Using the aforementioned three-layer hybrid game model as a framework, a local evaluation network and a global evaluation network are designed for each virtual synchronizer; In the local evaluation network and the global evaluation network, transient energy term and impedance matching degree term are introduced respectively, and transient energy function and impedance matching degree function are defined; An action network is designed in parallel with the local evaluation network and the global evaluation network. The action network outputs a virtual resistance adjustment amount based on the average of the local state and the neighboring states. An experience replay buffer is designed for each virtual synchronizer to store the state-action-reward-next state samples collected during the fault period, and a priority experience replay mechanism is used to weight the samples according to the time difference error. Using the weighted samples, the weights of the local evaluation network and the global evaluation network are updated using distributed temporal difference learning, and the weights of the action network are updated using the deterministic policy gradient method. The virtual resistance adjustment value output by the updated action network is used as the optimal control value.

9. The method according to claim 1, characterized in that, Step S6 includes: The optimal control value is used as the updated virtual resistance value and substituted into the core control equation of the virtual resistance to obtain the updated reference voltage. The updated reference voltage is injected into the voltage loop control equation and current loop control equation of the converter, and participates in the dual-loop closed-loop regulation calculation.

10. A system applied to the adaptive virtual resistance fault overcurrent suppression method for grid-type converters according to any one of claims 1-9, characterized in that, include: The basic modeling module is used to build the basic control model of the grid-type converter; The mechanism analysis module, which is connected to the basic modeling module, is used to establish an analytical model of transient current during power grid faults, analyze the influence mechanism of virtual resistance on fault current peak value, current change rate, DC decay rate and system damping, and output the analysis results of virtual resistance influence mechanism. The safety boundary determination module is connected to the basic modeling module and the mechanism analysis module respectively. It is used to receive the influence mechanism analysis results, combine the low voltage ride-through requirements and the converter overcurrent capacity, derive the adaptive upper and lower limits of the virtual resistance, and determine the safe control range of the virtual resistance. The optimal initial value calculation module is connected to the mechanism analysis module and the safety boundary determination module respectively. It is used to receive the safety control range and obtain the system inertia and line impedance. By constructing a multi-scenario weighted algorithm and comprehensively considering the system inertia, line impedance and transient power requirements, it calculates the optimal initial value of the virtual resistance under different scenarios. The collaborative optimization solution module is connected to the basic modeling module, the mechanism analysis module, and the optimal starting value calculation module, respectively. It is used to receive the optimal starting value, obtain the system state parameters in the basic control model, obtain the coupling characteristics of the multi-machine parallel system, construct a three-layer game model starting from the optimal starting value, and couple the dynamic programming algorithm to form a distributed game-dynamic programming collaborative optimization framework, and solve the optimal control value of the virtual resistance. The dual-loop embedded execution module is connected to the collaborative optimization solution module and the basic modeling module respectively. It receives the optimal control value and the virtual resistance core control equation, replaces the original virtual resistance setpoint with the optimal control value, substitutes it into the virtual resistance core control equation to obtain the updated reference voltage, and injects the reference voltage into the voltage loop control equation and current loop control equation of the converter to participate in the dual-loop closed-loop regulation calculation.

Images