A method and system for pre-synchronization and grid-connection of a grid-forming converter

Abstract

The application relates to a network-constructed converter pre-synchronization grid-connection method and system, and belongs to the technical field of new energy grid-connection control and smart grid. Real phase difference is directly restored through Clarke transformation and four-quadrant inverse tangent operation. In order to suppress pre-synchronization transient power impact, a transient power correction function containing nonlinear adaptive gain is designed to realize adaptive adjustment of automatic gain compression under large deviation and recovery of convergence speed under small deviation. A multi-element evaluation function of instantaneous phase difference, frequency difference and voltage amplitude difference is constructed, accurate determination of grid-connection closing conditions is realized through normalization and weighted synthesis of the comprehensive criterion, and the gradient descent algorithm is used for adaptive optimization of control parameters, so that the problems of impact current and power oscillation caused by phase period step and nonlinear distortion of error signal in the off-grid to grid-connection switching process of the light-storage network-constructed converter are effectively solved, and continuous linear pre-synchronization control in the whole phase interval and smooth grid-connection without impact are realized.

Description

Technical Field

[0001] This invention belongs to the field of new energy control technology, and more specifically relates to a method and system for pre-synchronization grid connection of grid-connected converters. Background Technology

[0002] With the rapid development of distributed photovoltaic (PV) power generation and microgrid technologies, grid-connected converters, as the key interface connecting distributed power sources and the grid, directly impact system stability and power quality through their control performance. Virtual synchronous machine (VSM) technology, by simulating the rotor motion equations and droop characteristics of a synchronous generator, provides inertia support and damping characteristics for the converter, effectively improving the frequency and voltage support capabilities of the microgrid under weak grid conditions. This has become the mainstream control strategy for grid-connected converters. In the actual operation of PV-storage microgrids, the converter needs to flexibly switch between off-grid and grid-connected operation modes. When the system switches from off-grid to grid-connected, because the converter output voltage during off-grid operation is generated by its own control law and lacks direct coupling with the grid voltage, deviations in voltage amplitude, frequency, and phase are inevitable at the switching instant. These deviations will cause inrush currents and power oscillations at the moment of closing, which may, in severe cases, trigger malfunctions of protection devices or even damage power devices, threatening the safe and stable operation of the system.

[0003] To suppress the impact during grid connection switching, pre-synchronization control of the converter output voltage is required before switching to ensure that the converter output voltage is consistent with the grid voltage in amplitude, frequency, and phase. Currently, several patents have proposed different technical solutions around pre-synchronization control strategies. Chinese invention patent CN110212590B addresses the problem of inaccurate frequency and phase angle synchronization caused by phase angle difference jumps during grid connection pre-synchronization. It proposes a pre-synchronization control system comprising a phase angle difference modulation module, a region judgment module, and a linear approximation modulation unit. By eliminating the disturbance of phase angle difference jumps to the frequency compensation signal, it improves the phase angle pre-synchronization accuracy. Chinese invention patent application CN117060488A proposes a technical solution for pre-synchronization control using a second-order generalized complex vector filter. By extracting the fundamental positive-sequence component of the grid voltage and calculating the voltage amplitude and phase angle compensation in real time, it can achieve smooth grid connection of grid-connected inverters under three-phase imbalance and grid background harmonic conditions. Chinese invention patent application CN119382224A discloses a pre-synchronization scheme that calculates the equivalent phase difference and amplitude deviation through coordinate transformation. Through iterative adjustment, the equivalent phase difference converges to zero, achieving frequency and amplitude pre-synchronization between the grid-connected converter and the power grid. The above scheme improves pre-synchronization control from the perspectives of phase disturbance elimination, harmonic suppression, and coordinate transformation. Based on this, pre-synchronization methods without phase-locked loops have also attracted widespread attention due to their simple structure and the elimination of the need for additional phase-locked loops.

[0004] Chinese invention patent CN111313469A discloses a pre-synchronization method without a phase-locked loop. This method involves acquiring the three-phase voltage of the power grid and the three-phase voltage output from a virtual synchronizing machine, and then performing a 3 / 2 coordinate transformation to obtain the pre-synchronization data. The sinusoidal value of the phase difference and the voltage amplitude difference in the coordinate system are then fed into a PI controller to achieve rapid pre-synchronization of the virtual synchronous machine without a phase-locked loop. Chinese invention patent CN116111867B proposes a pre-synchronization scheme that generates voltage difference compensation by constructing virtual instantaneous power. At the same time, it designs a nonlinear variable integral coefficient control method based on grid voltage difference, which can achieve rapid pre-synchronization without a phase-locked loop under the full range of phase differences.

[0005] In the specific implementation of the aforementioned pre-synchronization control strategy, the extraction and processing of the phase error signal is the core link determining the pre-synchronization performance. Because the phase angle output by the phase-locked loop has a periodic winding property of [0, 2π), when the phase difference crosses the period boundary, a first-type jump discontinuity of ±2π will occur, causing the pre-synchronization controller to output incorrect adjustment commands. Currently, there are two main methods to address this problem. The first method is the phase angle difference normalization compensation method, which constrains the phase error within the range of [-π, π] by writing piecewise logic. Although this method ensures the consistency of the phase difference sign and physical meaning throughout the entire period, its essence is a piecewise non-analytical function, with non-differentiable jump discontinuities at the ±π boundary. This non-analytical and non-differentiable mathematical characteristic not only makes small-signal modeling and stability analysis of the system difficult, but also easily introduces additional computational delays in actual digital implementation due to the piecewise logic's judgment and branch jumps. The second method is the trigonometric function transformation method, that is, using sin(Δ... The phase error is smoothly reconstructed, and the step pulses at ±π are eliminated by utilizing the continuous differentiability of trigonometric functions. However, this method implicitly relies on the small-angle approximation premise sin(Δπ / 2) / π. )≈Δ The linear feedback relationship can only be approximately satisfied when the phase difference approaches zero. Under conditions of large initial phase difference, the large-signal nonlinearity of the sine function causes the equivalent gain of the PI controller to decrease significantly as the phase difference increases, approaching zero when close to ±π. This results in severely insufficient transient regulation power and extremely slow system dynamic response. In summary, existing phase error processing methods suffer from the inherent defect of being unable to simultaneously achieve both continuity and linearity: the phase angle difference normalization compensation method guarantees linearity but destroys continuity, while the trigonometric function transformation method guarantees continuity but loses linearity. Neither of these methods can meet the requirements of pre-synchronous control for stability and rapid response across the entire operating range.

[0006] Furthermore, most existing pre-synchronization control strategies focus on optimizing the control law itself, lacking systematic optimization design for transient response characteristics of transient power correction during pre-synchronization and various control parameters (including the proportional coefficient, integral coefficient, acceleration factor, nonlinear exponent, and grid connection decision weight coefficient). Traditional schemes typically use empirical tuning or trial-and-error methods to determine these parameters, failing to fully explore the coupling relationships between parameters and making it difficult to achieve global optimum under multi-objective constraints (simultaneously considering impact suppression, convergence speed, and steady-state accuracy). Therefore, there is an urgent need for a pre-synchronization grid connection control scheme that can achieve continuous linear restoration of phase difference across the entire phase operating range, while possessing adaptive parameter optimization and transient power adjustment capabilities. Summary of the Invention

[0007] In view of this, the present invention provides a pre-synchronous grid connection method and system for grid-connected converters, which aims to solve the problems of excessive inrush current at the moment of closing and unstable grid connection transition caused by phase, frequency and amplitude deviations in the grid connection scenario of photovoltaic-storage microgrid.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: On the one hand, the present invention provides a method for pre-synchronization grid connection of a grid-connected converter, comprising the following specific steps: S1. Establish a pre-synchronization control architecture. During grid connection switching, the active power-frequency control loop of the converter is switched to transient power correction, and the transient power correction is used to dynamically compensate the active power reference value to achieve pre-synchronization adjustment of phase and frequency; at the same time, the control target of the reactive power-voltage control loop is switched from the off-grid operation set value to the actual grid voltage amplitude to achieve pre-synchronization adjustment of voltage amplitude.

[0009] The specific method for establishing the pre-synchronization control architecture is as follows: In off-grid operation mode, the converter uses a virtual synchronous machine (VSG) control to provide voltage and frequency support. The virtual synchronous machine control includes an active-frequency control loop and a reactive-voltage control loop. When the upper-level energy management system issues a grid connection switching command, the system switches from off-grid operation mode to pre-synchronization mode. In pre-synchronization mode, the active-frequency control loop introduces transient power correction, and achieves phase and frequency pre-synchronization by adjusting the phase and frequency of the virtual rotor. The reactive-voltage control loop switches the voltage amplitude reference value to the actual voltage amplitude of the grid and switches the reactive power reference value to the reactive power reference value of the grid, thereby achieving pre-synchronization of voltage amplitude.

[0010] S2. Global continuous linearization extraction of phase difference is achieved using complex vector arctangent operation. The voltage signals from the grid-connected side and the converter side are acquired, and after coordinate transformation, orthogonal components are constructed to characterize the phase difference information. The orthogonal components are solved using the four-quadrant arctangent function to obtain a corrected phase difference signal that has a continuous linear mapping relationship with the true phase difference. The corrected phase difference signal is used as a feedback quantity to participate in subsequent closed-loop regulation.

[0011] The specific steps to obtain the corrected phase difference signal are as follows: extract the orthogonal components of the grid voltage and the converter output voltage in the stationary two-phase coordinate system using coordinate transformation; construct the cross product and inner product of the complex voltage vector based on the orthogonal components, where the cross product represents the sine component of the phase difference between the two voltage vectors and the inner product represents the cosine component of the phase difference between the two voltage vectors; use the cross product as the first input of the four-quadrant arctangent function and the inner product as the second input to calculate the corrected phase difference signal.

[0012] The corrected phase difference signal maintains a strict one-to-one linear mapping relationship with the real physical phase difference throughout the entire working range, and its partial derivative remains constant throughout the entire working range, eliminating the step jump at the period boundary and the nonlinear gain attenuation under large deviation in the traditional phase extraction method.

[0013] S3. Calculate the transient power correction with nonlinear adaptive gain. Based on the current phase difference and frequency difference, calculate the transient power correction using a processing method that includes nonlinear adaptive gain.

[0014] The specific steps for calculating the transient power correction are as follows: Obtain the phase difference and frequency difference calculated from real-time sampling; calculate the phase difference proportional term based on the phase difference, which is obtained by applying a nonlinear mapping to the phase difference; calculate the frequency difference compensation term based on the frequency difference, which is composed of the superposition of the proportional and integral components of the frequency difference; dynamically and adaptively adjust the gain term based on the phase difference using a nonlinear power term and a hyperbolic tangent function—automatically compressing the gain under large deviations to suppress transient power surges, and maintaining the gain under small deviations to maintain convergence speed; multiply the phase difference proportional term, the frequency difference compensation term, and the nonlinear adaptive gain term to obtain the unlimited transient power correction; apply upper and lower limit constraints to the unlimited transient power correction and use a feedback method for anti-integral saturation processing to obtain the final output transient power correction, which is then introduced into the power command of the active-frequency control loop to drive the converter to adjust the output phase and frequency to track the power grid.

[0015] S4. Construct a multivariate evaluation function to determine grid connection and closing conditions. Obtain the normalized values ​​of phase difference, frequency difference, and voltage amplitude difference respectively, and weight and synthesize each normalized value to form a comprehensive evaluation value for grid connection and closing. Execute the grid connection and closing operation after confirming that both individual deviations are within their respective allowable threshold ranges and the comprehensive evaluation value meets preset conditions.

[0016] The specific steps of the grid connection and closing operation are as follows: According to the grid connection technical specifications, allowable thresholds for phase difference, frequency difference, and voltage amplitude difference are set respectively; the real-time monitored deviations are normalized to obtain normalized values ​​for each deviation; a multivariate evaluation function is constructed using the weighted Euclidean norm of the normalized values ​​of each deviation to obtain a comprehensive evaluation value for grid connection and closing; the grid connection and closing judgment adopts two-level criteria: the first level requires that the absolute value of each individual deviation does not exceed its respective preset maximum allowable threshold, and the second level requires that the comprehensive evaluation value is not greater than the set threshold; when both criteria are simultaneously satisfied and remain stable for multiple consecutive sampling periods, the pre-synchronization is determined to be complete and the grid connection and closing operation is executed.

[0017] S5. An adaptive optimization algorithm based on gradients is used to iteratively optimize the pre-synchronization control parameters. Taking the comprehensive performance index of the pre-synchronization process as the optimization objective, an adaptive optimization algorithm based on gradients is used to iteratively optimize the pre-synchronization control parameters. After each iteration update, constraints are applied to the parameters to make them meet the preset physical feasibility boundary conditions until the convergence condition is met, and then the optimal parameter configuration is output.

[0018] The specific steps of the gradient-based adaptive optimization algorithm for iteratively optimizing pre-synchronization control parameters are as follows: The control parameters to be optimized are organized into a parameter vector, which includes at least the phase difference proportional coefficient, frequency difference proportional coefficient, integral coefficient, acceleration factor, nonlinear exponent, and various weighting coefficients; the comprehensive performance index of the pre-synchronization process is used as the optimization objective function, which includes a weighted evaluation of dynamic overshoot, settling time, and steady-state error; the gradient of the objective function with respect to each parameter is estimated using the finite difference method, and the parameter vector is iteratively updated using the adaptive moment estimator (Adam); after each iteration, projection constraints are applied to the parameter vector: the weighting coefficients remain non-negative and their normalized sum is a set value, and each gain coefficient remains within a preset upper and lower limit range; when the relative change in the objective function value between two adjacent iterations is less than a set threshold or the change in the parameter vector is less than a set threshold, convergence is determined, and the current parameter configuration is output.

[0019] On the other hand, the present invention also provides a grid-connected converter pre-synchronous grid-connected system, wherein the method described in any of the above-mentioned embodiments includes: The pre-synchronization architecture establishment module is used to switch the active power-frequency control loop of the converter into transient power correction after receiving the grid connection switching command to achieve phase and frequency pre-synchronization adjustment, and at the same time switch the control target of the reactive power-voltage control loop to the actual voltage amplitude of the grid to achieve amplitude pre-synchronization. The phase difference continuous extraction module is used to acquire voltage signals from the grid side and the converter side, construct an orthogonal component group after coordinate transformation, and solve the orthogonal component group using the four-quadrant arctangent function to obtain a corrected phase difference signal that maintains a continuous linear mapping relationship with the true phase difference throughout the entire working range. The transient power correction module is used to generate a transient power correction amount with nonlinear gain adjustment characteristics based on the current phase difference and frequency difference. After the transient power correction amount is subjected to amplitude constraint and anti-integral saturation processing, it is introduced into the active-frequency control loop. The grid connection and closing determination module is used to obtain the normalized values ​​of phase difference, frequency difference and voltage amplitude difference respectively, and perform weighted summation to form a comprehensive evaluation value. It adopts a two-level criterion verification mechanism to determine the grid connection conditions. The grid connection and closing operation is executed after both levels of criteria are met and continuously confirmed. The parameter adaptive optimization module is used to adaptively adjust the pre-synchronization control parameters based on the overall performance of the pre-synchronization process, using an iterative optimization algorithm based on gradient estimation and adaptive step size adjustment, until the convergence condition is met and the parameter configuration is locked.

[0020] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a method and system for pre-synchronization grid connection of grid-connected converters, which has the following beneficial effects: (1) This invention designs a global continuous linear restoration method for phase difference based on the arctangent of a complex vector. The Clarke transform is used to extract the grid voltage and converter output voltage at rest. Orthogonal components in the coordinate system are used to construct the cross product and inner product of the complex voltage vector as continuous spatial projection components without periodic steps. The four-quadrant arctangent function atan2 is introduced to realize (- π This method achieves a strict 1:1 linear restoration of the phase difference across the entire interval [π]. It completely solves the phase period boundary step problem at the signal space transformation level, while overcoming the discontinuity of the phase angle difference normalization compensation method and the nonlinear gain attenuation defect of the trigonometric function transformation method, thus ensuring the continuity and linearity of the phase error feedback signal across the entire operating range.

[0021] (2) This invention proposes a transient power correction amount with nonlinear adaptive gain. Design Methodology. A transient power correction function integrating multi-parameter nonlinear adaptive gain is designed, using a nonlinear power term... The combination with the hyperbolic tangent function tanh enables dynamic adaptive gain adjustment. Under large initial deviations, the gain is automatically compressed to suppress transient power surges; under small deviations, the gain recovers, accelerating convergence. Simultaneously, the frequency difference is introduced into the integrator to form a predictive braking negative feedback mechanism, avoiding overshoot and oscillations in the integrator near the synchronization point.

[0022] (3) This invention proposes a pre-synchronization control parameter tuning method based on gradient descent adaptive optimization: a gradient descent parameter optimization method with the pre-synchronization comprehensive performance index as the objective function is proposed, which optimizes the pre-synchronization control parameters (the ratio coefficient of phase difference to frequency difference). K a1 , K a2 Integral coefficient K b Acceleration factor k Nonlinear exponent Grid connection determination weighting coefficient , , The parameters are organized as a parameter vector, and a weighted objective function containing the peak value of the impact current, convergence time, and steady-state accuracy is constructed. The Adam adaptive moment estimator optimizer is used for iterative solution, and the multi-objective global optimal tuning of the pre-synchronization control parameters is realized.

[0023] (4) This invention proposes a multivariate evaluation function that normalizes and weights the phase difference, frequency difference, and voltage difference. Q ( t This function, by setting allowable thresholds and weighting coefficients for each deviation component (which can be determined using the gradient descent adaptive optimization method described above), achieves a comprehensive judgment of grid connection conditions. This function can flexibly adjust the sensitivity of each component according to actual engineering constraints, improving the reliability and flexibility of grid connection and closing judgment.

[0024] The method proposed in this invention can significantly solve the stability problems such as inrush current, frequency fluctuations, and power overshoot caused by the defects of traditional pre-synchronization strategies during the switching of photovoltaic-storage grid-connected converters from off-grid mode to grid-connected mode. It improves the continuity and speed of phase tracking during the pre-synchronization process of the grid-connected converter, effectively suppresses pre-synchronization transient power surges and active power fluctuations at the moment of closing, and enhances the safe closing and stable operation capability of the grid-connected grid-connected converter system in pre-synchronization switching scenarios. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0026] Figure 1 This is a block diagram of the phase pre-synchronization process control. Figure 2 A schematic diagram of the spatial trajectory of the output quantity of the complex vector arctangent pre-synchronization strategy; Figure 3 The diagram shows the structure of the active power control loop based on complex vector arctangent correction. Figure 4 The control block diagram of VSG is an improvement based on the complex vector arctangent pre-synchronous grid-connected control strategy; Figure 5 The flowchart shows the pre-synchronization grid connection control based on complex vector arctangent optimization. Figure 6 Figure showing the experimental results of the pre-synchronization phase tracking process; Figure 7 The experimental waveforms comparing key parameters before and after optimization in the pre-synchronization process are shown (where, Figure 7 'a' represents the phase difference comparison. Figure 7 b represents the transient power comparison. Figure 7 c represents a comparison of system frequencies. Figure 8 The waveform diagrams are used to compare key parameters during grid connection operation. Figure 8 a represents a comparison of system frequencies. Figure 8 b represents a comparison of active power at the grid connection point. Figure 8 c represents the reactive power comparison at the grid connection point. Detailed Implementation

[0027] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0028] It is important to understand that the fundamental goal of pre-synchronization control is to bring the amplitude, frequency, and phase difference between the converter output voltage and the grid voltage within the allowable range before closing the circuit, so as to minimize the inrush current and achieve smooth grid connection. The active component of the inrush current at the moment of grid connection closing can cause faults such as transient deviation of DC bus voltage and virtual rotor power angle oscillation. The reactive component mainly affects the transient deviation of the grid connection point voltage, thus affecting grid connection safety.

[0029] This embodiment provides a method and system for pre-synchronization grid connection of grid-type converters based on complex vector arctangent optimization and parameter adaptive optimization, including the following steps: S1. Establish a pre-synchronization control architecture. When switching to grid connection, switch the active power-frequency control loop of the converter to transient power correction. Use the transient power correction to dynamically compensate the active power reference value to achieve pre-synchronization adjustment of phase and frequency. At the same time, switch the control target of the reactive power-voltage control loop from the off-grid operation set value to the actual voltage amplitude of the grid to achieve pre-synchronization adjustment of voltage amplitude. Specifically, in S1, the steps for establishing a pre-synchronization architecture include: S11. In off-grid operation mode, the converter uses Virtual Synchronous Machine (VSG) control to provide voltage and frequency support. The active-frequency control loop in the VSG control is mainly responsible for frequency and phase regulation, and its control equation is shown in the following formula. The reactive-voltage control loop achieves amplitude pre-synchronization by switching the reference command to the actual voltage amplitude of the grid and the reactive power reference value.

[0030] In the formula: H This is a virtual time constant; D p This is the per-unit value of the damping coefficient; P ref This is a reference value for active power. P e This refers to the output active power; k q Δ is the integral coefficient; ( t )for t The difference between the phase of the inverter and the grid at any given moment; For grid phase; Δ ω ( t )for t The difference between the inverter frequency and the grid frequency at any given time; ω grid The power grid frequency; ω ref The rated frequency; P 2 is t The transient reference power required to change the inverter output voltage and phase during the pre-synchronization process. P 2 is a function containing the output phase difference; Q ref This is a reactive power instruction; Q This refers to the output reactive power; E 0 represents the standard voltage value; Dq This is the voltage droop factor; E To control the amplitude of the output electromotive force.

[0031] S12. Grid connection switching command trigger and pre-synchronization mode start-up. When the upper-level energy management system issues a grid connection switching command, the system switches from off-grid operation mode to pre-synchronization mode. In this mode, the pre-synchronization control module starts working, actively adjusting the phase, frequency, and amplitude of the converter output voltage to gradually track the grid voltage, creating conditions for subsequent closing operations.

[0032] S13. Control loop switching strategy in pre-synchronization mode. After entering pre-synchronization mode, the active power-frequency control loop introduces transient power correction. P 2(t) achieves pre-synchronization of phase and frequency by adjusting the phase and frequency of the virtual rotor; the reactive power-voltage control loop switches the reference command to the actual voltage amplitude and reactive power reference value of the power grid, achieving pre-synchronization of voltage amplitude. The two loops work together to ensure that the converter output voltage simultaneously approaches the grid voltage in three dimensions: amplitude, frequency, and phase.

[0033] It's important to understand that traditional methods directly output the phase angle and take the difference using a phase-locked loop (PLL). However, due to the phase angle's [0, 2]... π The periodic winding characteristic will exhibit ±2 at the end of the cycle. π The step jump occurs. The tiny hysteresis angle within the jump interval is misinterpreted as a lead angle of nearly 2π, resulting in a huge reverse adjustment power output, forming a pulse-like oscillating excitation.

[0034] The existing phase difference normalization compensation method constrains the phase error to [-] using piecewise logic. π Within the range of [π], although the consistency between the sign and physical meaning of the phase difference is guaranteed, due to the ± π The existence of non-differentiable jump discontinuities at the boundary leads to its non-analytical and non-differentiable mathematical properties, posing difficulties for small-signal modeling and stability analysis. The trigonometric function transformation method utilizes sin( While smoothing the phase error eliminates the step pulse, the equivalent gain decreases under conditions of large initial phase difference. The power increases and then decreases, approaching zero when it is close to 2π, resulting in a severe deficiency in transient regulation power.

[0035] The complex vector arctangent optimization method proposed in this invention completely solves the above contradictions from the perspective of signal space transformation.

[0036] S2 acquires the voltage signals from the grid-connected side and the converter side, and constructs an orthogonal component to characterize the phase difference information after coordinate transformation. The orthogonal component is solved using the four-quadrant arctangent function to obtain a corrected phase difference signal that has a continuous linear mapping relationship with the true phase difference. The corrected phase difference signal is used as a feedback quantity to participate in the subsequent closed-loop regulation. Specifically, in S2, the steps to obtain the corrected phase difference signal that has a continuous linear mapping relationship with the true phase difference are as follows: S21. Complex Vector Construction and Clarke Transform: The proposed complex vector arctangent optimization method thoroughly resolves the above contradictions from the perspective of signal space transformation. Firstly, the Clarke transform is used to extract the grid voltage and converter output voltage at rest. αβ Orthogonal components in the coordinate system: in, For the grid voltage at rest coordinate system Axial components; For the grid voltage at rest coordinate system Axial components; For the converter output voltage at rest coordinate system Axial components; For the converter output voltage at rest coordinate system Axial components; This refers to the voltage amplitude of the power grid. This refers to the output voltage amplitude of the converter. The phase angle of the grid voltage; Converter output voltage phase angle; Further construct the cross product of complex voltage vectors inner product To obtain continuous spatial projection components that do not contain periodic steps: S22, Four-quadrant arctangent calculation and error signal output: The above orthogonal coordinate set with first-order continuous differentiability ( , As a bivariate variable, the cross product is... As the first input to the arctangent function in the four quadrants, the inner product... As the second input, it is processed through the four-quadrant arctangent function atan2( , This allows for the continuous calculation of the error phase angle. At this point, the independent variable changes from the phase angle, which originally possessed periodic properties, to a periodically non-circular, continuous planar coordinate. , ).

[0037] The final phase difference signal input to the pre-synchronous PI controller is: The spatial trajectory of the output quantity is as follows Figure 2 As shown: X =cosΔ , Y =sinΔ The radius of the formation is E grid E vsg The input plane is a circle, and the output phase principal value expands along the Z-axis, presenting an overall spatial curve that spirals tightly around the cylindrical surface. This curve is continuously differentiable in all four quadrants, but only in the Δ... There exists a step discontinuity at ±π.

[0038] S23. The partial derivatives of the corrected phase difference signal output by the function in the full-plane view strictly satisfy: Even under extreme transient conditions such as the initial stage of black start or significant phase difference before grid connection, the feedback error signal input to the PI controller maintains a strict 1:1 linear mapping with the actual physical phase difference, solving the nonlinearity problem of the traditional trigonometric function transformation method. Therefore, the PI controller can fully utilize the preset proportional gain across the entire phase operating range. K p Its control capability eliminates the transient response lag caused by nonlinear gain from a mechanistic perspective.

[0039] S3. Based on the current phase difference and frequency difference, calculate the transient power correction using a processing method that includes nonlinear adaptive gain; Specifically, construct the transient power correction quantity P The steps for 2(t) are as follows: S31. The system calculates the transient power correction based on the phase difference and frequency difference obtained from real-time sampling. P 2(t) is superimposed on the active power reference value of the active-frequency control loop. P The basic expression for 2(t) is shown below: In the formula: 、 These are the proportionality coefficients for the phase difference and the frequency difference, respectively.K b The integral coefficient; k As an acceleration factor; It is a constant less than 1.

[0040] S32. Predictive braking negative feedback mechanism of the integral term: The frequency difference in the integral term is essentially the derivative of the phase difference. Treating this term separately yields: Will Introducing negative feedback into the integrator utilizes its characteristic of having the opposite sign to the phase difference during the later stages of pre-synchronization. This allows negative feedback to be injected into the integrator, enabling the accumulated nonlinear terms to be offset in advance as the system approaches the synchronization point, thus achieving "predictive braking."

[0041] S33. Dynamic adjustment mechanism of nonlinear adaptive gain: This function adjusts through a nonlinear power term. and The function achieves dynamic adaptive adjustment of gain: it automatically restrains the gain under large initial errors, acting as a soft start to suppress transient power surges; and it can play a strong correction role when the error shrinks in the later stage, achieving precise pre-synchronization of phase and frequency.

[0042] S4 acquires the normalized values ​​of phase difference, frequency difference and voltage amplitude difference respectively, and performs weighted summation on each normalized value to form a comprehensive evaluation value for grid connection and closing; the grid connection and closing operation is performed after the conditions are met and continuously confirmed that each individual deviation is within its respective allowable threshold range and the comprehensive evaluation value meets the preset conditions. The specific steps are as follows: S41. During pre-synchronization adjustment, the system monitors the phase difference, frequency difference, and voltage amplitude difference in real time, and sets corresponding allowable deviation thresholds according to relevant grid connection technical specifications. These thresholds are primarily set based on the converter's rated capacity, short-circuit impedance, the voltage level of the connected grid, and the equipment's withstand capability. To unify physical quantities of different dimensions and orders of magnitude for comparison, each deviation is normalized: S42. After normalization, the values ​​of each component directly reflect their proportional relationship relative to the allowable threshold, facilitating dimensionless synthesis. Since the impact of each deviation on grid connection impact is independent, and its influence generally increases monotonically with the increase of the deviation, a multivariate evaluation function is constructed using a weighted Euclidean norm (i.e., the comprehensive evaluation value of grid connection): In the formula: for t The instantaneous phase difference after optimization and correction by the complex vector arctangent; fort Instantaneous frequency difference at any moment; for t Difference in voltage amplitude at different times; , , These are the allowable deviation thresholds; , , Let be the weight coefficients of each component, and satisfy . + + =1、 >0、 >0、 >0.

[0043] The above multivariate evaluation function Q The physical meaning of (t) is that when any deviation exceeds its allowable threshold, the corresponding normalization term is greater than 1, and under the action of the squared term, this deviation affects... Q The contribution of (t) is amplified, thus ensuring that the criterion is sufficiently sensitive to severe deviations. The introduction of weights reflects the relative importance of different deviation amounts to the severity of grid connection impacts. For example, in certain grid connection scenarios with high frequency stability requirements, the weight can be appropriately increased. w f The value of .

[0044] Preset corrected phase difference Frequency difference and voltage amplitude difference The maximum allowable limit for independent units is: 、| |≤ 、| |≤ If any of the following exceeds the limit, closing the circuit is prohibited. The following are the limits set according to national standards: =10°, =0.1Hz, =5%.

[0045] A multivariate evaluation function Q(t) is further introduced as a comprehensive margin criterion: grid connection and closing operation can only be performed when both criteria are met simultaneously.

[0046] According to the geometric meaning of the weighted Euclidean norm, Q (t)=1 corresponds to a hyperellipsoidal boundary in the three-dimensional normalized bias space, centered at the origin, with the semi-axis length determined by the weights. When Q When (t)≤1, the state point is located inside or on the boundary, indicating that the comprehensive weighted value of each deviation is within the allowable range; when QWhen (t)>1, the state point is located outside the boundary, indicating that the overall degree of bias exceeds the allowable range. It should be noted that due to the compression effect of the weights, when a single bias exceeds the threshold... Q (t) may still be no greater than 1, therefore Q (t)≤1 cannot be used as a standalone grid connection criterion. Therefore, the grid connection conditions are determined by two levels of criteria: the first level is the single-deviation criterion, which requires that the instantaneous phase difference, frequency difference, and voltage amplitude difference do not exceed their respective allowable thresholds, ensuring that each dimension is within a safe range; the second level is the comprehensive margin criterion, which requires... Q (t)≤1, to avoid situations where multiple deviations, although not exceeding their individual limits, still pose a significant grid connection risk when superimposed. A grid connection command and closing operation can only be issued and executed when both criteria are met simultaneously; otherwise, the pre-synchronization control should continue to operate to further reduce the deviation until both criteria are met.

[0047] S43. Continuous Confirmation Mechanism and Grid Connection Command Execution: Further, to improve the reliability of the grid connection criterion, the determination is made within multiple consecutive sampling periods. Q The value of (t). Let the sampling period be... T s If there exist N consecutive periods that satisfy Q If (t)≤1, the system is determined to have entered the grid connection permission window. N is set according to the system's dynamic response speed and anti-interference requirements, with a typical value range of 3~10. This continuous confirmation mechanism can effectively avoid misjudgments caused by single measurement noise or transient disturbances.

[0048] S5 takes the comprehensive performance index of the pre-synchronization process as the optimization target, and uses a gradient-based adaptive optimization algorithm to iteratively optimize the pre-synchronization control parameters. After each iteration update, constraints are applied to the parameters to make them meet the preset physical feasibility boundary conditions until the convergence condition is met, and then the optimal parameter configuration is output.

[0049] The specific method is as follows: S51, The parameter vector is defined as follows: The initial parameters are based on empirical values ​​set at the factory: K a1 =80、 K a2 =2,K b =15, k =0.8、 α =0.6, =0.6、 =0.25、 =0.15, with this initial value as the starting point for iteration, and subsequent optimizations will start from the parameters that converged in the previous iteration.

[0050] Online hard constraint upper and lower limits: K a1 ∈[10, 200]), K a2 ∈[0.2, 10], K b =∈[2, 60], k ∈[0.1, 3], α ∈[0.2, 0.95); the weight coefficients satisfy non-negativity and + + =1.

[0051] S52. To comprehensively evaluate the dynamic and steady-state performance of the pre-synchronization process, a multi-objective comprehensive performance index is constructed: In the formula: J overshoot It is an overshoot indicator used to suppress the maximum overshoot of phase, frequency, or voltage amplitude during pre-synchronization, so as to avoid excessive dynamic deviation. J settle To adjust the time index, in order to speed up the process of the system entering the grid connection permission window; J steady It is a steady-state error index used to ensure that all deviations before grid connection eventually converge to a sufficiently small range; λ 1. λ 2. λ 3 represents the weighting coefficients for each item, reflecting the relative importance of different performance indicators in the optimization objective. These coefficients are typically selected based on actual grid connection speed requirements and impact resistance requirements, and must meet certain conditions. λ 1+ λ 2+ λ 3 = 1.

[0052] Specifically, the three performance indicators can be further defined as follows: In the formula: T max This refers to the total duration of a single pre-synchronous simulation or experiment. T steady is the length of the steady-state evaluation window near the end time, and 1[·] is an indicator function that takes the value 1 when the condition is met and 0 otherwise. J overshoot Directly using the defined multivariate evaluation function Q ( t The maximum value during the entire process is taken as the overscheduled amount; J settle statistics Q ( tThe number of sampling points exceeding the allowable threshold of 1 reflects how quickly the system enters the grid connection window; J steady Then take within the steady-state window Q ( t The average value of ) is used to evaluate the final synchronization accuracy.

[0053] S53, Finite Difference Gradient Estimation: Due to the comprehensive performance index J ( It is usually not possible to directly obtain information about the parameter vector through analytical methods. The gradient is obtained, therefore the finite difference method is used to estimate the sensitivity of the objective function to each dimension parameter: In the formula: For parameter vectors The i One component; e i For the first i 1-dimensional unit vector; ε This is a small perturbation, usually taken as 1‰ to 1% of the current order of magnitude of the parameter, in order to balance numerical accuracy and gradient estimation stability.

[0054] S54. Adam Adaptive Moment Estimator Optimizer Parameter Update: After obtaining the gradient estimate, the Adam optimizer is used for parameter update. The Adam optimizer combines adaptive estimation of the first and second moments of the gradient, and has the advantages of fast convergence speed and low dependence on the learning rate. It is suitable for pre-synchronous parameter optimization problems with strong parameter coupling and non-convex objective functions. The parameter update formula is as follows: in n For the number of iterations, η For the learning rate. A more complete Adam update would be: In the formula: For the first n The gradient of the objective function in the next iteration; m n and v n These are the first and second moments of the gradient, respectively. and The exponential decay rate is estimated by moments, with typical values ​​of 0.9 and 0.999, respectively. and This is the moment estimate after bias correction; To prevent division by zero, small constants are usually taken as 10. -8 ; The typical value is 0.008; the number of iterations is constrained by the computing power of online operating hardware and real-time control. n Usually no more than 60 times; S55. Projection Constraints and Convergence Criteria: After each iteration completes the parameter update, projection constraints must be applied to the parameter vector to ensure it meets physical feasibility and constraint conditions. For example, for the weight coefficients... , w f , w V Perform normalized projection to keep it non-negative and sum to 1; control the gain. K a1 , K a2 , K b , k The upper and lower limits are adjusted to ensure they are within a reasonable range for engineering applications. Subsequently, the updated parameter vectors are used to perform online pre-synchronous process simulations to calculate the new comprehensive performance indicators. J ( θ The gradient estimation and parameter update process is repeated until the convergence condition is met.

[0055] The convergence criterion is set as follows: the relative change in the objective function value between two consecutive iterations is less than a threshold. ,Right now: Alternatively, it can be set to a maximum change in the parameter vector that is less than a threshold. ,Right now: In the example, take =1×10 -3 ; =5×10 -4 .

[0056] When one of the above convergence conditions is met, the final parameter vector is output: The aforementioned vectors are used as the parameters after convergence. This method, by automatically optimizing instead of traditional empirical trial-and-error, realizes the transformation of pre-synchronization parameters from empirical design to adaptive parameter solving, which can effectively improve pre-synchronization speed, suppress dynamic overshoot, and reduce steady-state grid connection error.

[0057] It also includes: S6, verifying the phase tracking, transient power surge and frequency offset suppression effects of the method of the present invention in the pre-synchronization stage, as well as the improvement effects of frequency fluctuation and active power surge in the grid connection switching stage through hardware-in-the-loop simulation experiments.

[0058] Figure 1This is a block diagram illustrating the control principle of the method of the present invention. The diagram shows the overall control flow, with the complex vector arctangent phase difference extraction module at its core, using a transient power correction variable containing nonlinear adaptive gain to perform feedforward correction on the VSG active-frequency control loop, and a multivariate evaluation function to determine the grid connection conditions. The complex vector arctangent module determines the grid connection conditions based on the grid voltage and the converter output voltage. The axial components construct a complex voltage vector, and the global continuous phase difference signal is directly restored using the four-quadrant arctangent operation. This signal is then fed into the transient power correction module for closed-loop regulation. When the multivariate evaluation function determines that the phase, frequency, and voltage amplitude differences have all converged to within the allowable range, the system issues a grid-connection command.

[0059] Figure 2 This is a spatial trajectory diagram extracted from the arctangent phase difference of a complex vector. The diagram illustrates the relationship between the grid voltage and the converter output voltage. The rotational trajectory of the voltage vector space in the complex plane and the phase difference between them. The three-phase voltages are mapped to [the plane] using the Clarke transform. A stationary coordinate system is used to construct a complex voltage vector containing phase difference information, and the cross product and inner product are calculated. Finally, the phase difference is continuously linearized and restored in the entire interval [-π, π] by using the four-quadrant arctangent function atan2(Y, X), which avoids the step jump problem at the periodic boundary of the traditional method.

[0060] Figure 3 This is a detailed block diagram of the active power-frequency control loop in pre-synchronization control. The diagram illustrates the transient power correction in pre-synchronization mode. P How does 2(t) superimpose onto the VSG active power reference value, and how is the converter output phase and frequency adjusted through the virtual rotor motion equation to gradually track the grid voltage? The figure clearly shows the sampling of phase and frequency differences, the calculation of transient power correction, and the dynamic adjustment process of nonlinear adaptive gain.

[0061] Figure 4 This is a block diagram of the VSG basic control architecture. The diagram illustrates the complete control structure of a grid-connected converter based on a virtual synchronous machine in off-grid operation mode, including an active power-frequency control loop and a reactive power-voltage control loop. In pre-synchronization mode, the active power-frequency control loop introduces transient power correction for active phase and frequency adjustment, while the reactive power-voltage control loop switches the reference command to the actual voltage amplitude and reactive power reference value of the grid. The two loops work together to ensure that the converter output voltage simultaneously approaches the grid voltage in terms of amplitude, frequency, and phase.

[0062] Figure 5This is a flowchart of the pre-synchronization control process. The flowchart illustrates the complete control sequence from issuing the grid connection command, entering pre-synchronization mode, executing transient power correction and complex vector arctangent phase difference extraction, to the multivariate evaluation function determining the closing conditions and executing the grid connection operation. In pre-synchronization mode, the complex vector arctangent module continuously outputs the phase difference signal, the transient power correction module dynamically compensates for the phase and frequency difference, and the multivariate evaluation function monitors the deviation status in real time; when the evaluation function value... Q When (t)≤1 and remains stable for multiple consecutive sampling periods, the system issues a grid connection command to execute the closing operation.

[0063] Figure 7 This is a comparison diagram of experimental waveforms during the pre-synchronization phase. Figure 7 A comparison shows the phase difference tracking waveforms of the complex vector arctangent method of the present invention and the existing strategy. The method of the present invention achieves global continuous linear restoration of the phase difference, with no periodic step transitions in the entire range, and the pre-synchronization completion time is about 2.5s, which is about 38% faster than the 4s of the existing strategy. Figure 7 b. The comparison shows the waveform of the transient power correction. The strategy of this invention has no periodic power pulses throughout the process, and only a transient power correction of about 2 p.u. occurs at the initial moment of pre-synchronization before smoothly converging to zero. Figure 7 The comparison in section c shows the frequency response waveforms. The strategy of this invention generates only a brief frequency offset of 0.06Hz in the initial pre-synchronization phase, subsequently smoothly converging to the rated value of 50Hz, providing stable frequency conditions for grid-connected switching. This demonstrates that the strategy of this invention has significant advantages in phase tracking accuracy, transient power suppression, and frequency stability during the pre-synchronization phase.

[0064] Figure 8 This is a comparison chart of experimental waveforms during the grid connection switching phase. Figure 8 The comparison shows the frequency response waveforms under two methods: pre-synchronization grid connection and direct grid connection. Under the pre-synchronization grid connection operation, the frequency only fluctuates slightly by 0.02Hz and recovers within 1 second. Compared with the 0.2Hz oscillation amplitude and approximately 4.5s recovery time of direct grid connection, the frequency fluctuation is reduced by 90% and the recovery time is shortened by 78%. Figure 8 b. The comparison shows the active power surge waveform. After pre-synchronization, the active power fluctuation amplitude under grid connection operation is about 0.14 pu, compared with the peak of 3.2 pu (overshoot 220%) of direct grid connection. The active power surge is reduced by about 86%, and the transient recovery time is shortened by about 78%. Figure 8 The comparison shows the reactive power response waveforms. Under pre-synchronization and grid connection operation, the reactive power transition is smoother, the fluctuation amplitude is significantly reduced, and the system can recover to a stable operating state more quickly. In contrast, under direct grid connection, reactive power exhibits significant oscillations, and the recovery time is longer. This fully verifies the excellent suppression effect of the pre-synchronization control strategy and adaptive parameter optimization of this invention on inrush current, frequency fluctuations, and reactive power oscillations during grid connection switching.

[0065] In this embodiment, the photovoltaic-storage microgrid is equivalent to a DC voltage source via a DC-DC converter. U dc =1200V; power reference value and load are both 2MW; switching frequency is 5kHz; filter inductor is 0.2mH; filter capacitor is 20... μ F; filter internal resistance is 1.5mΩ; simulation step size is 50μs; grid voltage level is 35kV.

[0066] This experiment was conducted on the OPAL-RT RT-LAB OP5700 hardware-in-the-loop platform, which is equipped with the eHS electrical solver, and the simulation step size was set to 10. μs The control system uses an RTU-BO204 controller, which forms a semi-physical closed-loop test environment with the simulation platform through the IO interface.

[0067] The experiment was set to run for 20 seconds. Initially, the photovoltaic-storage microgrid operated off-grid. A grid connection signal was issued at 10 seconds to initiate pre-synchronization, and grid connection was executed at 15 seconds. The initial phase difference between the converter and the grid at the initial pre-synchronization moment was approximately 45°. The pre-synchronization control strategy proposed in this invention adds a phase difference adjustment PI controller to the VSG basic control architecture. The proportional gain of this PI controller is 0.9, and the integral gain is 0.005; the remaining parameter settings are consistent with the VSG control. To verify the effectiveness of the method, a conventional pre-synchronization strategy, a pre-synchronization strategy based on trigonometric function transformation, and the pre-synchronization strategy proposed in this paper were compared and verified.

[0068] The experiment was set to run for 20 seconds. Initially, the photovoltaic-storage microgrid operated off-grid. At 10 seconds, a grid-connection signal was issued to initiate pre-synchronization, and grid connection was executed at 15 seconds. The initial phase difference between the converter and the grid at the pre-synchronization start-up point was approximately 45°. The pre-synchronization control strategy proposed in this invention adds a phase difference adjustment PI control loop to the VSG basic control architecture. The proportional coefficient is 0.9, and the integral coefficient is 0.005. The remaining parameters are consistent with the classic VSG control. To verify the effectiveness of the method, a conventional pre-synchronization strategy and a pre-synchronization strategy based on trigonometric function transformation were compared.

[0069] This embodiment constructs a multivariate evaluation function based on phase difference, frequency difference, and voltage amplitude difference, and designs a complex vector arctangent phase difference extraction method—converting the periodic phase in the traditional phase-locked loop into a complex vector domain arctangent operation to achieve global continuous linearization. Simultaneously, a transient power correction with nonlinear adaptive gain is introduced to dynamically compensate for the phase and frequency differences. Furthermore, the Adam gradient descent algorithm is employed for multi-objective adaptive optimization of control parameters to suppress inrush current and frequency overshoot during grid-connected switching of the grid-connected converter. This method not only effectively achieves a rapid and smooth transition during pre-synchronous grid connection of the grid-connected converter, ensuring safe switching of the power system, but also provides a key control scheme for the stable operation of the grid-connected converter system.

[0070] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for pre-synchronous grid connection of a grid-connected converter, characterized in that, The specific steps include the following: A pre-synchronization control architecture is established. During grid connection switching, the active power-frequency control loop of the converter is switched to transient power correction. The transient power correction is used to dynamically compensate the active power reference value to achieve pre-synchronization adjustment of phase and frequency. At the same time, the control target of the reactive power-voltage control loop is switched from the off-grid operation set value to the actual grid voltage amplitude to achieve pre-synchronization adjustment of voltage amplitude. The voltage signals from the grid-connected side and the converter side are acquired, and after coordinate transformation, an orthogonal component is constructed to characterize the phase difference information. The orthogonal component is solved using the four-quadrant arctangent function to obtain a corrected phase difference signal that has a continuous linear mapping relationship with the true phase difference. The corrected phase difference signal is used as a feedback quantity to participate in the subsequent closed-loop regulation. Based on the current phase difference and frequency difference, the transient power correction is calculated using a processing method that includes nonlinear adaptive gain. Normalized values ​​of phase difference, frequency difference and voltage amplitude difference are obtained respectively, and the normalized values ​​are weighted and combined to form a comprehensive evaluation value for grid connection and closing. The grid connection operation is performed after the individual deviations are both within their respective allowable threshold ranges and the comprehensive evaluation value meets the preset conditions and are continuously confirmed. Taking the comprehensive performance index of the pre-synchronization process as the optimization objective, a gradient-based adaptive optimization algorithm is used to iteratively optimize the pre-synchronization control parameters. After each iteration update, constraints are applied to the parameters to make them meet the preset physical feasibility boundary conditions until the convergence condition is met, and then the optimal parameter configuration is output.

2. The pre-synchronization grid connection method for a grid-connected converter according to claim 1, characterized in that, The steps to establish a pre-synchronization control architecture include: In off-grid operation mode, the converter uses virtual synchronous machine control to provide voltage and frequency support. The virtual synchronous machine control includes an active-frequency control loop and a reactive-voltage control loop. When the superior energy management system issues a grid connection switching command, the system switches from off-grid operation mode to pre-synchronization mode; In pre-synchronization mode, the active power-frequency control loop introduces transient power correction, and achieves pre-synchronization of phase and frequency by adjusting the phase and frequency of the virtual rotor; the reactive power-voltage control loop switches the voltage amplitude reference value to the actual voltage amplitude of the power grid, and switches the reactive power reference value to the reactive power reference value of the power grid, thereby achieving pre-synchronization of voltage amplitude.

3. The pre-synchronization grid connection method for a grid-connected converter according to claim 1, characterized in that, The specific steps to obtain the corrected phase difference signal are as follows: The orthogonal components of the grid voltage and the converter output voltage in the stationary two-phase coordinate system are extracted by using coordinate transformation. The cross product and inner product of the complex voltage vector are constructed based on the orthogonal components. The cross product represents the sinusoidal component of the phase difference between the two voltage vectors, and the inner product represents the cosine component of the phase difference between the two voltage vectors. The cross product is used as the first input to the four-quadrant arctangent function, and the inner product is used as the second input to calculate the corrected phase difference signal.

4. The pre-synchronization grid connection method for a grid-connected converter according to claim 1, characterized in that, The specific steps for calculating the transient power correction are as follows: Obtain the phase difference and frequency difference calculated in real time; The phase difference proportional term is calculated based on the phase difference, which is obtained by applying a nonlinear mapping to the phase difference; the frequency difference compensation term is calculated based on the frequency difference, which is composed of the superposition of the proportional component and the integral component of the frequency difference. Based on the phase difference, the gain term is dynamically and adaptively adjusted through a nonlinear power term and a tanh function; Multiply the phase difference ratio term, the frequency difference compensation term, and the nonlinear adaptive gain term to obtain the unlimited transient power correction amount; The amplitude upper and lower limits are constrained by applying upper and lower limits to the unlimited transient power correction amount to obtain the final output transient power correction amount.

5. The pre-synchronization grid connection method for a grid-connected converter according to claim 1, characterized in that, The specific steps for grid connection and shutdown are as follows: According to the grid connection technical specifications, allowable thresholds for phase difference, frequency difference, and voltage amplitude difference are set respectively; The real-time monitored deviations are normalized to obtain the normalized values ​​of each deviation. By using the normalized values ​​of each deviation, a multivariate evaluation function is constructed using the weighted Euclidean norm to obtain the comprehensive evaluation value for grid connection and closing. The grid connection and closing determination adopts a two-level criterion: the first level requires that the absolute value of each individual deviation does not exceed its respective preset maximum allowable threshold; the second level requires that the comprehensive evaluation value is not greater than the set threshold. When both criteria are met simultaneously and the system remains stable for multiple consecutive sampling periods, the pre-synchronization is determined to be complete and the grid connection operation is executed.

6. The pre-synchronization grid connection method for a grid-connected converter according to claim 1, characterized in that, The specific steps of the gradient-based adaptive optimization algorithm for iteratively optimizing the pre-synchronization control parameters are as follows: The control parameters to be optimized are organized into a parameter vector, which includes at least the phase difference proportional coefficient, the frequency difference proportional coefficient, the integral coefficient, the acceleration factor, the nonlinear exponent, and each weight coefficient. The comprehensive performance index of the pre-synchronization process is used as the optimization objective function, and the comprehensive performance index includes a weighted evaluation of dynamic overshoot, settling time and steady-state error. The gradient of the objective function with respect to each parameter is estimated using the finite difference method, and the parameter vector is iteratively updated using an adaptive moment estimator optimizer. After each iteration, a projection constraint is applied to the parameter vector: the weight coefficients remain non-negative and the normalized sum is a set value, and each gain coefficient remains within the preset upper and lower limits. When the relative change in the objective function value between two consecutive iterations is less than a set threshold or the change in the parameter vector is less than a set threshold, convergence is determined and the current parameter configuration is output.

7. A grid-connected converter pre-synchronization system, employing the grid-connected converter pre-synchronization method according to any one of claims 1-6, characterized in that, include: The pre-synchronization architecture establishment module is used to switch the active power-frequency control loop of the converter into transient power correction after receiving the grid connection switching command to achieve phase and frequency pre-synchronization adjustment, and at the same time switch the control target of the reactive power-voltage control loop to the actual voltage amplitude of the grid to achieve amplitude pre-synchronization. The phase difference continuous extraction module is used to acquire voltage signals from the grid side and the converter side, construct an orthogonal component group after coordinate transformation, and solve the orthogonal component group using the four-quadrant arctangent function to obtain a corrected phase difference signal that maintains a continuous linear mapping relationship with the true phase difference throughout the entire working range. The transient power correction module is used to generate a transient power correction amount with nonlinear gain adjustment characteristics based on the current phase difference and frequency difference. After the transient power correction amount is subjected to amplitude constraint and anti-integral saturation processing, it is introduced into the active-frequency control loop. The grid connection and closing determination module is used to obtain the normalized values ​​of phase difference, frequency difference and voltage amplitude difference respectively, and perform weighted summation to form a comprehensive evaluation value. It adopts a two-level criterion verification mechanism to determine the grid connection conditions. The grid connection and closing operation is executed after both levels of criteria are met and continuously confirmed. The parameter adaptive optimization module is used to adaptively adjust the pre-synchronization control parameters based on the overall performance of the pre-synchronization process, using an iterative optimization algorithm based on gradient estimation and adaptive step size adjustment, until the convergence condition is met and the parameter configuration is locked.

8. A grid-connected converter pre-synchronous grid-connected system according to claim 7, characterized in that, The phase difference continuous extraction module includes: The coordinate transformation unit is used to transform the grid voltage and the converter output voltage to a stationary two-phase coordinate system to obtain the corresponding orthogonal components. Complex vector construction unit, used to construct the cross product and inner product of complex voltage vectors based on the orthogonal components; The arctangent solving unit is used to take the cross product as the first input of the four-quadrant arctangent function and the inner product as the second input to solve for the corrected phase difference signal. A linear holding unit is used to ensure that the corrected phase difference signal maintains a strict one-to-one linear mapping relationship with the real physical phase difference throughout the entire working range.

9. A grid-connected converter pre-synchronous grid-connected system according to claim 7, characterized in that, The transient power correction module includes: The deviation compensation calculation unit is used to generate a phase difference proportional term and a frequency difference compensation term based on the phase difference and frequency difference. The nonlinear gain adjustment unit is used to construct a nonlinear adaptive gain term through a nonlinear function, so as to achieve automatic gain compression under large deviation and adaptive adjustment to maintain gain under small deviation. The first transient power correction is obtained by multiplying the phase difference ratio term, the frequency difference compensation term, and the nonlinear adaptive gain term to obtain the unlimited transient power correction. The limiting and anti-saturation unit is used to apply amplitude constraints to the transient power correction and to perform anti-integral saturation processing using a feedback method to obtain the final output transient power correction.

10. A grid-connected converter pre-synchronous grid-connected system according to claim 7, characterized in that, The grid connection and closing determination module includes: The normalization processing unit is used to normalize each deviation to obtain the normalized value of each deviation. The weighted synthesis unit is used to construct a multivariate evaluation function using the normalized values ​​of each deviation quantity and the weighted Euclidean norm to obtain the comprehensive evaluation value. A two-level criterion determination unit is used to verify the first-level single-item deviation criterion and the second-level comprehensive evaluation value criterion; The stability confirmation unit is used to maintain stability confirmation for multiple consecutive sampling cycles after both levels of criteria are met, and finally issue a grid connection and closing command.

Images