Optimization Method for Sequential Dual-Loop PI Control in Three-Phase Four-Arm Photovoltaic Inverter

Abstract

This invention relates to the field of photovoltaic power generation technology and discloses a method for optimizing the sequence-sequence dual-loop PI control of a three-phase four-arm photovoltaic inverter. The method first obtains the inverter topology parameters and control strategy requirements, constructing a single-loop control model and a sequence-sequence decoupling model. Then, it extracts the strategy separation signal using DC input to obtain the positive-sequence and negative-sequence control components. Based on the dual-loop parameter matching rules and dynamic response constraints, it adjusts the controller parameters to obtain optimized PI parameters. Subsequently, it performs dynamic performance optimization and static performance verification, integrates the results to form an optimization scheme, and generates an implementation guide. This method achieves precise separation and independent adjustment of the positive-sequence and negative-sequence components, optimizes PI parameter matching, improves the inverter's dynamic response speed and steady-state performance under complex operating conditions, ensures output power quality, and is applicable to the control optimization of three-phase four-arm inverters in photovoltaic power generation systems.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation technology, specifically to an optimization method for sequential dual-loop PI control of a three-phase four-arm photovoltaic power generation inverter. Background Technology

[0002] As the global energy structure shifts towards clean energy, photovoltaic (PV) power generation, as an important form of renewable energy utilization, continues to see its installed capacity grow. In PV power generation systems, the inverter, as a key device connecting the DC generation side to the AC grid, directly impacts power quality, system stability, and energy conversion efficiency. Three-phase four-arm inverters, due to their ability to flexibly handle three-phase unbalanced loads and zero-sequence components, are widely used in distributed generation, microgrids, and other scenarios.

[0003] Currently, the control strategies for three-phase four-arm inverters are mostly based on traditional PI control algorithms, using single-loop or dual-loop control structures to regulate voltage and current. However, in actual operation, grid voltage distortion, load fluctuations, and the randomness of photovoltaic array output power often lead to a mixture of positive-sequence, negative-sequence, and zero-sequence components in the inverter output current. Traditional single-loop control struggles to effectively separate these components, easily causing problems such as output voltage harmonic distortion and dynamic response lag. Especially when the three-phase load is unbalanced, the presence of negative-sequence components exacerbates current fluctuations and reduces energy conversion efficiency.

[0004] While existing dual-loop control strategies offer improvements in dynamic response, their controller parameter design largely relies on empirical trial-and-error methods, lacking systematic parameter matching rules. This leads to coupling interference between the positive-sequence and negative-sequence control loops. When the grid experiences sudden voltage spikes, drops, or load changes, the control loop parameters struggle to adapt quickly, resulting in insufficient dynamic regulation and even oscillations. Furthermore, traditional control methods have limited ability to suppress negative-sequence components. Under grid imbalance conditions, the harmonic content of the output current often exceeds grid standards, affecting grid compatibility.

[0005] Existing verification methods for control strategies mostly focus on static performance testing, neglecting performance evaluation under dynamic operating conditions. In the actual operation of photovoltaic power generation systems, rapid changes in irradiance and ambient temperature can cause drastic fluctuations in inverter input power. Traditional control strategies struggle to maintain stable output characteristics under such dynamic scenarios, easily leading to problems such as large overshoot and long settling times.

[0006] With the development of power electronics technology, the switching frequency of inverters is constantly increasing, highlighting the growing contradiction between the computational complexity and real-time requirements of traditional control algorithms. Existing control strategies often fail to adequately consider the dynamic response differences between positive-sequence and negative-sequence components during parameter optimization, leading to decreased controller adaptability under complex operating conditions. These problems not only restrict the grid-connected performance of photovoltaic power generation systems but may also shorten equipment lifespan and increase operation and maintenance costs. Therefore, developing a control strategy that can effectively separate positive-sequence and negative-sequence components, optimize PI parameter matching, and improve dynamic and static performance has become an urgent need for the current development of photovoltaic power generation inverter technology. Summary of the Invention

[0007] The purpose of this invention is to provide a sequential dual-loop PI control optimization method for a three-phase four-arm photovoltaic inverter to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, this invention provides a method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter, the method comprising:

[0009] S1. In the photovoltaic power generation system operating environment, obtain inverter topology parameters and control strategy requirements information;

[0010] S2. Based on mathematical modeling tools, coordinate transformation theory and system identification methods, a single-loop control model and a sequence decoupling model are constructed according to the inverter topology parameters.

[0011] S3. Based on the direct flow extraction strategy, the signal is separated according to the single-loop control model and the sequence decoupling model to obtain the positive sequence control component and the negative sequence control component.

[0012] S4. Based on the dual-loop parameter matching rules and dynamic response constraints, the controller parameters are adjusted according to the positive-sequence control components and negative-sequence control components to obtain optimized PI parameters.

[0013] S5. Based on the optimized PI parameters, dynamic performance optimization is performed according to the single-loop control model, the sequential decoupling model, and the inverter topology parameters to obtain the dynamic optimization results.

[0014] S6. Based on the static verification method, the performance is verified by comparing historical data based on the single-loop control model, the sequence decoupling model and the control strategy requirements, and the static verification results are obtained.

[0015] S7. Based on the dynamic optimization results and static verification results, the scheme is integrated to obtain the optimized scheme of the sequence dual-loop PI control.

[0016] S8. Generate a control strategy implementation guide based on the optimized scheme of the sequenced dual-loop PI control.

[0017] Preferably, the construction of a single-loop control model and a sequence decoupling model based on inverter topology parameters, using mathematical modeling tools, coordinate transformation theory, and system identification methods, includes:

[0018] S21. Based on the inverter topology parameters, obtain the main circuit topology model and load characteristic model;

[0019] S22. Perform time-domain characteristic analysis on the main circuit topology model to obtain the differential expressions for voltage and current.

[0020] S23. Based on the main circuit topology model, convert the differential expression into a frequency domain transfer function; determine the frequency domain transfer function as the basis of the single-loop control model; determine the main circuit topology model as the boundary of the single-loop control model; determine the single-loop control model based on the single-loop control model basis and the single-loop control model boundary.

[0021] S24. Perform symmetric component decomposition on the load characteristic model to obtain the positive-sequence load component and the negative-sequence load component.

[0022] S25. Based on the load characteristic model, the positive-sequence load component is determined as the positive-sequence decoupling starting point; the negative-sequence load component is determined as the negative-sequence decoupling starting point; and the sub-sequence decoupling nodes are determined based on the positive-sequence decoupling starting point and the negative-sequence decoupling starting point.

[0023] S26. Perform zero-sequence component suppression analysis on the load characteristic model to obtain the zero-sequence suppression constraint conditions; determine the decoupling node as the constraint propagation starting point; determine the zero-sequence suppression constraint conditions as the constraint propagation ending point; determine the constraint propagation path based on the constraint propagation starting point and the constraint propagation ending point.

[0024] S27. Based on the single-loop control model, the sequential decoupling nodes, and the constraint transmission path, the main circuit topology model is decoupled and integrated to obtain the sequential decoupling model.

[0025] S28. Based on the rotation matrix operation of coordinate transformation theory, the frequency domain characteristics of the sequence decoupling model are verified to obtain the sequence decoupling model.

[0026] Preferably, the step of using a direct current extraction strategy to separate signals based on a single-loop control model and a sequence decoupling model to obtain positive-sequence control components and negative-sequence control components includes:

[0027] S31. Perform signal characteristic checks based on the single-loop control model to obtain the first separation signal;

[0028] S32. Based on the sequence decoupling model, perform signal integrity analysis on the single-loop control model to obtain the second separation signal;

[0029] S33. Based on the single-loop control model, identify and analyze the feature extraction points of the DC flow extraction to obtain the extraction constraints and feature extraction statements;

[0030] S34. Based on the sequence decoupling model, identify the signal separation points protected by the extraction strategy to obtain the separation target signal;

[0031] S35. Based on the principle of double second-order generalized integrals, construct a direct flow extraction model according to the feature extraction statement and the separated target signal;

[0032] S36. Using the direct flow extraction model, compare the extraction constraints to perform an extraction delay check and obtain the first separation component;

[0033] S37. Based on the signal transmission path of the sequence decoupling model, component distortion is checked according to the extraction constraints and the DC extraction model to obtain the second separated component.

[0034] Preferably, the step of adjusting the controller parameters based on the dual-loop parameter matching rules and dynamic response constraints, according to the positive-sequence control components and the negative-sequence control components, to obtain optimized PI parameters includes:

[0035] S41. Obtain the reference variable for the outer loop control input based on the positive sequence control component; determine the reference variable as the parameter adjustment source;

[0036] S42. Based on the optimized PI parameters, and according to the sequence decoupling model, parameter transfer analysis is performed on the single-loop control model to obtain the parameter transfer path;

[0037] S43. Based on the parameter passing path, perform dynamic response entry checks on the optimized PI parameters to obtain a list of parameter adjustments with dynamic response entry points.

[0038] S44. Based on the parameter adjustment list, track steady-state variables according to the order decoupling model to obtain constrained steady-state variables;

[0039] S45. Determine the optimal PI parameters based on the parameter adjustment list and the constrained steady-state variables.

[0040] Preferably, the static verification method, based on the single-loop control model, the sequential decoupling model, and control strategy requirements, performs performance verification using a historical data comparison algorithm to obtain static verification results, including:

[0041] S61. Based on the single-loop control model and the sequence decoupling model, the static verification method is used to construct the data tuples and obtain the data quintuples.

[0042] S62. Based on the historical data chain, construct a set of verification indicators according to the data quintuple;

[0043] S63. Based on the control strategy requirements and the data quintuple, convert each verification indicator in the verification indicator set into a comparison sequence to obtain the initial verification pool.

[0044] Preferably, the step of constructing a set of verification indicators based on historical data chains and data quintuples includes:

[0045] S621. Obtain indicator information based on the data quintuple; set the total number of indicators in the indicator information to M, the number of the current indicator to j, and let j=1; verify the indicator set to T, which is initially an empty set; retrieve the historical data chain set to Recorded, which is initially an empty set; the historical data chain is a quadruple representing the performance correlation between indicators.

[0046] S622. Determine if j is greater than M. If j is greater than M, proceed to step S627. If j is less than or equal to M, proceed to step S623.

[0047] S623. Based on the verification indicator set T, determine whether the current indicator is in T according to the indicator information. If the current indicator is in T, let j = j + 1 and execute step S622. If indicator j is not in T, create a verification indicator sequence ind based on the current indicator.

[0048] S624. Based on the indicator information, retrieve the performance correlation according to the verification indicator sequence ind to obtain a new set of historical data chains;

[0049] S625. Based on the retrieved historical data chain set Recorded, determine whether the new historical data chain set is a subset of Recorded. If the new historical data chain set is a subset of Recorded, update T according to ind, let j = j + 1, and execute step S622. If the new historical data chain set is not a subset of Recorded, execute step S626.

[0050] S626. Update ind according to the new historical data chain set; update Recorded using the new historical data chain set, and proceed to step S624.

[0051] S627, Output the set of verification metrics T.

[0052] Preferably, the step of integrating the schemes based on dynamic optimization results and static verification results to obtain a sequence dual-loop PI control optimization scheme includes:

[0053] S71. Based on the random perturbation strategy and the gray box verification strategy, the schemes are verified alternately according to the dynamic optimization results to obtain a verification scheme pool.

[0054] S72. Based on optimization and integration technology, a comprehensive performance analysis is performed according to the verification scheme pool and control strategy requirements to obtain integrated feedback information and scheme matching degree.

[0055] S73. Based on the sequence decoupling model, the verification scheme pool is updated according to the integrated feedback information and the scheme matching degree to obtain the updated scheme pool.

[0056] S74. Based on the control strategy requirement information and according to the updated scheme pool, the final scheme of the single-loop control model is determined to obtain the optimized scheme of the sequenced double-loop PI control.

[0057] Preferably, the step of performing a dynamic response entry check on the optimized PI parameters based on the parameter transmission path to obtain a list of parameter adjustments with dynamic response entry points includes:

[0058] S431. Based on the parameter passing path, obtain the main parameter passing segment and the secondary parameter passing segment;

[0059] S432. Perform response time analysis on the main parameter transfer segment to obtain the first response threshold;

[0060] S433. Based on the main parameter transfer segment, the first response threshold is determined as the entry check start point; the secondary parameter transfer segment is determined as the entry check end point; the dynamic response entry is determined based on the entry check start point and the entry check end point.

[0061] S434. Perform overshoot analysis on the secondary parameter transfer segment to obtain the second response threshold.

[0062] S435. Based on the secondary parameter transfer segment, the dynamic response entry point is determined as the threshold comparison start point; the second response threshold is determined as the threshold comparison end point; the response entry constraint is determined based on the threshold comparison start point and the threshold comparison end point.

[0063] S436. Based on the response entry constraint, perform segmented verification on the optimized PI parameters to obtain a list of parameter adjustments with dynamic response entry points.

[0064] Preferably, the step of determining the final scheme for the single-loop control model based on the control strategy requirement information and according to the updated scheme pool to obtain the sequential dual-loop PI control optimization scheme includes:

[0065] S741. Based on the updated solution pool, obtain the set of candidate solutions and performance evaluation metrics;

[0066] S742. Perform steady-state accuracy evaluation on the candidate scheme set to obtain the first evaluation result;

[0067] S743. Based on the candidate scheme set, the first evaluation result is determined as the starting point for scheme screening; the performance evaluation index is determined as the ending point for scheme screening; and the steady-state screening conditions are determined based on the starting point and the ending point for scheme screening.

[0068] S744. Perform dynamic follow-up evaluation on the candidate solution set to obtain the second evaluation result;

[0069] S745. Based on the candidate scheme set, the steady-state screening conditions are determined as the starting point for comprehensive screening; the second evaluation result is determined as the ending point for comprehensive screening; and the dynamic screening conditions are determined based on the comprehensive screening starting point and the comprehensive screening ending point.

[0070] S746. Based on dynamic screening conditions, the candidate scheme set is finally screened to obtain the optimized scheme of the sequenced double-loop PI control.

[0071] Preferably, the rotation matrix operation based on coordinate transformation theory for frequency domain characteristic verification of the sequence decoupling model includes:

[0072] S281. Obtain the positive-order component transfer function and the negative-order component transfer function based on the order decoupling model.

[0073] S282. Perform frequency response analysis on the positive sequence component transfer function to obtain the positive sequence amplitude-frequency characteristic curve;

[0074] S283. Based on the sequence decoupling model, the positive sequence amplitude-frequency characteristic curve is determined as the starting point of frequency domain verification; the negative sequence component transfer function is determined as the ending point of frequency domain verification; the frequency domain verification path is determined based on the starting point and the ending point of frequency domain verification.

[0075] S284. Perform phase delay analysis on the transfer function of the negative sequence component to obtain the negative sequence phase frequency characteristic curve;

[0076] S285. Based on the rotation matrix operation rules, characteristic matching verification is performed according to the positive sequence amplitude frequency characteristic curve and the negative sequence phase frequency characteristic curve to obtain the frequency domain characteristic verification results of the sequence decoupling model.

[0077] Compared with the prior art, the present invention has the following beneficial effects:

[0078] By constructing a sequence-decoupling model and a dual-loop control structure, precise separation and independent adjustment of positive-sequence and negative-sequence control components are achieved, solving the problem of coupling interference between positive and negative-sequence components in traditional control strategies. When there is grid voltage distortion or load imbalance, the positive-sequence and negative-sequence components in the mixed signal can be effectively separated through a DC extraction strategy. This allows the controller to make differentiated adjustments based on the characteristics of different sequence components, avoiding the decrease in adjustment accuracy caused by component mixing in traditional single-loop control, and ensuring that the inverter can still output high-quality power under complex operating conditions.

[0079] PI parameter optimization based on dual-loop parameter matching rules and dynamic response constraints breaks away from the traditional parameter design mode that relies on trial and error based on experience. By systematically analyzing the dynamic characteristics of positive-sequence and negative-sequence control components, a quantitative basis for parameter adjustment is established, enabling the controller parameters to adapt to the response requirements of different sequence components. During load abrupt changes or grid voltage fluctuations, the optimized PI parameters can accelerate the system's dynamic adjustment speed, reduce overshoot, and improve the inverter's ability to quickly adapt to changes in operating conditions, avoiding the dynamic response lag problem caused by insufficient adaptability in traditional parameter design.

[0080] The combined approach of dynamic performance optimization and static verification ensures the reliability of the control strategy from multiple dimensions. Dynamic performance optimization, based on a single-loop control model, a sequential decoupling model, and inverter topology parameters, adjusts the system's response characteristics during transient processes, improving the inverter's stability under scenarios of rapid power changes. Static verification, through historical data comparison algorithms, verifies the performance of the control strategy during steady-state operation, ensuring that indicators such as harmonic content and voltage deviation of the output power meet grid standards, thus solving the problem of balancing dynamic and static performance in traditional control strategies.

[0081] The control strategy implementation guide generated by the sequence dual-loop PI control optimization scheme provides clear operational guidelines for practical engineering applications, reducing the difficulty of technology implementation. Compared with traditional control methods, this method does not require complex hardware modifications; it can significantly improve inverter performance simply by optimizing the control algorithm. While ensuring improved system stability, it reduces downtime caused by equipment failures, providing strong support for the efficient operation of photovoltaic power generation systems from multiple aspects. Attached Figure Description

[0082] Figure 1 This is a schematic diagram illustrating the working principle of the sequential dual-loop PI control optimization method for a three-phase four-arm photovoltaic inverter described in this invention.

[0083] Figure 2 Flowcharts for constructing single-loop control models and sequence decoupling models;

[0084] Figure 3 A flowchart for separating positive and negative sequence control components from a signal;

[0085] Figure 4 A flowchart for adjusting controller parameters to obtain optimized PI parameters;

[0086] Figure 5 This is a flowchart for checking the dynamic response entry point. Detailed Implementation

[0087] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0088] Please see Figure 1-Figure 5 This invention provides an optimization method for sequential dual-loop PI control in a three-phase four-arm photovoltaic inverter, the method comprising:

[0089] S1. In the photovoltaic power generation system operating environment, obtain the inverter topology parameters and control strategy requirements. The inverter topology parameters include hardware parameters such as the number of bridge arms, switching device models, filter inductance value, filter capacitor value, and DC side voltage; the control strategy requirements include performance indicators such as output voltage stability accuracy requirements, dynamic response speed requirements, and load disturbance immunity requirements.

[0090] S2. Based on mathematical modeling tools, coordinate transformation theory, and system identification methods, a single-loop control model and a sequence decoupling model are constructed according to the inverter topology parameters. Equivalent circuit analysis of the inverter main circuit is performed using mathematical modeling tools. The three-phase AC quantities are converted into DC quantities using coordinate transformation theory for easier control. The unknown parameters in the model are determined using system identification methods, thus completing the construction of the single-loop control model and the sequence decoupling model.

[0091] S3. Based on the DC extraction strategy, signal separation is performed according to the single-loop control model and the sequence decoupling model to obtain the positive-sequence control component and the negative-sequence control component. The DC extraction strategy uses a specific filtering algorithm to extract the control-related DC component from the mixed voltage and current signal, and then, based on the characteristics of the single-loop control model and the sequence decoupling model, separates the signal into two control components: positive-sequence and negative-sequence.

[0092] S4. Based on the dual-loop parameter matching rule and dynamic response constraints, the controller parameters are adjusted according to the positive-sequence control components and the negative-sequence control components to obtain optimized PI parameters. The dual-loop parameter matching rule ensures that the PI parameters of the inner and outer loops cooperate with each other, while the dynamic response constraints limit the system's response time and overshoot range when the load changes or voltage fluctuates. By repeatedly adjusting the parameters, the system is made to meet these conditions.

[0093] S5. Based on optimized PI parameters, dynamic performance optimization is performed according to the single-loop control model, the sequential decoupling model, and the inverter topology parameters to obtain dynamic optimization results. Dynamic performance optimization mainly targets the adjustment of dynamic processes such as the system's step response and ramp response. Through simulation and actual testing, the dynamic performance of the system under different operating conditions is optimized.

[0094] S6. Based on the static verification method, performance verification is performed using a historical data comparison algorithm based on the single-loop control model, the sequential decoupling model, and the control strategy requirements, to obtain the static verification results. The static verification method analyzes the output voltage and current waveforms of the system during steady-state operation, calculates its harmonic content, voltage deviation, and other indicators, and compares them with historical data to verify whether the static performance of the system meets the requirements.

[0095] S7. Based on the dynamic optimization results and static verification results, the scheme is integrated to obtain the optimized scheme for the sequence dual-loop PI control. The effective information obtained from dynamic optimization and static verification is combined to adjust and improve the control scheme, ensuring that the scheme can meet both dynamic performance requirements and static performance stability.

[0096] S8. Generate a control strategy implementation guide based on the optimized scheme for the sequence dual-loop PI control. The implementation guide details the specific operation steps, parameter setting methods, system debugging procedures, etc., providing clear guidance for engineers in practical applications.

[0097] Example 1:

[0098] S21. Based on the inverter topology parameters, obtain the main circuit topology model and load characteristic model. The inverter topology parameters include the number of bridge arms, the model and parameters of the switching transistors in each bridge arm, the values ​​of the filter inductor and capacitor, the DC-side capacitor capacity, and the line resistance. The main circuit topology model is an abstract representation of the main circuit structure of a three-phase four-bridge-arm inverter, encompassing the connection methods of the four bridge arms, the connection relationship between the filter circuit and the bridge arms, and the connection path between the DC-side power supply and the bridge arms. These topology parameters allow for the accurate construction of a model that accurately reflects the physical structure of the main circuit. The load characteristic model is constructed based on the actual load types that may be connected to the photovoltaic power generation system, including the load's impedance characteristics, power factor, whether it is a nonlinear load, and the load's variation under different operating conditions. For example, when the load is a motor, the differences in its characteristics during startup, operation, and braking stages must be reflected in the model.

[0099] S22. Perform time-domain characteristic analysis on the main circuit topology model to obtain differential expressions for voltage and current. Within the time domain, analyze the voltage and current changes of each component in the main circuit at different times. Based on Kirchhoff's voltage and current laws, and combined with the time-domain characteristics of components such as inductors and capacitors (e.g., inductor voltage is proportional to the rate of change of current, and capacitor current is proportional to the rate of change of voltage), derive differential equations describing the changes in node voltage and branch current over time in the main circuit. These differential expressions can intuitively reflect the dynamic changes in voltage and current in the main circuit during transient processes. For example, the sudden changes in voltage and current in the circuit at the instant the switch is turned on or off can be represented by differential expressions.

[0100] S23. Based on the main circuit topology model, the differential expression is converted into a frequency domain transfer function; the frequency domain transfer function is used as the basis for the single-loop control model; the main circuit topology model is used as the boundary of the single-loop control model; the single-loop control model is determined based on the single-loop control model basis and the single-loop control model boundary. The Laplace transform is used to convert the voltage and current differential expressions in the time domain into transfer functions in the frequency domain. The transfer function describes the output response characteristics of the main circuit under different frequency input signals and is the core part of the single-loop control model, i.e., the basis of the single-loop control model. The circuit structure and component parameter ranges defined by the main circuit topology model constitute the boundary conditions of the single-loop control model. For example, the maximum switching frequency of the bridge arm switch, the minimum and maximum values ​​of the filter inductor, etc., limit the applicability of the model. By combining the frequency domain transfer function and the boundary conditions of the main circuit topology model, a complete single-loop control model is constructed. This model can be used to analyze the stability, dynamic response, and other characteristics of the system under the single-loop control strategy.

[0101] S24. Perform symmetrical component decomposition on the load characteristic model to obtain the positive-sequence load component and the negative-sequence load component. The symmetrical component decomposition method decomposes the three-phase unbalanced load characteristics into three components: positive-sequence, negative-sequence, and zero-sequence. In the control of a three-phase four-arm inverter, the focus is on the positive-sequence and negative-sequence load components. The positive-sequence load component is the symmetrical component with the same phase sequence as the power supply, while the negative-sequence load component is the symmetrical component with the opposite phase sequence. Through decomposition, the complex unbalanced load characteristics can be simplified to the superposition of two symmetrical components, facilitating separate control. For example, when the three-phase load is unbalanced, the positive-sequence and negative-sequence components causing the asymmetry can be clearly distinguished through symmetrical component decomposition.

[0102] S25. Based on the load characteristic model, the positive-sequence load component is determined as the positive-sequence decoupling starting point; the negative-sequence load component is determined as the negative-sequence decoupling starting point; and the sub-sequence decoupling node is determined based on the positive-sequence and negative-sequence decoupling starting points. In the load characteristic model, the point of application of the positive-sequence load component is set as the starting point of positive-sequence decoupling, from which independent control and decoupling processing of the positive-sequence component begins; similarly, the point of application of the negative-sequence load component is set as the starting point of negative-sequence decoupling. The sub-sequence decoupling node is the intersection point of the positive-sequence and negative-sequence decoupling starting points in the control model, and it is also the key node for realizing the separate control of the positive-sequence and negative-sequence components. Through this node, the positive-sequence and negative-sequence control paths can be distinguished, ensuring that the control of the two does not interfere with each other.

[0103] S26. Perform zero-sequence component suppression analysis on the load characteristic model to obtain zero-sequence suppression constraints; determine the decoupling node as the constraint propagation starting point; determine the zero-sequence suppression constraints as the constraint propagation endpoint; determine the constraint propagation path based on the constraint propagation starting point and endpoint. Zero-sequence components in a three-phase four-arm inverter may cause neutral point shift and other problems, requiring suppression. Analyze the causes, magnitude, and impact on the system of zero-sequence components on the load characteristic model, and formulate zero-sequence suppression constraints, such as the maximum allowable value of zero-sequence voltage and the limitation range of zero-sequence current. The decoupling node serves as the starting point for constraint propagation, transmitting the zero-sequence suppression constraints along a specific path to various relevant components of the system. This path is the constraint propagation path, ensuring the effective implementation of zero-sequence suppression measures in the system.

[0104] S27. Based on the single-loop control model, the sequence decoupling nodes, and the constraint propagation path, the main circuit topology model is decoupled and integrated to obtain a sequence decoupling model. Using the single-loop control model as the basic framework, and considering the positions of the sequence decoupling nodes and the requirements of the constraint propagation path, the main circuit topology model is adjusted and integrated. During the integration process, it is necessary to ensure that the positive-sequence and negative-sequence components can be independently propagated and controlled through the sequence decoupling nodes, while the zero-sequence suppression constraint can function in the circuit through the constraint propagation path. Through this integration, the model acquires the function of sequence decoupling, meaning it can control the positive-sequence and negative-sequence components separately without affecting each other.

[0105] S28. Based on coordinate transformation theory, rotation matrix operations are used to verify the frequency domain characteristics of the sequence decoupling model, thus obtaining the sequence decoupling model. The rotation matrix in coordinate transformation theory can convert variables in a three-phase coordinate system into variables in a two-phase rotating coordinate system or a stationary coordinate system, facilitating the analysis of the system's frequency domain characteristics. Rotation matrix operations are used to perform coordinate transformations on the sequence decoupling model, obtaining model expressions in different coordinate systems, and then analyzing the model's amplitude-frequency and phase-frequency characteristics in the frequency domain. By comparing the theoretically calculated frequency domain characteristics with the actual expected characteristics, the response of the sequence decoupling model at different frequencies is verified to meet the design requirements. If deviations exist, the model is adjusted until it passes the frequency domain characteristic verification, ultimately determining the sequence decoupling model.

[0106] Example 2:

[0107] S31. Based on the single-loop control model, perform signal characteristic checks to obtain the first separation signal. The single-loop control model includes the transmission patterns of inverter output voltage, current, and other signals. By analyzing the frequency components, amplitude variation range, phase relationship, and harmonic content of these signals, identify the effective signal components that conform to the single-loop control logic. For example, for three-phase AC signals, check whether they contain fundamental and major harmonic components, and screen out those signals that accurately reflect the system operating state and are suitable for single-loop control processing. These signals are then determined as the first separation signal.

[0108] S32. Based on the sequence decoupling model, signal integrity analysis is performed on the single-loop control model to obtain the second separated signal. The sequence decoupling model requires the signal to contain complete information of both positive and negative sequence components to achieve independent control of the two. From the perspective of sequence decoupling, it is checked whether the signal in the single-loop control model has lost key features of the positive or negative sequence components during transmission, such as whether the amplitude of the negative sequence component has been excessively attenuated or whether the phase of the positive sequence component has undergone abnormal shift. Through this integrity analysis, it is ensured that the extracted signal can fully reflect the actual situation of the positive and negative sequence components in the system, thereby obtaining the second separated signal.

[0109] S33. Based on the single-loop control model, identify and analyze the feature extraction points for DC output extraction to obtain extraction constraints and feature extraction statements. The feature extraction points for DC output extraction are the locations where DC characteristics are most prominent in the signal. In the single-loop control model, these points are typically located after the filtering stage or at the feedback signal acquisition point. Analyze these feature extraction points to determine the constraints that must be followed during DC output extraction, such as ensuring that the signal delay during extraction does not exceed a certain time and that the extracted DC output error is controlled within a specific range. Simultaneously, combine the signal processing logic of the single-loop control model to form feature extraction statements that specifically describe how to identify and extract DC characteristics from the feature extraction points, such as "At the filter capacitor voltage signal acquisition point, extract the DC component with a frequency of 0Hz, ignoring AC components above twice the fundamental frequency."

[0110] S34. Based on the sequence-based decoupling model, identify the signal separation points protected by the extraction strategy to obtain the target signals to be separated. The signal separation points protected by the extraction strategy refer to key nodes that significantly affect the separation accuracy of positive and negative sequence components during the signal separation process. The signal quality of these nodes directly determines the accuracy of the separation results. In the sequence-based decoupling model, these nodes may be located at the branches of positive and negative sequence components or at the input ports of the decoupling algorithm. By identifying these nodes, the target signals that need to be separated from these nodes are determined. For example, signals related to positive sequence voltage and current are separated from positive sequence branch nodes, and signals related to negative sequence voltage and current are separated from negative sequence branch nodes. These signals are the target signals to be separated.

[0111] S35. Based on the principle of dual second-order generalized integrators, a DC component extraction model is constructed according to the feature extraction statement and the target signal to be separated. The dual second-order generalized integrator principle has good filtering characteristics and DC component extraction capability, and can accurately extract the DC component in the signal while suppressing AC interference. Based on the extraction rules described in the feature extraction statement, such as extraction frequency and filtering bandwidth, and the characteristics of the target signal to be separated, such as amplitude range and dynamic change rate, the parameters of the dual second-order generalized integrator are designed, including resonant frequency and damping coefficient. By integrating these parameters into the model, a DC component extraction model that can specifically extract the DC component in the target signal is constructed.

[0112] S36. Using a direct current extraction model, perform an extraction delay check by comparing the extraction constraints to obtain the first separated component. Apply the direct current extraction model to actual signal processing to extract the direct current from the target signal. During this process, monitor the time delay of the extraction process and compare it with the maximum delay time specified in the extraction constraints. If the delay time meets the constraint requirements, the extracted direct current component is taken as the first separated component; if it does not meet the requirements, adjust the parameters of the direct current extraction model, such as reducing the order of the filtering stage, to reduce the delay until the constraint conditions are met, and then determine the first separated component.

[0113] S37. Based on the signal transmission path of the sequence decoupling model, component distortion checks are performed according to the extraction constraints and the DC extraction model to obtain the second separated component. The signal transmission path of the sequence decoupling model describes the transmission routes of the positive and negative sequence components in the system. Along these paths, it is checked whether the signal processed by the DC extraction model is distorted, such as excessive amplitude attenuation or phase distortion. At the same time, the extracted components are evaluated in conjunction with the signal distortion limits in the extraction constraints. Components that meet the distortion requirements are determined as the second separated component. If there is excessive distortion, it is corrected by adjusting the filtering parameters or signal compensation method of the DC extraction model until the second separated component that meets the requirements is obtained. Through further processing of the first and second separated components, the positive sequence control component and the negative sequence control component are finally obtained.

[0114] Example 3:

[0115] S41. Obtain the reference variables for the outer loop control input based on the positive sequence control components; determine the reference variables as the source of parameter adjustment. The positive sequence control components contain the main control information of the system under symmetrical operation. By analyzing these components, the reference variables required for outer loop control are extracted. These variables typically include the desired output voltage amplitude, frequency, and active and reactive power commands. For example, in voltage outer loop control, the reference variable can be the desired effective value of the output phase voltage; in power outer loop control, the reference variable can be the set active and reactive power values. These reference variables directly determine the controller's adjustment target and are therefore determined as the source of parameter adjustment. Subsequent parameter adjustment processes will revolve around these reference variables to ensure that the system output can track these reference values.

[0116] S42. Based on optimized PI parameters and according to the sequence decoupling model, parameter transfer analysis is performed on the single-loop control model to obtain the parameter transfer path. Optimized PI parameters include the proportional coefficient and integral time constant. These parameters need to be transferred between the sequence decoupling model and the single-loop control model to achieve separate control of the positive-sequence and negative-sequence components. By analyzing the structure of the positive and negative-sequence channels in the sequence decoupling model and the connection methods of the voltage and current loops in the single-loop control model, the links and paths that the PI parameters pass through from input to output are determined. For example, positive-sequence PI parameters may first pass through the positive-sequence decoupling module and then enter the inner current loop and outer voltage loop of the single-loop control model; negative-sequence PI parameters pass through the negative-sequence decoupling module and then enter the corresponding control loop. Through this analysis, the parameter transfer route in the entire control system is clearly identified, forming the parameter transfer path.

[0117] S43. Based on the parameter propagation path, perform dynamic response entry point checks on the optimized PI parameters to obtain a parameter adjustment list containing dynamic response entry points. There are key nodes along the parameter propagation path; changes in the parameters at these nodes directly affect the system's dynamic response, i.e., dynamic response entry points. Check these entry points to determine which optimized PI parameters are effective at them. For example, changes in the PI parameters of the inner current loop directly affect the system's tracking speed of current commands, thus constituting a dynamic response entry point; the PI parameters of the outer voltage loop primarily affect the system's steady-state accuracy, but can also affect the dynamic response during voltage surges, similarly potentially becoming dynamic response entry points. Compile these PI parameters related to dynamic response entry points into a list, i.e., a parameter adjustment list.

[0118] S44. Based on the parameter adjustment list, track steady-state variables using the sequential decoupling model to obtain constrained steady-state variables. Steady-state variables include the steady-state error of the output voltage, the harmonic content of the output current, and the steady-state values ​​of active and reactive power. Based on the PI parameters in the parameter adjustment list, simulate the steady-state operation of the system under different parameter values ​​in the sequential decoupling model, tracking the changes in these steady-state variables. Analyze which steady-state variables change significantly with the parameters in the parameter adjustment list; these variables are the constrained steady-state variables. For example, when adjusting the integral time constant of the outer voltage loop, the steady-state error of the output voltage will change accordingly; therefore, the output voltage steady-state error is a constrained steady-state variable. Similarly, when adjusting the proportional gain of the inner current loop, the harmonic content of the current may change; therefore, the current harmonic content may also become a constrained steady-state variable.

[0119] S45. Determine the optimized PI parameters based on the parameter adjustment list and the constrained steady-state variables. In determining the optimized PI parameters, it is necessary to comprehensively consider the value range of the parameters in the parameter adjustment list and the allowable variation range of the constrained steady-state variables. For each parameter, first set an initial value, then simulate its impact on the dynamic response along the parameter propagation path, and observe the changes in the constrained steady-state variables through a sequence decoupling model. If the parameter value causes the dynamic response to fail to meet requirements or the constrained steady-state variables to exceed the allowable range, adjust the parameter value and repeat the above process. During this process, the following formula can be used to assist in calculating the degree of influence of parameter adjustment on the steady-state variables:

[0120]

[0121] in, This indicates the sensitivity of a constrained steady-state variable to the proportionality parameter. This represents the change in a constrained steady-state variable. This represents the change in the proportional parameter. By calculating the sensitivity, the degree of impact of parameter adjustments on the steady-state variable can be determined, allowing for more targeted parameter optimization. For example, parameters with high sensitivity require fine-tuning to avoid large fluctuations in the steady-state variable; parameters with low sensitivity can be adjusted within a wider range to optimize the dynamic response. Through iterative iteration, the PI parameter that satisfies both the dynamic response requirements and keeps the constrained steady-state variable within the allowable range is finally determined—that is, the optimized PI parameter. This process requires combining the characteristics of the sequence decoupling model and the single-loop control model to ensure that the parameter exerts good control effects in both positive and negative sequence channels, achieving overall system optimization.

[0122] Example 4:

[0123] S61. Based on the single-loop control model and the sequence decoupling model, a static verification method is used to construct data tuples, obtaining a five-tuple data set. The static verification method focuses on the system's performance during steady-state operation. By collecting the voltage and current signals output from the single-loop control model and the positive and negative sequence component signals separated from the sequence decoupling model, combined with the load power data during actual operation, a data tuple containing five elements is formed. For example, under a certain steady-state condition, the collected three-phase output voltage RMS value is 220V, the three-phase output current RMS value is 5A, the positive sequence voltage component amplitude is 218V, the negative sequence voltage component amplitude is 2V, and the load active power is 1100W. These five data points combined constitute a data tuple, fully reflecting the static operating characteristics of the system under this condition.

[0124] S62. Based on the historical data chain, construct a set of verification indicators according to the data quintuples. S621. Obtain indicator information based on the data quintuples; set the total number of indicators to M, the current indicator number to j, and let j=1; the set of verification indicators to T, initially empty; the retrieved historical data chain set to Recorded, initially empty; the historical data chain is a quadruple representing the performance correlation between indicators. Indicator information includes voltage deviation rate, current distortion rate, positive-sequence component ratio, negative-sequence component ratio, power factor, etc. Each indicator reflects system performance from different dimensions. The quadruple of the historical data chain contains two related indicators, the correlation strength value between the indicators, and the correlation type (e.g., positive or negative correlation). For example, "voltage deviation rate - current distortion rate - 0.3 - positive correlation" indicates a weak positive correlation between the two indicators.

[0125] S622. Determine if j is greater than M. If j is greater than M, proceed to step S627. If j is less than or equal to M, proceed to step S623. Assume the total number of indicators M is 5, and the initial j=1. At this time, j is less than M, so proceed to step S623.

[0126] S623. Based on the verification indicator set T, determine whether the current indicator is in T according to the indicator information. If the current indicator is in T, let j = j + 1 and execute step S622; if indicator j is not in T, create a verification indicator sequence ind based on the current indicator. If the verification indicator set T is initially empty, and the current indicator j = 1 is "voltage deviation rate", which is not in T, then create a verification indicator sequence ind containing "voltage deviation rate".

[0127] S624. Based on the indicator information, retrieve the performance correlation according to the verification indicator sequence ind to obtain a new set of historical data chains. Starting with "voltage deviation rate" in the verification indicator sequence ind, retrieve the correlation related to this indicator in the historical data to obtain a new set of historical data chains, such as data chains including "voltage deviation rate - positive sequence component proportion - 0.8 - negative correlation" and "voltage deviation rate - power factor - 0.2 - negative correlation".

[0128] S625. Based on the retrieved historical data chain set Recorded, determine whether the new historical data chain set is a subset of Recorded. If the new historical data chain set is a subset of Recorded, update T according to ind, let j = j + 1, and proceed to step S622. If the new historical data chain set is not a subset of Recorded, proceed to step S626. Since Recorded is initially empty, the new historical data chain set is clearly not a subset of it, therefore proceed to step S626.

[0129] S626. Update ind based on the new historical data chain set; update Recorded using the new historical data chain set, and execute step S624. Add the "positive sequence component ratio" and "power factor" involved in the new historical data chain to the verification index sequence ind. At this time, ind contains "voltage deviation rate, positive sequence component ratio, and power factor"; at the same time, store the new historical data chain set into Recorded, and execute step S624 again to retrieve the new historical data chain related to the three indicators in ind.

[0130] Repeat steps S624 to S626, continuously expanding the verification index sequences ind and Recorded. For example, if subsequent data chains such as "positive sequence component percentage - negative sequence component percentage - 0.9 - negative correlation" and "power factor - current distortion rate - 0.4 - positive correlation" are retrieved, add "negative sequence component percentage" and "current distortion rate" to ind until ind contains all 5 indices. At this point, j=5, execute step S622 again, j equals M, and proceed to step S627.

[0131] S627, Output verification index set T. At this time, the verification index set T includes "voltage deviation rate, current distortion rate, positive sequence component ratio, negative sequence component ratio, and power factor", which fully covers the key indicators for evaluating the static performance of the system.

[0132] S63. Based on the control strategy requirements and the data quintuple, each verification indicator in the verification indicator set is converted into a comparison sequence to obtain the initial verification pool. The control strategy requirements specify the allowable ranges for each indicator, such as voltage deviation rate ≤ ±2%, current distortion rate ≤ 5%, positive sequence component percentage ≥ 95%, negative sequence component percentage ≤ 5%, and power factor ≥ 0.95. For each verification indicator, the actual values ​​collected in the data quintuple are arranged in chronological order with the required values ​​to form a comparison sequence. For example, the comparison sequence for "voltage deviation rate" might be "1.2% / 1.5% / 1.0% vs ±2%", and the comparison sequence for "positive sequence component percentage" might be "96% / 95.5% / 97% vs 95%". The comparison sequences of all indicators together constitute the initial verification pool, providing basic data for subsequent performance verification.

[0133] Example 5:

[0134] S71. Based on a random disturbance strategy and a gray-box verification strategy, alternative verification schemes are performed according to the dynamic optimization results to obtain a verification scheme pool. The random disturbance strategy applies small random changes to various parameters in the dynamic optimization results, such as the proportional coefficient and integral time constant of the PI controller, generating multiple control schemes with subtle differences. The gray-box verification strategy combines the internal structural information of the inverter, such as the main circuit topology and filter parameters, with externally measured input and output data, such as DC-side voltage and AC-side output voltage and current, to verify these generated schemes. During the verification process, the stability of the output voltage and the distortion of the current waveform under disturbance are observed, and those schemes that can still maintain basic control performance after disturbance are collected to form the verification scheme pool.

[0135] S72. Based on optimization and integration technology, a comprehensive performance analysis is performed according to the verification scheme pool and control strategy requirements to obtain integrated feedback information and scheme matching degree. The optimization and integration technology evaluates each scheme in the verification scheme pool from multiple dimensions, including the steady-state fluctuation range of the output voltage, the response speed during load changes, and the current harmonic content under different loads. These evaluation dimensions are determined based on the control strategy requirements. For each scheme, its performance in each dimension is analyzed. For example, one scheme may have smaller voltage fluctuations in steady state but slower dynamic response, while another scheme may have faster dynamic response but slightly higher harmonic content. These analysis results form integrated feedback information. Simultaneously, based on the degree to which each scheme meets the requirements of each dimension, such as whether it reaches the set voltage fluctuation threshold and whether the response time is within the specified range, the matching degree between each scheme and the control strategy requirements is calculated. A higher matching degree indicates that the scheme better meets the overall requirements.

[0136] S73. Based on the sequence decoupling model, the verification scheme pool is updated according to the integrated feedback information and the scheme matching degree to obtain an updated scheme pool. The sequence decoupling model requires that the control scheme can effectively separate and control the positive and negative sequence components. Therefore, the schemes in the verification scheme pool are screened and adjusted based on the evaluation of the control effect of the positive and negative sequence components in the integrated feedback information, such as whether the negative sequence component is effectively suppressed and the tracking accuracy of the positive sequence component. For schemes with low matching degree and obvious defects indicated by the integrated feedback information, such as insufficient negative sequence suppression, parameter adjustments or structural modifications are made; for schemes that still cannot meet the sequence decoupling requirements after adjustment, they are removed from the pool. Schemes with high matching degree and better performance after adjustment are retained to form the updated scheme pool.

[0137] S74. Based on the control strategy requirement information and according to the updated scheme pool, the final scheme for the single-loop control model is determined, resulting in the optimized scheme for the sequential dual-loop PI control. S741. Based on the updated scheme pool, a set of candidate schemes and performance evaluation indicators are obtained. The set of candidate schemes consists of all schemes in the updated scheme pool that have undergone preliminary screening. Performance evaluation indicators include the steady-state error range of the output voltage, the overshoot limit of the dynamic response, and the upper limit of the total harmonic distortion rate of the current, all of which are derived from the control strategy requirement information.

[0138] S742. Perform steady-state accuracy evaluation on the candidate scheme set to obtain the first evaluation result. Through simulation or experimentation, operate each candidate scheme under different steady-state conditions such as rated load, light load, and heavy load, measure the deviation between the output voltage and the target voltage, and calculate the steady-state error. For example, under rated load, the voltage error of one scheme is 1V, and that of another scheme is 1.5V. Compile these error data to form the first evaluation result.

[0139] S743. Based on the candidate scheme set, the first evaluation result is determined as the starting point for scheme screening; the performance evaluation index is determined as the ending point for scheme screening; and the steady-state screening conditions are determined based on the starting point and the ending point for scheme screening. Using the steady-state error data in the first evaluation result as a basis, and combining it with the maximum allowable steady-state error specified in the performance evaluation index (e.g., not exceeding 2V), the steady-state screening conditions are determined, meaning that only schemes with a steady-state error less than or equal to 2V can proceed to the next stage of screening.

[0140] S744. Perform dynamic tracking evaluation on the candidate scheme set to obtain a second evaluation result. By changing the load or input command, such as suddenly increasing the load to 150% of the rated value, observe the dynamic changes in the output voltage and current of each candidate scheme, and record parameters such as response time and overshoot. For example, one scheme has a voltage overshoot of 3% and a response time of 0.1 seconds when the load changes abruptly, while another scheme has an overshoot of 5% and a response time of 0.08 seconds. These dynamic parameters constitute the second evaluation result.

[0141] S745. Based on the candidate solution set, the steady-state screening conditions are determined as the starting point for comprehensive screening; the second evaluation result is determined as the ending point for comprehensive screening; and the dynamic screening conditions are determined based on the comprehensive screening starting point and the comprehensive screening ending point. Taking the steady-state screening conditions as a premise, i.e., the solution already meets the steady-state error requirements, and combining the dynamic parameters in the second evaluation result and the dynamic performance requirements in the performance evaluation indicators, such as overshoot not exceeding 5% and response time not exceeding 0.1 seconds, the dynamic screening conditions are determined, i.e., the solution must simultaneously satisfy overshoot ≤ 5% and response time ≤ 0.1 seconds.

[0142] S746. Based on the dynamic screening conditions, the candidate scheme set is finally screened to obtain the optimized scheme for the sequential dual-loop PI control. From the schemes that have passed the steady-state screening, the schemes that simultaneously meet the dynamic screening conditions are selected. If there are multiple schemes that meet the conditions, their performance on other secondary indicators, such as efficiency and parameter robustness, is further compared. Finally, the scheme with the most balanced performance in all aspects is selected as the optimized scheme for the sequential dual-loop PI control.

[0143] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0144] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter, characterized in that, The method includes: S1. In the photovoltaic power generation system operating environment, obtain inverter topology parameters and control strategy requirements information; S2. Based on mathematical modeling tools, coordinate transformation theory and system identification methods, a single-loop control model and a sequence decoupling model are constructed according to the inverter topology parameters. S3. Based on the direct flow extraction strategy, the signal is separated according to the single-loop control model and the sequence decoupling model to obtain the positive sequence control component and the negative sequence control component. S4. Based on the dual-loop parameter matching rules and dynamic response constraints, the controller parameters are adjusted according to the positive-sequence control components and negative-sequence control components to obtain optimized PI parameters. S5. Based on the optimized PI parameters, dynamic performance optimization is performed according to the single-loop control model, the sequential decoupling model, and the inverter topology parameters to obtain the dynamic optimization results. S6. Based on the static verification method, the performance is verified by comparing historical data based on the single-loop control model, the sequence decoupling model and the control strategy requirements, and the static verification results are obtained. S7. Based on the dynamic optimization results and static verification results, the scheme is integrated to obtain the optimized scheme of the sequence dual-loop PI control. S8. Generate a control strategy implementation guide based on the optimized scheme of the sequenced dual-loop PI control.

2. The method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 1, characterized in that, The method, based on mathematical modeling tools, coordinate transformation theory, and system identification, constructs a single-loop control model and a sequence decoupling model according to the inverter topology parameters, including: S21. Based on the inverter topology parameters, obtain the main circuit topology model and load characteristic model; S22. Perform time-domain characteristic analysis on the main circuit topology model to obtain the differential expressions for voltage and current. S23. Based on the main circuit topology model, convert the differential expression into a frequency domain transfer function; determine the frequency domain transfer function as the basis of the single-loop control model; determine the main circuit topology model as the boundary of the single-loop control model; determine the single-loop control model based on the single-loop control model basis and the single-loop control model boundary. S24. Perform symmetric component decomposition on the load characteristic model to obtain the positive-sequence load component and the negative-sequence load component. S25. Based on the load characteristic model, the positive-sequence load component is determined as the positive-sequence decoupling starting point; the negative-sequence load component is determined as the negative-sequence decoupling starting point; and the sub-sequence decoupling nodes are determined based on the positive-sequence decoupling starting point and the negative-sequence decoupling starting point. S26. Perform zero-sequence component suppression analysis on the load characteristic model to obtain the zero-sequence suppression constraint conditions; determine the decoupling node as the constraint propagation starting point; determine the zero-sequence suppression constraint conditions as the constraint propagation ending point; determine the constraint propagation path based on the constraint propagation starting point and the constraint propagation ending point. S27. Based on the single-loop control model, the sequential decoupling nodes, and the constraint transmission path, the main circuit topology model is decoupled and integrated to obtain the sequential decoupling model. S28. Based on the rotation matrix operation of coordinate transformation theory, the frequency domain characteristics of the sequence decoupling model are verified to obtain the sequence decoupling model.

3. The method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 1, characterized in that, The DC-based extraction strategy, based on a single-loop control model and a sequence decoupling model, separates the signal to obtain positive-sequence control components and negative-sequence control components, including: S31. Perform signal characteristic checks based on the single-loop control model to obtain the first separation signal; S32. Based on the sequence decoupling model, perform signal integrity analysis on the single-loop control model to obtain the second separation signal; S33. Based on the single-loop control model, identify and analyze the feature extraction points of the DC flow extraction to obtain the extraction constraints and feature extraction statements; S34. Based on the sequence decoupling model, identify the signal separation points protected by the extraction strategy to obtain the separation target signal; S35. Based on the principle of double second-order generalized integrals, construct a direct flow extraction model according to the feature extraction statement and the separated target signal; S36. Using the direct flow extraction model, compare the extraction constraints to perform an extraction delay check and obtain the first separation component; S37. Based on the signal transmission path of the sequence decoupling model, component distortion is checked according to the extraction constraints and the DC extraction model to obtain the second separated component.

4. The method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 1, characterized in that, The process of adjusting controller parameters based on dual-loop parameter matching rules and dynamic response constraints, according to positive-sequence and negative-sequence control components, to obtain optimized PI parameters includes: S41. Obtain the reference variable for the outer loop control input based on the positive sequence control component; determine the reference variable as the parameter adjustment source; S42. Based on the optimized PI parameters, and according to the sequence decoupling model, parameter transfer analysis is performed on the single-loop control model to obtain the parameter transfer path; S43. Based on the parameter passing path, perform dynamic response entry checks on the optimized PI parameters to obtain a list of parameter adjustments with dynamic response entry points. S44. Based on the parameter adjustment list, track steady-state variables according to the order decoupling model to obtain constrained steady-state variables; S45. Determine the optimal PI parameters based on the parameter adjustment list and the constrained steady-state variables.

5. The method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 1, characterized in that, The static verification method, based on a single-loop control model, a sequence decoupling model, and control strategy requirements, performs performance verification using a historical data comparison algorithm to obtain static verification results, including: S61. Based on the single-loop control model and the sequence decoupling model, the static verification method is used to construct the data tuples and obtain the data quintuples. S62. Based on the historical data chain, construct a set of verification indicators according to the data quintuple; S63. Based on the control strategy requirements and the data quintuple, convert each verification indicator in the verification indicator set into a comparison sequence to obtain the initial verification pool.

6. The optimization method for sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 5, characterized in that, The verification indicator set, constructed based on historical data chains and data quintuples, includes: S621. Obtain indicator information based on the data quintuple; set the total number of indicators in the indicator information to M, the number of the current indicator to j, and let j=1; verify the indicator set to T, which is initially an empty set; retrieve the historical data chain set to Recorded, which is initially an empty set; the historical data chain is a quadruple representing the performance correlation between indicators. S622. Determine if j is greater than M. If j is greater than M, proceed to step S627. If j is less than or equal to M, proceed to step S623. S623. Based on the verification indicator set T, determine whether the current indicator is in T according to the indicator information. If the current indicator is in T, let j = j + 1 and execute step S622. If indicator j is not in T, create a verification indicator sequence ind based on the current indicator. S624. Based on the indicator information, retrieve the performance correlation according to the verification indicator sequence ind to obtain a new set of historical data chains; S625. Based on the retrieved historical data chain set Recorded, determine whether the new historical data chain set is a subset of Recorded. If the new historical data chain set is a subset of Recorded, update T according to ind, let j = j + 1, and execute step S622. If the new historical data chain set is not a subset of Recorded, execute step S626. S626. Update ind according to the new historical data chain set; update Recorded using the new historical data chain set, and proceed to step S624. S627, Output the set of verification metrics T.

7. The method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 1, characterized in that, The scheme integration based on dynamic optimization results and static verification results yields a sequence dual-loop PI control optimization scheme, including: S71. Based on the random perturbation strategy and the gray box verification strategy, the schemes are verified alternately according to the dynamic optimization results to obtain a verification scheme pool. S72. Based on optimization and integration technology, a comprehensive performance analysis is performed according to the verification scheme pool and control strategy requirements to obtain integrated feedback information and scheme matching degree. S73. Based on the sequence decoupling model, the verification scheme pool is updated according to the integrated feedback information and the scheme matching degree to obtain the updated scheme pool. S74. Based on the control strategy requirement information and according to the updated scheme pool, the final scheme of the single-loop control model is determined to obtain the optimized scheme of the sequenced double-loop PI control.

8. The method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 4, characterized in that, The step of performing a dynamic response entry check on the optimized PI parameters based on the parameter transmission path, and obtaining a list of parameter adjustments with dynamic response entry points, includes: S431. Based on the parameter passing path, obtain the main parameter passing segment and the secondary parameter passing segment; S432. Perform response time analysis on the main parameter transfer segment to obtain the first response threshold; S433. Based on the main parameter transfer segment, the first response threshold is determined as the entry check start point; the secondary parameter transfer segment is determined as the entry check end point; the dynamic response entry is determined based on the entry check start point and the entry check end point. S434. Perform overshoot analysis on the secondary parameter transfer segment to obtain the second response threshold. S435. Based on the secondary parameter transfer segment, the dynamic response entry point is determined as the threshold comparison start point; the second response threshold is determined as the threshold comparison end point; the response entry constraint is determined based on the threshold comparison start point and the threshold comparison end point. S436. Based on the response entry constraint, perform segmented verification on the optimized PI parameters to obtain a list of parameter adjustments with dynamic response entry points.

9. The method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 7, characterized in that, Based on the control strategy requirement information and according to the updated scheme pool, the final scheme determination for the single-loop control model is obtained, resulting in a sequence-based dual-loop PI control optimization scheme, including: S741. Based on the updated solution pool, obtain the set of candidate solutions and performance evaluation metrics; S742. Perform steady-state accuracy evaluation on the candidate scheme set to obtain the first evaluation result; S743. Based on the candidate scheme set, the first evaluation result is determined as the starting point for scheme screening; the performance evaluation index is determined as the ending point for scheme screening; and the steady-state screening conditions are determined based on the starting point and the ending point for scheme screening. S744. Perform dynamic follow-up evaluation on the candidate solution set to obtain the second evaluation result; S745. Based on the candidate scheme set, the steady-state screening conditions are determined as the starting point for comprehensive screening; the second evaluation result is determined as the ending point for comprehensive screening; and the dynamic screening conditions are determined based on the comprehensive screening starting point and the comprehensive screening ending point. S746. Based on dynamic screening conditions, the candidate scheme set is finally screened to obtain the optimized scheme of the sequenced double-loop PI control.

10. The method for optimizing the sequential dual-loop PI control of a three-phase four-arm photovoltaic inverter according to claim 2, characterized in that, The rotation matrix operation based on coordinate transformation theory is used to verify the frequency domain characteristics of the sequence decoupling model, including: S281. Obtain the positive-order component transfer function and the negative-order component transfer function based on the order decoupling model. S282. Perform frequency response analysis on the positive sequence component transfer function to obtain the positive sequence amplitude-frequency characteristic curve; S283. Based on the sequence decoupling model, the positive sequence amplitude-frequency characteristic curve is determined as the starting point of frequency domain verification; the negative sequence component transfer function is determined as the ending point of frequency domain verification; the frequency domain verification path is determined based on the starting point and the ending point of frequency domain verification. S284. Perform phase delay analysis on the transfer function of the negative sequence component to obtain the negative sequence phase frequency characteristic curve; S285. Based on the rotation matrix operation rules, characteristic matching verification is performed according to the positive sequence amplitude frequency characteristic curve and the negative sequence phase frequency characteristic curve to obtain the frequency domain characteristic verification results of the sequence decoupling model.

Images