Multi-agent-based virtual power plant layered voltage coordination control system and method

Abstract

The invention discloses a multi-agent-based virtual power plant layered voltage coordination control system and method, and relates to the technical field of coordination control, and the method comprises the steps: obtaining the original operation data of a distributed energy unit in a virtual power plant, and constructing a dynamic mapping model representing the relation between node voltage and control input; decomposing the dynamic mapping model into a fast dynamic sub-model and a slow dynamic sub-model; generating a fast loop control law based on the fast dynamic sub-model, and performing residual calculation and online correction on the slow dynamic sub-model according to the fast loop control law; establishing a global optimization objective function according to the corrected slow dynamic sub-model, and solving a reference value of each node through a distributed algorithm; issuing the node reference value to a local inverter, and generating a control input signal in combination with a fast loop control law; according to the method, the problems of voltage oscillation, tracking errors and slow convergence caused by virtual power plant fast and slow control conflicts are solved through fast and slow dynamic layering cooperative control.

Description

Technical Field

[0001] The present invention relates to the technical field of coordinated control, and more specifically, to a multi-agent-based virtual power plant hierarchical voltage coordinated control system and method. Background Art

[0002] Virtual power plants (VPPs) integrate different types of distributed energy resources (photovoltaic, energy storage, electric vehicle charging stations, flexible loads, etc.) to achieve unified scheduling and voltage control. The control objectives are: maintaining voltage at each node within acceptable limits; maintaining consistency in voltage trends across multiple resources (avoiding excessively high voltages at some nodes and excessively low voltages at others); and ensuring stable operation after renewable energy integration.

[0003] However, due to resource diversity, communication delays, and imprecise modeling, common control methods can exhibit flaws at the control level. For example, the fast control of local inverters and the slower commands issued by the virtual power plant coordinator can conflict on a timescale, leading to oscillations (periodic voltage deviations), tracking errors, or slow convergence. This present invention proposes a solution to these problems. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a hierarchical voltage coordination control system and method for a virtual power plant based on multi-agents, which solves the voltage oscillation, tracking error and slow convergence problems caused by the fast and slow control conflicts of the virtual power plant through fast and slow dynamic hierarchical collaborative control, and realizes rapid voltage response of each node in the virtual power plant, global coordinated stability and efficient integration of multiple types of distributed energy.

[0005] To achieve the above object, the present invention provides the following technical solutions: In the first aspect, the present application provides a hierarchical voltage coordination control method for a virtual power plant based on multi-agents, which includes: obtaining the original operating data of the distributed energy units in the virtual power plant, and constructing a dynamic mapping model that characterizes the relationship between the node voltage and the control input; decomposing the dynamic mapping model into a fast dynamic sub-model and a slow dynamic sub-model; based on the fast dynamic sub-model, generating a fast-loop control law, and performing residual calculation and online correction on the slow dynamic sub-model according to the fast-loop control law; establishing a global optimization objective function based on the corrected slow dynamic sub-model, and solving the reference value of each node through a distributed algorithm; sending the node reference value to the local inverter, and generating a control input signal in combination with the fast-loop control law.

[0006] In one embodiment, a dynamic mapping model characterizing the relationship between node voltage and control input is constructed, specifically: preprocessing the original operating data, and performing time series analysis on the preprocessed data to distinguish system dynamic variables with different time scales and extract key eigenvectors; based on the key eigenvectors, determining the model structure form, the model structure includes an autoregressive part related to the historical output and an exogenous input part related to the control input; according to the determined model structure, the node voltage output time series and the input signal are constructed into a regression model; the model structure parameters of the regression model are solved for the preprocessed data through a parameter identification algorithm; and according to the solved parameters, a dynamic mapping model between the node voltage and the control input is generated.

[0007] In one embodiment, the dynamic mapping model is decomposed into a fast dynamic sub-model and a slow dynamic sub-model, specifically: the small singular perturbation parameters are adaptively adjusted based on the Lyapunov adaptive law; the mapping model is scale-separated according to the adjusted small singular perturbation parameters to obtain a fast dynamic subsystem containing fast dynamic variables and a slow dynamic subsystem containing slow dynamic variables; the fast dynamic subsystem is subjected to variable freezing processing to obtain a fast dynamic sub-model that evolves with a fast time scale; the slow dynamic subsystem is subjected to asymptotic expansion processing to obtain a slow dynamic sub-model that evolves with a slow time scale.

[0008] In one embodiment, the small parameters of the singular perturbation are adaptively adjusted based on the Lyapunov adaptive law, specifically: obtaining the system control error and constructing a Lyapunov function; derivatizing the Lyapunov function and using the derivative result as a stability constraint; setting an adaptive law according to the stability constraint, and updating the value of the small parameters of the singular perturbation in real time according to the adaptive law for dynamic application.

[0009] In one embodiment, a fast loop control law is generated based on the fast dynamic sub-model, specifically: According to the fast dynamic submodel, the fast dynamic variables and discretized scheduling variables of the fast dynamic subsystem are obtained; based on the fast dynamic variables and scheduling variables, the fast dynamic subsystem is locally linearized to establish an LPV state-space model; for each group of LPV state-space models, the closed-loop pole positions are set according to the fast dynamic response performance; based on the pole configuration method and the closed-loop pole positions, the corresponding state feedback gain matrix is ​​solved, and each group of gain matrices is integrated to form a gain library; according to the scheduling variables, selection or multi-point interpolation is performed in the gain library to obtain the real-time control gain matrix, and combined with the fast dynamic variables to calculate the fast-loop control law.

[0010] In one embodiment, the slow dynamic sub-model is subjected to residual calculation and online correction according to the fast-loop control law, specifically: a slow dynamic prediction residual signal is obtained based on the difference between the actual node voltage under the fast-loop control law and the predicted output value of the slow dynamic sub-model; the predicted residual signal is preliminarily screened through wavelet transformation to obtain a preliminary residual signal; the preliminary residual signal is input into a low-pass filter for processing to obtain a low-pass residual signal; the slow dynamic sub-model is corrected according to the low-pass residual signal, and the residual is added as a feedback correction term to the slow dynamic sub-model state equation to obtain a corrected slow dynamic sub-model.

[0011] In one embodiment, the predicted residual signal is preliminarily screened through wavelet transformation to obtain a preliminary residual signal, specifically: the sampling frequency and the typical frequency of fast dynamics are obtained and the number of wavelet decomposition layers is calculated; based on the number of wavelet decomposition layers, the predicted residual signal is subjected to multi-scale wavelet decomposition to obtain wavelet coefficients of each layer, wherein the wavelet coefficients include high-frequency coefficients and low-frequency coefficients; according to the fast and slow system characteristics of the virtual power plant, the high-frequency coefficients are regarded as fast dynamic disturbances and the low-frequency coefficients are regarded as slow dynamic components; the high-frequency coefficients are removed and the low-frequency coefficients are retained as the preliminary residual signal.

[0012] In one embodiment, a global optimization objective function is established based on the modified slow dynamic sub-model, and the reference value of each node is solved by a distributed algorithm. Specifically, the global optimization objective function is constructed based on the state vector output by the modified slow dynamic sub-model and the voltage measurement value, and constraints are set. The global optimization objective function includes node voltage deviation, power loss and economic indicators, and the constraints include node voltage constraints, energy storage SOC constraints, and upper and lower limits of inverter power; the global optimization problem is split into local optimization sub-problems for each node to obtain the local objective function and neighbor node constraints of each node; a distributed optimization algorithm is used to iteratively optimize each node according to the local objective function and neighbor node constraints to obtain the local reference value of the current iteration round; each node sends the local reference value to the adjacent node through the communication network; the local constraints are updated according to the local reference value and local optimization is performed again to obtain a new round of reference values, until the preset global convergence conditions are met, and the final node reference value of each node is output, and the node reference value includes a node voltage reference value, a current reference value or a power reference value.

[0013] In one embodiment, a node reference value is sent to a local inverter, and a control input signal is generated in combination with a fast-loop control law, specifically: the node reference value is smoothed to generate a smoothed reference value; the smoothed reference value is sent to a local inverter controller through a control communication interface; the local inverter performs data verification after receiving the smoothed reference value; the control input signal is calculated according to the fast-loop control law and the verified smoothed reference value; and after receiving the control input signal, the local inverter adjusts the output power, current or voltage.

[0014] In a second aspect, the present application provides a multi-agent-based virtual power plant hierarchical voltage coordination control system and method system, the system comprising: The model building module is used to obtain the original operating data of the distributed energy units in the virtual power plant and build a dynamic mapping model that represents the relationship between node voltage and control input; A model decomposition module, used for decomposing the dynamic mapping model into a fast dynamic sub-model and a slow dynamic sub-model; The fast-loop control law generation and model correction module is used to generate the fast-loop control law based on the fast dynamic sub-model, and perform residual calculation and online correction on the slow dynamic sub-model according to the fast-loop control law; The slow dynamic optimization module is used to establish a global optimization objective function based on the modified slow dynamic sub-model and solve the reference value of each node through a distributed algorithm; The control execution module is used to send the node reference value to the local inverter and generate the control input signal in combination with the fast loop control law.

[0015] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages: 1. Through a mapping model based on system identification, the fast and slow dynamic decomposition of singular perturbation theory, and the LPV fast-loop control law, the virtual power plant node voltage is optimized and coordinated with the fast response at the millisecond to second level and the slow dynamic optimization at the minute to hour level. This not only takes into account the transient and steady-state characteristics of the voltage, avoiding oscillations and tracking errors caused by conflicts between fast and slow control, but also improves the global regulation accuracy and real-time control capabilities of multiple types of distributed energy while maintaining system stability.

[0016] 2. By combining the fast-loop control law with the residual correction of the slow-dynamic sub-model, the virtual power plant node voltage can achieve a rapid response in milliseconds to seconds. At the same time, wavelet decomposition and low-pass filtering are used to extract the slow-dynamic error signal for precise correction. The global node reference value is then solved through a distributed optimization algorithm and smoothly sent to the local inverter. This not only improves the model prediction accuracy and system stability, but also takes into account the minimization of power loss and economic optimization, achieving efficient coordination and global optimization control of fast and slow dynamics. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of a multi-agent-based hierarchical voltage coordination control method for a virtual power plant provided in an embodiment of the present application.

[0018] Figure 2 Schematic diagram of the structure of a multi-agent based virtual power plant hierarchical voltage coordination control system provided in an embodiment of the present application.

[0019] Figure 3 This is a scatter plot of scale separation of the fast and slow dynamic subsystems provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The following will provide a clear and complete description of 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. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0021] Reference Figure 1 As shown in FIG, the flowchart of the multi-agent-based virtual power plant hierarchical voltage coordination control method provided by the present invention includes the following steps: S1, obtains the original operating data of various distributed energy units in the virtual power plant, and constructs a dynamic mapping model that characterizes the relationship between node voltage and control input based on the parameter identification algorithm.

[0022] In this embodiment, various distributed energy units in the virtual power plant include photovoltaics, energy storage, electric vehicle charging piles, and flexible loads, and the original operating data includes control inputs (such as inverter reference power and current instructions) and node voltage outputs.

[0023] Among them, a dynamic mapping model that characterizes the relationship between node voltage and control input is constructed based on the parameter identification algorithm, specifically: Cleaning the original running data to obtain preprocessed data, wherein the cleaning includes denoising, outlier removal, missing data filling and normalization; Performing time series analysis on preprocessed data to distinguish fast-dynamic variables from slow-dynamic variables and extracting key feature vectors, including signal trend, periodicity, fluctuation amplitude, and hysteresis characteristics; Based on the key eigenvectors, the ARX (autoregressive with exogenous input) model structure is selected, where the ARX model structure includes the ARX model parameters of the autoregressive coefficients related to the historical output and the exogenous input coefficients related to the control input; According to the ARX model structure, the node voltage output time series and input signal are constructed into a regression equation, and the ARX model parameters are calculated based on the preprocessed data and the ARX model structure through the parameter identification algorithm; The specific calculation formula of the regression equation is as follows:

[0024] Where, is the observed value of the node voltage at time t, The node voltage is first The historical value at each sampling moment, is the autoregressive coefficient of the corresponding node voltage history value, ,..., is the current and historical value of the control input signal after a lag of d, where d is the input delay. is the exogenous input coefficient, is the error term.

[0025] A dynamic mapping model between node voltage and control input is generated according to ARX model parameters.

[0026] Among them, fast dynamic variables refer to signals that change significantly within a short time scale (usually milliseconds to seconds), and their change speed is much faster than the overall system regulation or slow-loop regulation capabilities, such as inverter current and voltage transients; slow dynamic variables refer to signals that change significantly on a longer time scale (usually minutes to hours), and their change speed is slower than the fast dynamic response, such as energy storage SOC and photovoltaic power prediction.

[0027] It should be noted that the mapping model of node voltage and control input constructed through the ARX system identification method can directly utilize the input and output data collected in the virtual power plant without establishing a complex physical mechanism model, so as to capture the dynamic relationship between node voltage and control input changes; at the same time, the ARX model has a clear structure and low computational complexity, and its parameters can be quickly obtained through the parameter identification algorithm, which is convenient for updating and application in real-time scheduling; in addition, by distinguishing between fast dynamic variables and slow dynamic variables and extracting key eigenvectors, the ARX model can take into account both transient and steady-state characteristics of voltage, thereby ensuring the sensitivity of control response and improving the accuracy of system operation prediction and regulation.

[0028] S2, the dynamic mapping model is decomposed into a fast dynamic sub-model and a slow dynamic sub-model through singular perturbation theory.

[0029] In this example, the dynamic mapping model is decomposed into a fast dynamic sub-model and a slow dynamic sub-model using singular perturbation theory. Specifically: Based on Lyapunov adaptive law for singular perturbation of small parameters Adaptively adjust the small singular perturbation parameters and introduce them into the dynamic mapping model to characterize the time scale difference between fast and slow dynamics. like Figure 3 As shown, the mapping model is scale-separated according to the singular perturbation small parameter to obtain a fast dynamic subsystem containing fast dynamic variables and a slow dynamic subsystem containing slow dynamic variables; Freezing the variables of the fast dynamic subsystem, that is, treating the slow dynamic variables as constants, to obtain a fast dynamic submodel that evolves only with a fast time scale; The specific calculation formula of the fast dynamic sub-model is as follows:

[0030] Where, is the fast dynamic sub-model, is a fast dynamic variable, including inverter current, voltage transient and other fast dynamic variables. represents the fast time scale, For slow dynamic variables, such as energy storage SOC and photovoltaic power prediction, is the control input vector, It is a fast dynamic output function that maps fast dynamic variables, slow dynamic variables and control input vectors into fast dynamic node voltage outputs.

[0031] The slow dynamic subsystem is subjected to asymptotic expansion, that is, the fast dynamic variables are regarded as algebraic constraints that instantaneously reach a steady state, and a slow dynamic submodel that evolves only with a slow time scale is obtained.

[0032] The specific calculation formula of the slow dynamic sub-model is as follows:

[0033] Where, The state vector output by the slow dynamic sub-model includes the energy storage SOC (battery state of charge), photovoltaic power forecast value and flexible load power trend. is the algebraic solution of the fast dynamic subsystem, that is, after the dynamic subsystem reaches the quasi-steady state condition, the fast variable By slow variables and the control input vector The function represented by It is a slow dynamic system function.

[0034] The fast dynamic subsystem reflects the rapid transient regulation of voltage on a time scale, while the slow dynamic subsystem reflects the long-term evolution trend of voltage on a time scale. The singular perturbation parameter is a dimensionless small quantity in singular perturbation theory that represents the difference in time scales between the fast and slow dynamics in a system.

[0035] Furthermore, the singular perturbation small parameters are adaptively adjusted based on the Lyapunov adaptive law, specifically: Obtain the system control error and construct the Lyapunov function; The specific calculation formula of the Lyapunov function is as follows:

[0036] Where, is the Lyapunov function, is the system control error, that is, the deviation vector between the node voltage output and the expected voltage, is a small singular perturbation parameter, is the preset expected singular perturbation small parameter range, is the regulating factor, is the quadratic form of the system control error, that is, the sum of the squares of the errors, is the transpose of the system error vector.

[0037] Derivative the Lyapunov function and use the derivative result as the stability constraint ; The specific calculation formula for the derivation is as follows:

[0038] Where, is the derivative of the Lyapunov function, is the rate of change of the system error over time, is the rate of change of the singular perturbation parameter with time.

[0039] An adaptive law is set according to the stability constraint condition, and the value of the small singular perturbation parameter is updated in real time according to the adaptive law and applied dynamically, so that the fast dynamic sub-model and the slow dynamic sub-model always maintain a reasonable time scale separation.

[0040] The specific calculation formula of the adaptive law is as follows:

[0041] Where, is a regression vector composed of preprocessed data, reflecting the sensitivity of the system control error to the small singular perturbation parameters.

[0042] It should be noted that decomposing the mapping model into a fast dynamic sub-model and a slow dynamic sub-model can effectively distinguish between the fast transient response and the slow long-term evolution trend in the system, thereby realizing the decoupling of time scales in analysis and control and simplifying the complexity of the system; at the same time, based on the Lyapunov adaptive law, the small parameters of the singular perturbation are adjusted in real time, which can adaptively optimize the time scale separation of the fast and slow dynamic subsystems according to the system control error, so that the fast dynamic sub-model and the slow dynamic sub-model always maintain a reasonable scale difference, which not only ensures the stability of the system, but also improves the control accuracy and dynamic performance, and achieves a balance between fast response and robust control.

[0043] S3, establish an LPV state space model that depends on the adjustment parameters for the fast dynamic sub-model and output the fast loop control law.

[0044] In this embodiment, an LPV state space model that depends on the adjustment parameters is established for the fast dynamic sub-model, and a fast loop control law is output, specifically: According to the fast dynamic sub-model, fast dynamic variables of the fast dynamic sub-system are obtained, wherein the fast dynamic variables include inverter current and voltage transients; Obtain the scheduling variables defined by the uncertainties and time-varying characteristic parameters (such as node load changes, PV output fluctuations, and battery SOC) of the fast dynamic subsystem; Among them, scheduling variables refer to external or internal parameters that affect the dynamic characteristics of the system but are not control inputs themselves; Based on the fast dynamic variables and discretized scheduling variables, the fast dynamic subsystem is locally linearized to establish an LPV (Linear Parameter Variation) state space model; The specific calculation formula of the LPV state space model is as follows:

[0045]

[0046] Where, is the fast dynamic state change rate, is the scheduling variable, is the linear mapping matrix of the fast dynamic state to its own changes, is the linear influence matrix of the control input on the fast dynamic state change, For fast dynamic output, is the linear mapping matrix of fast dynamic variables to output, For control input Output Direct impact matrix.

[0047] For each set of LPV state-space models, setting closed-loop pole positions based on fast dynamic response performance, wherein the fast dynamic response performance includes damping ratio and natural frequency; The specific calculation formula of the closed-loop pole is as follows:

[0048] Where, are the two closed-loop poles of the system, is the damping ratio, is the natural frequency, and j is the imaginary unit.

[0049] Based on the pole placement method combined with the closed-loop pole solution, the corresponding state feedback gain matrix is ​​obtained. , integrate each group of gain matrices to form a gain library; The specific calculation formula of the gain matrix is ​​as follows:

[0050] Where, is the linear mapping matrix of the fast dynamic state under the i-th scheduling variable to its own change, is the linear influence matrix of the control input on the fast dynamic state change under the i-th scheduling variable, is the nth closed-loop pole, is the closed-loop system matrix, is the set of eigenvalues ​​of the closed-loop system matrix.

[0051] According to the scheduling variables, select or interpolate in the gain library to obtain the real-time control gain matrix , and the fast loop control law is obtained by combining the fast dynamic variable calculation.

[0052] The specific calculation formula of the fast loop control law is as follows:

[0053] Where, is the fast loop control law.

[0054] Among them, the fast-loop control law refers to the rule or formula for mapping fast dynamic variables to control inputs in real time. It is the core of the fast-loop controller to achieve rapid stabilization of node voltage. It is used to quickly adjust the node voltage, allowing the system to quickly adjust the node voltage within a time scale of milliseconds to seconds, so that the voltage is close to the desired value.

[0055] S4, performing residual calculation on the slow dynamic sub-model according to the fast loop control law to obtain a residual signal, and correcting the slow dynamic sub-model.

[0056] In this embodiment, the slow dynamic sub-model is subjected to residual calculation according to the fast loop control law to obtain a residual signal, and the slow dynamic sub-model is corrected, specifically as follows: Obtaining a slow dynamic prediction residual signal based on the difference between the actual node voltage under the fast loop control law and the output value predicted by the slow dynamic sub-model, wherein the prediction residual signal indicates the magnitude and direction of the deviation of the slow dynamic sub-model output from the actual operating state; Performing preliminary screening on the prediction residual signal through wavelet transformation to obtain a preliminary residual signal; The preliminary residual signal is input into a low-pass filter for processing to suppress the interference of fast dynamic fluctuations on slow dynamic correction, so that the residual reflects the system error under the slow time scale, and a low-pass residual signal is obtained; The specific calculation formula of the low-pass residual signal is as follows: ,

[0057] Where, is the low-pass residual signal, is the filter coefficient, is the slow motion prediction residual vector.

[0058] The slow dynamic sub-model is corrected according to the low-pass residual signal, and the residual is added into the state equation of the slow dynamic sub-model as a feedback correction term to obtain the corrected slow dynamic sub-model.

[0059] The specific calculation formula of the modified slow dynamic sub-model is as follows:

[0060] Where, is the state vector output by the modified slow dynamic sub-model, is the residual feedback gain matrix, which can be determined by empirical adjustment or robust optimization.

[0061] It is important to note that using the actual response of the fast-loop control law as the basis for slow-dynamic model corrections allows for the decoupling of fast and slow dynamics, improving model prediction accuracy. Low-pass filtering prevents interference from fast-dynamic fluctuations and enhances the stability of slow-dynamic corrections. The corrected slow-dynamic sub-model can be directly used in distributed optimization calculations, ensuring the stability and economic efficiency of virtual power plant node voltages on slow timescales.

[0062] Furthermore, the prediction residual signal is preliminarily screened by wavelet transformation to obtain a preliminary residual signal, specifically: According to the characteristics of the virtual power plant system, a wavelet basis with orthogonality and good time-frequency localization is selected, such as Daubechies db4; Obtain the sampling frequency and the typical frequency of fast dynamics and calculate the number of wavelet decomposition layers; The specific calculation formula for the number of wavelet decomposition layers is as follows:

[0063] Where, is the number of wavelet decomposition layers, is the sampling frequency, is the typical frequency of fast dynamics, It is a rounding function that ensures that the number of wavelet decomposition levels is a practical integer.

[0064] Perform multi-scale wavelet decomposition on the prediction residual signal based on the number of wavelet decomposition layers to obtain wavelet coefficients of each layer, wherein the wavelet coefficients include high-frequency coefficients and low-frequency coefficients; Among them, the wavelet coefficients represent the components of the prediction residual signal in different frequency bands; According to the fast and slow system characteristics of the virtual power plant, the high-frequency coefficient is regarded as a fast dynamic disturbance, and the low-frequency coefficient is regarded as a slow dynamic component; The high-frequency coefficients are removed and the low-frequency coefficients are retained as the preliminary residual signal.

[0065] It should be noted that according to the fast and slow system characteristics of the virtual power plant, the high-frequency coefficients are regarded as fast dynamic disturbances, and the low-frequency coefficients are regarded as slow dynamic components. It can be understood that the signal is decomposed into different frequency bands using wavelet multi-scale decomposition. The fast dynamics are mainly manifested as high-frequency components (obvious transient fluctuations), and the slow dynamics are mainly manifested as low-frequency components (trend changes).

[0066] S5, establish the global optimization objective function according to the modified slow dynamic sub-model, and solve the reference value of each node through the distributed algorithm.

[0067] In this example, a global optimization objective function is established based on the modified slow dynamic sub-model, and the reference value of each node is solved through a distributed algorithm. Specifically, Based on the state vector output by the modified slow dynamic sub-model and the voltage measurement value, a global optimization objective function is constructed and constraints are set. The global optimization objective function includes node voltage deviation, power loss, and economic indicators (the weighted sum of energy storage charging and discharging costs and renewable energy utilization rate). The constraints include node voltage constraints, energy storage SOC constraints, and inverter power upper and lower limits. The specific calculation formula of the global optimization objective function is as follows:

[0068] Where, is the global optimization objective function, is the set of all nodes in the virtual power plant, is the actual voltage value of the i-th node at time t, is the voltage reference value of the i-th node, 、 、 are weight coefficients, is the power loss of the i-th node at time t, is the economic index of the i-th node at time t.

[0069] Split the global optimization problem into local optimization sub-problems for each node, and obtain the local objective function of each node and neighbor node constraints; The specific calculation formula of the local objective function is as follows:

[0070] Using a distributed optimization algorithm, each node performs iterative optimization based on the local objective function and neighbor node constraints to obtain the local reference value of the current iteration round. ; Each node sends the calculated local reference value to the adjacent nodes through the communication network; The local constraints are updated according to the local reference values ​​and local optimization is performed again to obtain a new round of reference values ​​until the preset global convergence conditions are met, and the final node reference values ​​of each node are output. The node reference values ​​include node voltage reference values, current reference values ​​or power reference values.

[0071] The global convergence condition is specifically:

[0072] Where, is the preset convergence threshold.

[0073] The constraints are as follows: Node voltage constraints:

[0074] Where, is the minimum allowable voltage, is the maximum allowable voltage; Energy storage SOC constraints:

[0075] Where, is the minimum energy storage SOC, For energy storage SOC, is the maximum energy storage SOC; Inverter power upper and lower limits:

[0076] Where, is the minimum inverter power, is the inverter power, is the maximum inverter power.

[0077] It should be noted that through distributed iterative optimization, the virtual power plant can achieve global voltage stability, minimize power loss and economic optimization under slow time scales. The distributed strategy reduces dependence on the central controller and improves the scalability and robustness of the system.

[0078] S6: Send the node reference value to the local inverter and generate the control input signal in combination with the fast loop control law.

[0079] In this embodiment, the node reference value is sent to the local inverter, and the control input signal is generated in combination with the fast loop control law, specifically: Smoothing the node reference value to generate a smoothed reference value ; The specific calculation formula for the smoothing process is as follows:

[0080] Where, is the smoothing coefficient.

[0081] Sending the smoothing reference value to the local inverter controller through the control communication interface; After receiving the smoothed reference value, the local inverter performs data verification, which includes data integrity check and data range legitimacy check to ensure that the reference value is within the executable range of the inverter; The control input signal is calculated based on the fast loop control law and the smooth reference value after verification. ;

[0082] Where, is the fast loop feedback gain matrix, is the fast loop feedforward control quantity.

[0083] Among them, the fast-loop control law compares the smoothed reference value with the actual node status and generates inverter control instructions within a time scale of milliseconds to seconds to achieve rapid adjustment of node voltage or power.

[0084] After receiving the control input signal, the local inverter adjusts the output power, current or voltage.

[0085] It should be noted that the smoothed reference value reduces the impact of sudden changes on the fast-loop control and improves system stability. The fast-loop control law combined with the smoothed reference value realizes rapid response of node voltage or power. Combined with slow dynamic optimization, the virtual power plant can take into account the coordination of fast and slow dynamics and global optimization.

[0086] Reference Figure 2As shown in the figure, the structure diagram of the multi-agent-based virtual power plant hierarchical voltage coordination control system provided by the present invention includes a model construction module, a model decomposition module, a fast-loop control law generation and model correction module, a slow dynamic optimization module, and a control execution module. There are connections between the modules: The model building module is used to obtain the original operating data of the distributed energy units in the virtual power plant and build a dynamic mapping model that represents the relationship between node voltage and control input; A model decomposition module, used for decomposing the dynamic mapping model into a fast dynamic sub-model and a slow dynamic sub-model; The fast-loop control law generation and model correction module is used to generate the fast-loop control law based on the fast dynamic sub-model, and perform residual calculation and online correction on the slow dynamic sub-model according to the fast-loop control law; The slow dynamic optimization module is used to establish a global optimization objective function based on the modified slow dynamic sub-model and solve the reference value of each node through a distributed algorithm; The control execution module is used to send the node reference value to the local inverter and generate the control input signal in combination with the fast loop control law.

[0087] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0088] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0089] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0090] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0091] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0092] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A multi-agent-based virtual power plant hierarchical voltage coordination control method, characterized in that: include: Obtain the original operating data of distributed energy units in the virtual power plant and build a dynamic mapping model that characterizes the relationship between node voltage and control input; Decompose the dynamic mapping model into a fast dynamic sub-model and a slow dynamic sub-model; Based on the fast dynamic sub-model, a fast loop control law is generated, and residual calculation and online correction of the slow dynamic sub-model are performed according to the fast loop control law; Based on the modified slow dynamic sub-model, a global optimization objective function is established, and the reference value of each node is solved through a distributed algorithm; The node reference value is sent to the local inverter and combined with the fast loop control law to generate the control input signal.

2. The multi-agent-based virtual power plant hierarchical voltage coordination control method according to claim 1 is characterized in that: The dynamic mapping model that characterizes the relationship between node voltage and control input is constructed as follows: Preprocess the original operating data and perform time series analysis on the preprocessed data to distinguish system dynamic variables with different time scales and extract key feature vectors; Determining a model structure based on the key eigenvectors, wherein the model structure includes an autoregressive part related to historical outputs and an exogenous input part related to control inputs; According to the determined model structure, the node voltage output time series and the input signal are constructed into a regression model; Solving the model structure parameters of the regression model by using a parameter identification algorithm on the preprocessed data; A dynamic mapping model between the node voltage and the control input is generated according to the solved parameters.

3. The multi-agent-based virtual power plant hierarchical voltage coordination control method according to claim 2 is characterized in that: The dynamic mapping model is decomposed into a fast dynamic sub-model and a slow dynamic sub-model, specifically: Adaptive adjustment of singular perturbation small parameters based on Lyapunov adaptive law; The mapping model is scale-separated according to the adjusted singular perturbation small parameters to obtain a fast dynamic subsystem containing fast dynamic variables and a slow dynamic subsystem containing slow dynamic variables; The fast dynamic subsystem is subjected to variable freezing to obtain a fast dynamic submodel that evolves with a fast time scale; The slow dynamic subsystem is asymptotically expanded to obtain a slow dynamic submodel that evolves with a slow time scale.

4. The multi-agent-based virtual power plant hierarchical voltage coordination control method according to claim 3 is characterized in that: The adaptive adjustment of the singular perturbation small parameters based on the Lyapunov adaptive law is specifically as follows: Obtain the system control error and construct the Lyapunov function; Derivative the Lyapunov function, and use the derivative result as a stability constraint; An adaptive law is set according to the stability constraint condition, and the value of the small parameter of the singular perturbation is updated in real time according to the adaptive law for dynamic application.

5. The multi-agent-based virtual power plant hierarchical voltage coordination control method according to claim 4 is characterized in that: The fast-loop control law is generated based on the fast dynamic sub-model, specifically: According to the fast dynamic sub-model, the fast dynamic variables of the fast dynamic sub-system and the discretized scheduling variables are obtained; Based on fast dynamic variables and scheduling variables, the fast dynamic subsystem is locally linearized to establish the LPV state space model; For each set of LPV state space models, the closed-loop pole positions are set according to the fast dynamic response performance; Based on the pole placement method and the closed-loop pole position, the corresponding state feedback gain matrix is ​​solved and each group of gain matrices is integrated to form a gain library. According to the scheduling variables, selection or multi-point interpolation is performed in the gain library to obtain the real-time control gain matrix, and the fast loop control law is obtained by combining it with the fast dynamic variable calculation.

6. The multi-agent-based virtual power plant hierarchical voltage coordination control method according to claim 5 is characterized in that: The residual calculation and online correction of the slow dynamic sub-model according to the fast loop control law are specifically as follows: The slow dynamic prediction residual signal is obtained according to the difference between the actual node voltage under the fast loop control law and the output value predicted by the slow dynamic sub-model; The prediction residual signal is preliminarily screened by wavelet transformation to obtain a preliminary residual signal; Inputting the preliminary residual signal into a low-pass filter for processing to obtain a low-pass residual signal; The slow dynamic sub-model is corrected according to the low-pass residual signal, and the residual is added into the state equation of the slow dynamic sub-model as a feedback correction term to obtain the corrected slow dynamic sub-model.

7. The multi-agent-based virtual power plant hierarchical voltage coordination control method according to claim 6 is characterized in that: The prediction residual signal is preliminarily screened by wavelet transformation to obtain a preliminary residual signal, specifically: Obtain the sampling frequency and the typical frequency of fast dynamics and calculate the number of wavelet decomposition layers; Perform multi-scale wavelet decomposition on the prediction residual signal based on the number of wavelet decomposition layers to obtain wavelet coefficients of each layer, wherein the wavelet coefficients include high-frequency coefficients and low-frequency coefficients; According to the fast and slow system characteristics of the virtual power plant, the high-frequency coefficient is regarded as a fast dynamic disturbance, and the low-frequency coefficient is regarded as a slow dynamic component; The high-frequency coefficients are removed and the low-frequency coefficients are retained as the preliminary residual signal.

8. The multi-agent-based virtual power plant hierarchical voltage coordination control method according to claim 7 is characterized in that: According to the modified slow dynamic sub-model, a global optimization objective function is established, and the reference value of each node is solved by a distributed algorithm, specifically: Based on the state vector output by the modified slow dynamic sub-model and the voltage measurement value, a global optimization objective function is constructed and constraints are set. The global optimization objective function includes node voltage deviation, power loss and economic indicators. The constraints include node voltage constraints, energy storage SOC constraints, and inverter power upper and lower limits. Split the global optimization problem into local optimization sub-problems for each node, and obtain the local objective function of each node and the constraints of neighboring nodes; Using a distributed optimization algorithm, each node is iteratively optimized according to the local objective function and the constraints of neighboring nodes to obtain the local reference value of the current iteration round; Each node sends the local reference value to the adjacent nodes through the communication network; The local constraints are updated according to the local reference values ​​and local optimization is performed again to obtain a new round of reference values ​​until the preset global convergence conditions are met, and the final node reference values ​​of each node are output. The node reference values ​​include node voltage reference values, current reference values ​​or power reference values.

9. The multi-agent-based virtual power plant hierarchical voltage coordination control method according to claim 8, characterized in that: The node reference value is sent to the local inverter, and the control input signal is generated in combination with the fast loop control law, specifically: Smoothing the node reference value to generate a smoothed reference value; Sending the smoothing reference value to the local inverter controller through the control communication interface; The local inverter performs data verification after receiving the smooth reference value; The control input signal is calculated based on the fast loop control law and the verified smooth reference value; After receiving the control input signal, the local inverter adjusts the output power, current or voltage.

10. A system using the multi-agent-based virtual power plant hierarchical voltage coordination control method according to any one of claims 1 to 9, characterized in that: include: The model building module is used to obtain the original operating data of the distributed energy units in the virtual power plant and build a dynamic mapping model that represents the relationship between node voltage and control input; A model decomposition module, used for decomposing the dynamic mapping model into a fast dynamic sub-model and a slow dynamic sub-model; The fast-loop control law generation and model correction module is used to generate the fast-loop control law based on the fast dynamic sub-model, and perform residual calculation and online correction on the slow dynamic sub-model according to the fast-loop control law; The slow dynamic optimization module is used to establish a global optimization objective function based on the modified slow dynamic sub-model and solve the reference value of each node through a distributed algorithm; The control execution module is used to send the node reference value to the local inverter and generate the control input signal in combination with the fast loop control law.

Images