A photovoltaic cluster control method, device, and equipment based on distributed optimization

Abstract

This invention discloses a photovoltaic (PV) cluster control method, apparatus, and equipment based on distributed optimization. The method includes: updating the original PV cluster topology when a target PV unit is detected to be connected to the distribution network, obtaining the target PV cluster; determining frequency performance indicators, energy performance indicators, and energy storage performance indicators based on the operating data of the target PV cluster, and determining global optimization indicators; constructing PV objective functions and energy storage objective functions based on the global optimization indicators, and jointly solving the PV objective functions and energy storage objective functions using a fractional-order optimization method and a multi-objective genetic algorithm to obtain a global control sequence; verifying the global control sequence to obtain the target control sequence and converting it into target control commands for coordinated control of the target PV cluster. This technical solution realizes coordinated regulation of PV and energy storage in a PV cluster, improves the rationality of global optimization, and enhances the overall efficiency and stability of the cluster operation.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation control technology, and in particular to a photovoltaic cluster control method, device and equipment based on distributed optimization. Background Technology

[0002] With the large-scale development of the photovoltaic industry, distributed photovoltaic clusters have become an important carrier for the power grid to absorb new energy. However, the current cluster control technology has many bottlenecks, making it difficult to meet the needs of flexible and efficient operation.

[0003] Distributed photovoltaic (PV) clusters comprise a massive number of distributed generation units, supporting energy storage units, and power conversion equipment. They can be flexibly integrated into the distribution network architecture, effectively improving the local consumption capacity of new energy and alleviating grid transmission pressure. However, PV output is highly volatile and random due to factors such as sunlight intensity and environmental weather. Coupled with the dynamic changes in load-side electricity demand, nonlinear energy storage charging and discharging losses, and operational deviations of power electronic equipment, the limitations of traditional cluster control models are becoming increasingly apparent, making it difficult to meet the comprehensive operational requirements of flexible integration, stable operation, efficient utilization, and economic dispatch for distributed PV clusters. Moreover, existing PV cluster optimization control schemes are clearly one-sided in design. Most technologies focus solely on power consumption or power dispatch, failing to comprehensively consider multi-dimensional operational needs. This easily leads to local optima rather than global optima, resulting in a lack of a sound global coordination mechanism. Local optimization is disconnected from overall dispatch, leading to insufficient cluster operating efficiency and stability. Summary of the Invention

[0004] This invention provides a photovoltaic cluster control method, device, and equipment based on distributed optimization. By constructing a global optimization target through multi-dimensional indicators, the target control command is determined, realizing the coordinated regulation of photovoltaic and energy storage in the photovoltaic cluster, improving the rationality of global optimization, and enhancing the efficiency and stability of the overall cluster operation.

[0005] According to one aspect of the present invention, a photovoltaic cluster control method based on distributed optimization is provided, comprising: When a target photovoltaic unit is detected to be connected to the distribution network, the adaptability of the target photovoltaic unit is determined, and the original photovoltaic cluster topology is updated based on the adaptability to obtain the target photovoltaic cluster. The operation data of the target photovoltaic cluster is obtained, and the frequency performance index, energy performance index and energy storage performance index are determined based on the operation data. The global optimization index is determined based on the frequency performance index, the energy performance index and the energy storage performance index. Based on the global optimization index, photovoltaic objective functions and energy storage objective functions are constructed respectively. The photovoltaic objective functions and energy storage objective functions are jointly solved by combining fractional optimization method and multi-objective genetic algorithm to obtain global control sequence. The global control sequence is verified based on the established verification method to obtain the target control sequence, and the target control sequence is converted into target control instructions to perform coordinated control of the target photovoltaic cluster.

[0006] According to another aspect of the present invention, a photovoltaic cluster control device based on distributed optimization is provided, comprising: The cluster update module is used to determine the adaptability of the target photovoltaic unit when it is detected that the target photovoltaic unit is connected to the distribution network, and update the original photovoltaic cluster topology based on the adaptability to obtain the target photovoltaic cluster. The indicator determination module is used to acquire the operating data of the target photovoltaic cluster, determine the frequency performance indicator, energy performance indicator and energy storage performance indicator based on the operating data, and determine the global optimization indicator based on the frequency performance indicator, the energy performance indicator and the energy storage performance indicator. The global control sequence determination module is used to construct photovoltaic objective functions and energy storage objective functions based on the global optimization index, and to jointly solve the photovoltaic objective functions and the energy storage objective functions by combining fractional optimization method and multi-objective genetic algorithm to obtain the global control sequence; The collaborative control module is used to verify the global control sequence based on a set verification method to obtain the target control sequence, and to convert the target control sequence into target control commands to perform collaborative control on the target photovoltaic cluster.

[0007] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to execute the distributed optimization-based photovoltaic cluster control method according to any embodiment of the present invention.

[0008] The technical solution of this invention involves determining the suitability of a target photovoltaic unit when it is detected that the unit is connected to the distribution network, updating the original photovoltaic cluster topology based on the suitability, and obtaining a target photovoltaic cluster. The solution then acquires the operating data of the target photovoltaic cluster, determines frequency performance indicators, energy performance indicators, and energy storage performance indicators based on the operating data, and determines global optimization indicators based on these indicators. Based on the global optimization indicators, photovoltaic objective functions and energy storage objective functions are constructed respectively. These objective functions are then jointly solved using a fractional-order optimization method and a multi-objective genetic algorithm to obtain a global control sequence. The global control sequence is verified using a set verification method to obtain a target control sequence, which is then converted into target control commands for coordinated control of the target photovoltaic cluster. This technical solution, through the construction of a global optimization objective using multi-dimensional indicators and the determination of target control commands, achieves coordinated regulation of photovoltaic and energy storage within the photovoltaic cluster, improves the rationality of global optimization, and enhances the overall efficiency and stability of the cluster's operation.

[0009] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart of a photovoltaic cluster control method based on distributed optimization according to Embodiment 1 of the present invention.

[0012] Figure 2 This is a flowchart of a photovoltaic cluster control method based on distributed optimization according to Embodiment 2 of the present invention.

[0013] Figure 3 This is a schematic diagram of a photovoltaic cluster control device based on distributed optimization according to Embodiment 3 of the present invention.

[0014] Figure 4 This is a schematic diagram of the structure of an electronic device provided according to Embodiment 4 of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0016] It should be noted that the terms "target," "original," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] Example 1 Figure 1 This is a flowchart of a photovoltaic cluster control method based on distributed optimization according to Embodiment 1 of the present invention. This embodiment is applicable to the collaborative control of plug-and-play photovoltaic clusters. The method can be executed by a photovoltaic cluster control device based on distributed optimization. This photovoltaic cluster control device based on distributed optimization can be implemented in hardware and / or software, and can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes: S110. When the target photovoltaic unit is detected to be connected to the distribution network, the adaptability of the target photovoltaic unit is determined, and the original photovoltaic cluster topology is updated based on the adaptability to obtain the target photovoltaic cluster.

[0018] In this context, the target photovoltaic (PV) unit can refer to a newly connected PV unit or a distributed PV power generation device to be integrated into the PV cluster. In this embodiment, the target PV unit may include PV modules, inverters, and local controllers. Adaptability refers to a quantitative parameter representing the degree of compatibility between the newly connected PV unit and the original PV cluster. In this embodiment, adaptability can be used to measure the compatibility of the new PV unit with the original PV cluster in terms of electrical parameters, communication protocols, operating characteristics, and control logic. The original PV cluster can refer to an existing PV cluster, including PV units, energy storage units, and gateway nodes. In this embodiment, the original PV cluster topology can refer to the electrical connections, communication network architecture, and physical distribution structure between PV units, energy storage units, and gateway nodes in the existing PV cluster. The target PV cluster can refer to the PV cluster obtained after the new unit has been adapted and integrated.

[0019] In this embodiment, the communication interface and power access port of the distribution network can be monitored in real time to identify the newly added grid-connected target photovoltaic unit. After completing the identification of the access device, the rated power, operating parameters, communication protocol and voltage of the target photovoltaic unit are collected and compared with the benchmark operating parameters, communication networking rules and electrical adaptation standards of the existing photovoltaic cluster. The access adaptation degree of the target photovoltaic unit is quantified. Based on the comparison between the adaptation degree and the preset adaptation degree threshold, if the adaptation degree is greater than the preset adaptation degree threshold, the target photovoltaic unit can be directly connected to the existing photovoltaic cluster. The communication link and electrical connection relationship of the new unit are added, the cluster connection architecture is reconstructed, and the original topology is updated to form a target photovoltaic cluster containing the new unit, realizing plug-and-play photovoltaic units.

[0020] S120. Obtain the operating data of the target photovoltaic cluster, determine the frequency performance index, energy performance index and energy storage performance index based on the operating data, and determine the global optimization index based on the frequency performance index, energy performance index and energy storage performance index.

[0021] The operational data refers to the electrical parameters, equipment status parameters, and environmental condition parameters collected in real time during the operation of the photovoltaic cluster. In this embodiment, the operational data may include grid frequency parameters, photovoltaic output data, load power consumption data, energy storage charging and discharging parameters, equipment loss data, and operational data related to photovoltaic-energy storage synergy. In this embodiment, the target photovoltaic cluster includes each photovoltaic unit, energy storage unit, and gateway node, and the operational data may include the relevant actual operational data of each photovoltaic unit, energy storage unit, and gateway node. Frequency performance indicators can be quantitative evaluation indicators characterizing the photovoltaic cluster's ability to support the distribution network frequency, the effect of frequency fluctuation suppression, and the virtual inertia response level. Energy performance indicators can be comprehensive indicators used to evaluate the photovoltaic power generation absorption efficiency, the rationality of output distribution, and the utilization rate of photovoltaic energy. Energy storage performance indicators can be evaluation indicators used to measure the degree of energy storage unit loss, lifetime decay rate, economic benefits of photovoltaic-energy storage synergy, and the rationality of charging and discharging operation. Global optimization indicators can be comprehensive evaluation indicators formed by integrating the three categories of sub-indicators: frequency performance indicators, energy performance indicators, and energy storage performance indicators, which can coordinate and integrate multiple objectives of grid security, energy efficiency, and energy storage lifetime optimization.

[0022] In this embodiment, the operating data of each unit in the cluster can be collected in real time by the acquisition device in the cluster. Combined with the grid frequency regulation requirements and virtual inertia control logic, the frequency deviation of the grid connection point and the inertia support capability are determined to determine the frequency performance index. The corresponding energy performance index is determined based on the actual photovoltaic output, theoretical power generation, photoelectric conversion efficiency and output distribution balance of all photovoltaic units. The corresponding energy storage performance index is determined based on the energy storage unit parameters and photovoltaic-storage operation data. Then, these three types of index data are weighted and fused to obtain the corresponding global optimization index.

[0023] S130. Based on the global optimization index, construct the photovoltaic objective function and the energy storage objective function respectively. Combine the fractional optimization method and the multi-objective genetic algorithm to jointly solve the photovoltaic objective function and the energy storage objective function to obtain the global control sequence.

[0024] The photovoltaic objective function can be an optimization function constructed by taking the operating characteristics of the photovoltaic unit as the objective, combined with output stability, power generation efficiency, and grid connection constraints. The energy storage objective function can be an optimization function constructed by taking the operating characteristics of the energy storage unit as the core, combined with charging and discharging losses, lifetime protection, and photovoltaic-storage synergistic benefits. The fractional-order optimization method can be an optimization modeling method based on fractional-order calculus theory, which, unlike integer-order modeling, can accurately characterize complex industrial processes with nonlinear, hysteresis, and fluctuating characteristics. The multi-objective genetic algorithm can be an intelligent optimization algorithm based on an evolutionary iteration mechanism, which iteratively seeks the optimal solution for multiple mutually constrained optimization objectives. The global control sequence can refer to the set of control commands that include the output commands of each photovoltaic unit, the charging and discharging commands of the energy storage unit, and the adjustment parameters. In this embodiment, the fractional-order optimization algorithm can be embedded into the local controller, and the photovoltaic and energy storage objective functions can be reconstructed. Then, the optimal solution sequence can be generated by optimizing and iterating subproblems through a non-dominated sorting genetic algorithm, thus obtaining the global control sequence.

[0025] In this embodiment, global optimization indicators are used as constraints and optimization guidelines. A photovoltaic objective function, centered on stable photovoltaic output and efficient grid integration, and an energy storage objective function, centered on low energy loss and high synergistic benefits, are constructed. Then, a fractional optimization method is employed, using Caputo fractional derivatives to reconstruct the two objective functions. A multi-objective genetic algorithm is used to generate candidate solution sets corresponding to the dual objective functions. Multi-objective evolutionary iteration is then carried out using fractional iteration relationships and fractional crowding constraints. When the iteration accuracy meets preset requirements, the optimal solution set is output. The power commands and control parameters of each unit within the solution set are analyzed and integrated to form a global control sequence covering the entire cluster.

[0026] S140. Verify the global control sequence based on the set verification method to obtain the target control sequence, and convert the target control sequence into target control instructions to perform coordinated control of the target photovoltaic cluster.

[0027] The setting verification method can refer to an algorithm that verifies the control instructions contained in the global control sequence. In this embodiment, the setting verification method can be a digital twin-driven instruction verification algorithm. The target control sequence can refer to the final optimized control solution set obtained after verification or correction by the setting verification method. The target control instruction can be understood as a control instruction that can be recognized and executed by various control hardware devices.

[0028] In this embodiment, the optimal solution sequence can be verified by a digital twin-driven instruction verification algorithm, and the verification result can be compensated by a sliding mode control algorithm to obtain the corresponding target control sequence. Then, the target control sequence is processed by format conversion and signal conversion to be transformed into target control instructions that can be recognized by each power electronic device. By sending the control instructions to the photovoltaic inverter, energy storage PCS, and cluster controller, the operating status of each photovoltaic and energy storage unit is uniformly regulated to realize photovoltaic output regulation, energy storage coordinated charging and discharging, and cluster power coordinated management and control.

[0029] The technical solution of this invention involves determining the suitability of a target photovoltaic (PV) unit when it is detected that the PV unit is connected to the distribution network, updating the original PV cluster topology based on the suitability, and obtaining the target PV cluster. It then acquires the operating data of the target PV cluster, determines frequency performance indicators, energy performance indicators, and energy storage performance indicators based on the operating data, and determines global optimization indicators based on these indicators. Based on the global optimization indicators, it constructs PV objective functions and energy storage objective functions respectively, and jointly solves these functions using a fractional-order optimization method and a multi-objective genetic algorithm to obtain a global control sequence. Finally, it verifies the global control sequence using a set verification method to obtain the target control sequence, and transforms the target control sequence into target control commands for coordinated control of the target PV cluster. This technical solution, through the global optimization objective constructed from multi-dimensional indicators, and subsequently determines the target control commands, achieves coordinated regulation of PV and energy storage in the PV cluster, improves the rationality of global optimization, and enhances the overall efficiency and stability of the cluster operation.

[0030] Example 2 Figure 2 This is a flowchart of a photovoltaic cluster control method based on distributed optimization according to Embodiment 2 of the present invention. This embodiment is based on the above embodiment with optimizations. Specifically, the optimization is as follows: when a target photovoltaic unit is detected to be connected to the distribution network, the adaptability of the target photovoltaic unit is determined, including: when a target photovoltaic unit is detected to be connected to the distribution network, multiple photovoltaic units within a set range are determined through a routing algorithm, and a communication link is established between the target photovoltaic unit and the multiple photovoltaic units; based on the communication link, the original parameter set of the target photovoltaic unit is sent to the gateway node, and the adaptability between the target photovoltaic unit and the gateway node is determined based on the original parameter set and the cluster baseline parameters. Figure 2 As shown, the method includes: S210. When the target photovoltaic unit is detected to be connected to the distribution network, multiple photovoltaic units within a set range are determined by the routing algorithm, and a communication link is established between the target photovoltaic unit and the multiple photovoltaic units.

[0031] The routing algorithm can refer to a networking algorithm used for distributed node addressing, neighbor node discovery, and communication path planning. The defined range can be a pre-defined range. In this embodiment, the defined range can be used to limit the screening boundaries of neighboring photovoltaic units. The communication link can refer to a bidirectional data transmission channel between photovoltaic units used for transmitting operating parameters and exchanging data.

[0032] In this embodiment, a new photovoltaic (PV) unit is detected. An improved AODV routing algorithm is used to search for multiple existing PV units within a set range, centered on the target PV unit. A bidirectional communication link is established between the target PV unit and multiple surrounding PV units, enabling interconnection of operational data and control signaling. It should be noted that traditional AODV is a routing protocol designed for wireless ad hoc networks. It initiates route discovery only when nodes need to communicate. After the target node replies with a route reply packet, a path is established, and the path validity is monitored through a route maintenance mechanism. However, in distributed PV clusters, the locations of PV units are fixed, but their numbers change dynamically, and the communication environment is subject to industrial electromagnetic interference. The indiscriminate broadcasting and simple path selection mechanisms of traditional AODV are not suitable for this environment.

[0033] Specifically, in this embodiment, when a new photovoltaic (PV) unit is received connecting to the distribution network, a link quality detection frame is sent and broadcast to devices within the physical layer's reachable range. Surrounding existing nodes receive this frame and return a scan response frame, which may include their own signal strength and current CPU load rate. The new PV unit records all responding existing PV information to form an initial neighbor candidate list. Then, the new PV unit generates a routing request frame and sends it to the initial neighbor candidate list. It obtains the normalized value of the current signal strength and the normalized value of the CPU load rate of each PV node in the initial neighbor candidate list, and calculates the forwarding weight by weighted summation. If the forwarding weight of a PV node is greater than or equal to a preset forwarding weight threshold, the current routing request frame is forwarded to other nodes. If the forwarding weight of a PV node is less than the preset forwarding weight threshold, the routing request frame is discarded and not forwarded. When the gateway node receives the routing request frame, it generates a path confirmation frame, rolls back to the new PV unit, and establishes a PV cluster path connection, thus constructing the corresponding communication link.

[0034] It should be noted that the normalized signal strength value is the original physical quantity received by the photovoltaic unit from the neighboring node, and the value obtained after normalization. The CPU load rate is a real-time operating status parameter of the photovoltaic unit's embedded controller, reflecting the current CPU workload, and is obtained after normalizing the original value. In this embodiment, multi-hop forwarding is used to allow the routing request frame to reach the gateway node, completing the establishment of the communication link between the new photovoltaic unit and the cluster; when the gateway node receives the routing request frame, it generates a path confirmation frame, rolls back to the new photovoltaic unit, and establishes a path connection for the photovoltaic cluster.

[0035] S220. Based on the communication link, send the original parameter set of the target photovoltaic unit to the gateway node, and determine the compatibility between the target photovoltaic unit and the gateway node based on the original parameter set and the cluster reference parameters.

[0036] The original parameter set can refer to the collection of basic parameters of the target photovoltaic unit itself. In this embodiment, the original parameter set may include parameters such as rated power, output voltage, operating frequency, and control response delay. The gateway node can refer to the edge aggregation and management node of the photovoltaic cluster, which is responsible for the aggregation of parameters of each distributed unit, protocol conversion, and cluster collaboration determination. The cluster baseline parameters can be baseline parameter information pre-stored in the gateway node, used to adapt to the unified standard parameter thresholds and control constraint benchmarks for the stable operation of the current photovoltaic cluster.

[0037] In this embodiment, the original parameter set of the target photovoltaic unit can be sent to the gateway node based on the established local communication link. The power deviation rate, voltage deviation rate, frequency deviation, and response delay difference are determined based on the original parameter set and the cluster reference parameters. Then, the power deviation rate, voltage deviation rate, frequency deviation, and response delay difference are input into the BP neural network to output the fit between the new photovoltaic node and the gateway node.

[0038] It should be noted that a BP neural network can consist of an input layer, a hidden layer, and an output layer. There are two hidden layers, and the activation function is ReLU. It can be trained using 1000 sets of historical access data as the training sample set.

[0039] Specifically, in this embodiment, the new photovoltaic unit sends its original parameter set to the gateway node. ,in, Rated power, For output voltage, For operating frequency, To control response latency, cluster baseline parameters, including average rated power, are obtained through the gateway node. Standard voltage of the power grid Rated frequency of power grid and standard response delay The power deviation rate is calculated based on the original parameter set and the corresponding cluster baseline parameters. Voltage deviation rate Frequency deviation and response latency difference The calculation methods are as follows: , , and ; In this embodiment, power deviation rate, voltage deviation rate, frequency deviation, and response delay difference can be input into a BP neural network to output the fit between the new photovoltaic node and the gateway node.

[0040] S230. Update the original photovoltaic cluster topology based on the adaptability to obtain the target photovoltaic cluster.

[0041] In this embodiment, optionally, updating the original photovoltaic cluster topology based on the fit degree to obtain the target photovoltaic cluster includes: if the fit degree is greater than or equal to a preset threshold, updating the original photovoltaic cluster topology based on the target photovoltaic unit to obtain the target photovoltaic cluster; if the fit degree is less than the preset threshold, determining the parameter adjustment amount of the target photovoltaic unit, adjusting the parameters of the target photovoltaic unit based on the parameter adjustment amount to obtain the adjusted photovoltaic unit, and updating the original photovoltaic cluster topology based on the adjusted photovoltaic unit to obtain the target photovoltaic cluster.

[0042] The preset threshold refers to a pre-defined threshold for adaptability, used to measure the critical quantification value for whether a new photovoltaic unit can be directly integrated into the cluster, and can be set according to actual needs. The parameter adjustment amount refers to the specific numerical value adjusted to the original parameters of the current target photovoltaic unit. The adjusted photovoltaic unit refers to the photovoltaic unit obtained after correcting the original parameter set.

[0043] In this embodiment, the calculated fit of the target photovoltaic unit can be compared with a preset threshold. When the fit is greater than or equal to the preset threshold, it is determined that the parameters of the target photovoltaic unit are fully compatible with the existing cluster. The target photovoltaic unit can then be directly included in the cluster management node, and corresponding communication connections and electrical topology associations can be added. The topology of the original photovoltaic cluster is updated as a whole, and the target photovoltaic cluster containing the newly added unit is integrated. If the fit is less than the preset threshold, the corresponding parameter adjustment amount can be determined by comparing the deviation between the original parameters of the target photovoltaic unit and the cluster baseline parameters. Based on the parameter adjustment amount, the parameters of the new photovoltaic node are corrected to obtain the adjusted photovoltaic unit. Then, based on the adjusted photovoltaic unit, the topology of the original photovoltaic cluster is updated as a whole, and the target photovoltaic cluster containing the newly added unit is integrated.

[0044] Specifically, in this embodiment, if the fit is greater than or equal to a preset fit threshold, an access confirmation message is broadcast to neighboring nodes. The topology of the original photovoltaic cluster can be directly updated based on the target photovoltaic unit, integrating to obtain the target photovoltaic cluster. If the fit is greater than or equal to the preset fit threshold, the power deviation rate, voltage deviation rate, frequency deviation, and response delay difference are mapped to a proportional-integral adjustment algorithm. The target value of a parameter to be adjusted for a photovoltaic unit is set to T, and the actual measured value is y(k). The difference between the actual measured value and the target value is used to obtain the error. The current parameter adjustment amount is obtained based on the error amount, and its calculation method can be as follows: ; in, This is the adjustment amount for the k-th iteration. This is the proportionality coefficient. The integral coefficient is... Let be the integral term of the error, i be the iteration number, and k be the total number of iterations. The sampling period.

[0045] Then, the parameter set of the target photovoltaic unit is corrected by the determined parameter adjustment amount to obtain the adjusted photovoltaic unit. Based on the adjusted photovoltaic unit, the topology of the original photovoltaic cluster is updated as a whole and integrated to obtain the target photovoltaic cluster.

[0046] In this embodiment, by setting up a dual-path access mechanism with the adaptability threshold as the judgment boundary, a seamless plug-and-play access for photovoltaic units is achieved, ensuring dynamic adaptive updates of the cluster topology and improving the flexibility and operational compatibility of distributed photovoltaic cluster networking.

[0047] S240. Obtain the operating data of the target photovoltaic cluster, determine the frequency performance index, energy performance index and energy storage performance index based on the operating data, and determine the global optimization index based on the frequency performance index, energy performance index and energy storage performance index.

[0048] In this embodiment, optionally, the operating data includes the rated frequency data of the photovoltaic unit and the virtual inertia reference coefficient; determining the frequency performance index based on the operating data includes: acquiring the real-time frequency data of the grid connection point and the inertia adjustment coefficient; determining the instantaneous frequency deviation based on the real-time frequency data and the rated frequency data; determining the virtual inertia coefficient corresponding to each photovoltaic unit based on the instantaneous frequency deviation, the inertia adjustment coefficient and the virtual inertia reference coefficient; and determining the frequency performance index based on the virtual inertia coefficient and the instantaneous frequency deviation.

[0049] In this context, the rated frequency data refers to the preset standard frequency parameters during normal grid-connected operation of the photovoltaic unit. The virtual inertia reference coefficient can be a pre-set system reference coefficient used to measure the virtual inertia support capability of the photovoltaic unit. The grid connection point refers to the interface node connecting the photovoltaic cluster and the distribution network. The real-time frequency data can be the actual operating frequency value of the distribution network currently being collected in real time by sensors at the grid connection point. The inertia adjustment coefficient is a coefficient used to adjust the virtual inertia response sensitivity of the photovoltaic unit, which can be dynamically set according to the grid frequency stability requirements. The instantaneous frequency deviation refers to the difference between the real-time frequency of the grid connection point and the rated frequency of the photovoltaic unit at a certain moment, used to reflect the instantaneous fluctuation of the grid frequency. The virtual inertia coefficient is a coefficient used to characterize the virtual inertia support capability that a single photovoltaic unit can provide. In this embodiment, the virtual inertia coefficient can be determined by the virtual inertia reference coefficient, the instantaneous frequency deviation, and the inertia adjustment coefficient. It is understood that the larger the virtual inertia coefficient, the stronger the frequency regulation support capability of the photovoltaic unit.

[0050] In this embodiment, real-time frequency data of the photovoltaic cluster connected to the distribution network can be collected in real time by a frequency sensor installed at the grid connection point, and a preset inertia adjustment coefficient can be retrieved from the gateway node. The difference between the collected real-time frequency data at the grid connection point and the rated frequency data of the photovoltaic unit is calculated to obtain the instantaneous frequency deviation. Based on the virtual inertia reference coefficient of each photovoltaic unit, combined with the magnitude of the instantaneous frequency deviation and the weight of the inertia adjustment coefficient, the virtual inertia coefficient corresponding to each photovoltaic unit is determined by a preset quantitative calculation formula. Then, the virtual inertia coefficient and instantaneous frequency deviation of each photovoltaic unit are weighted and summed, and normalized to obtain the frequency performance index.

[0051] In this embodiment, the frequency performance index can also be called inertia-coordinated frequency stability. By adopting a quantitative numerical form, the closer the value is to the preset optimal value, the stronger the frequency support capability of the photovoltaic cluster and the smaller the frequency fluctuation. Inertia-coordinated frequency stability is used to characterize the ability of a distributed photovoltaic cluster to provide frequency stability support for the power grid through inverter virtual inertia adjustment, as well as the coordination and consistency of inertia adjustment among photovoltaic units within the cluster. From a physical point of view, traditional synchronous generators rely on the mechanical inertia of their own rotating parts to spontaneously smooth frequency changes when the grid frequency fluctuates. However, photovoltaics are connected to the grid through power electronic devices and do not have natural mechanical inertia. They need to simulate virtual inertia through inverter control. The value of this indicator directly reflects two aspects: the higher the value, the better the matching degree between the virtual inertia adjustment of each photovoltaic unit and the grid frequency fluctuation, the more consistent the inertia response of each unit in the cluster, the absence of internal friction problems such as local over-adjustment and local under-adjustment, the stronger the ability to smooth grid frequency fluctuations, the less likely the grid frequency will exceed the limit risk, and the better the support effect of the entire photovoltaic cluster on the frequency stability of the grid; conversely, the lower the value, the worse the coordination of inertia adjustment of the photovoltaic cluster, the weaker the ability to suppress grid frequency fluctuations, and the worse the grid frequency stability.

[0052] Specifically, in this embodiment, real-time operating data of all photovoltaic units and grid connection points is collected. The real-time operating data includes the real-time frequency of the grid connection point, the rated power of each photovoltaic unit, and the virtual inertia reference coefficient of the photovoltaic unit inverter. Maximum permissible frequency deviation of the power grid and inertia adjustment coefficient Extract the rated frequency from the database, and calculate the instantaneous frequency deviation by subtracting the real-time frequency from the rated frequency at the network point. The virtual inertia coefficient of each photovoltaic unit is obtained based on the instantaneous frequency deviation and the inertia adjustment coefficient. The calculation method can be as follows: ; Where i is the number of the photovoltaic unit. It should be noted that this calculation formula is based on the adaptation logic that the smaller the frequency deviation, the stronger the inertia reserve, and the larger the frequency deviation, the more conservative the inertia.

[0053] In this embodiment, the instantaneous frequency deviation and virtual inertia coefficient of each photovoltaic unit can be weighted and summed, and then normalized to obtain the inertia-coordinated frequency stability, i.e., the frequency performance index. This configuration allows for accurate capture of grid frequency fluctuations, improves the real-time frequency regulation response of the photovoltaic cluster, and uses multi-parameter coupling to construct the frequency performance index, providing a reliable basis for global optimization.

[0054] In this embodiment, optionally, the operating data includes real-time output data and theoretical power generation data of the photovoltaic units; determining energy performance indicators based on the operating data includes: determining the output ratio data of each photovoltaic unit based on the real-time output data and theoretical power generation data; determining the global output distribution disorder based on the output ratio data and the information entropy relationship; determining the global energy efficiency based on the photoelectric conversion efficiency of each photovoltaic unit and the energy entropy relationship; and determining the energy performance indicators based on the global output distribution disorder and the global energy efficiency.

[0055] Real-time output data refers to the actual power generation output per unit time during the actual operation of the photovoltaic unit. Theoretical power generation data refers to the theoretical power generation data determined based on parameters such as the rated power, irradiance conditions, and conversion efficiency of the photovoltaic unit. Output ratio data refers to the proportion of the real-time output of a single photovoltaic unit to the total output of all photovoltaic units. The information entropy relationship can be a calculation logic relationship used to quantify the degree of disorder in data distribution, which can transform the disordered state of output distribution into a quantifiable value. The global output distribution disorder degree can measure the degree of balance in the output distribution of all photovoltaic units. Understandably, the lower the global output distribution disorder degree, the more reasonable the output allocation; conversely, the output allocation needs to be adjusted. Photovoltaic conversion efficiency refers to the efficiency with which the photovoltaic unit converts solar energy into electrical energy. The energy entropy relationship can be used to quantitatively evaluate the balance of energy utilization, converting energy efficiency into a quantifiable value. Global energy efficiency can be a comprehensive evaluation value of the energy conversion efficiency of all photovoltaic units.

[0056] In this embodiment, real-time output data of each photovoltaic unit can be collected, and theoretical power generation data of each photovoltaic unit can be obtained. The average real-time output and average theoretical output during the collection period can be calculated. Then, the output ratio of each photovoltaic unit can be determined by a formula. Substituting the output ratio of each photovoltaic unit into the information entropy calculation formula, the disorder of the output distribution of each unit can be obtained. In this embodiment, the theoretical output of incident light of each photovoltaic unit can be divided by the corresponding average theoretical output to obtain the actual photoelectric conversion efficiency. Then, the conversion efficiency data can be substituted into the energy entropy calculation formula to obtain the global energy efficiency. The global energy efficiency and global disorder are weighted and summed to obtain the energy performance index.

[0057] In this embodiment, the frequency performance index can also be called dual-entropy ordered energy efficiency. Dual-entropy ordered energy efficiency is a quantitative indicator that combines the two dimensions of information entropy and energy entropy to comprehensively characterize the orderliness and energy utilization efficiency of the distributed photovoltaic cluster's operating state. It is a core evaluation standard that takes into account both the cluster's operational stability and the efficiency of new energy utilization. The information entropy dimension is used to quantify the uniformity of the output distribution of each photovoltaic unit within the cluster, which is the global disorder of the system operation. The more uneven the output distribution and the greater the local output fluctuations, the higher the information entropy value, the higher the system's disorder, and the greater the risk of impact on the power grid. The energy entropy dimension is used to quantify the photoelectric conversion efficiency and energy utilization level of the entire photovoltaic cluster. The higher the photoelectric conversion efficiency and the less energy transmission loss, the higher the energy entropy value, and the more fully the new energy is utilized. The value of dual-entropy ordered energy efficiency is the result of a weighted integration of the two dimensions. The higher the value, the better the dual-entropy synergy effect: on the one hand, the more uniform the output distribution of each photovoltaic unit within the cluster, the more orderly the system operation, and the lower the system stability risk caused by output fluctuations; on the other hand, the higher the photoelectric conversion and energy utilization efficiency of the entire cluster, the better the new energy absorption effect. Conversely, the lower the value, the more likely the cluster has uneven output distribution, high degree of disorder in operation, and poor stability.

[0058] Specifically, in this embodiment, the real-time output of all photovoltaic units is collected through the inverter data interface. Then, the theoretical power output of the incident light for each photovoltaic unit is obtained through an irradiance sensor; the average real-time power output during the acquisition period is calculated. The average theoretical output was calculated, and the output percentage of each photovoltaic unit was obtained. The calculation method can be as follows: ; Among them, P i It can be the output percentage of the i-th photovoltaic unit; N is the total number of photovoltaic units.

[0059] This embodiment can embed the output ratio into the information entropy formula to obtain the global disorder. The calculation method can be as follows: ,in, The actual photoelectric conversion efficiency is obtained by dividing the theoretical output of incident light from each photovoltaic unit by the corresponding average theoretical output. The actual photoelectric conversion efficiency is embedded into the energy entropy formula to calculate the global energy efficiency. The calculation method can be as follows: .

[0060] It should be noted that the global disorder is quantified by information entropy, reflecting the uniformity of the output distribution of each unit in the photovoltaic cluster. The lower the value, the smaller the output fluctuation and the more regular and orderly the system operation. The global energy efficiency is characterized by energy entropy, reflecting the overall high efficiency of photoelectric conversion and energy utilization. The higher the value, the less energy loss and the more fully utilized. The two are weighted and summed by preset entropy weighting coefficients to obtain the dual-entropy ordered energy efficiency. This weighted integration is not a simple superposition, but rather a balance between orderly operation and efficient energy use through coefficients, avoiding the impact of disorder fluctuations on system stability and preventing resource waste caused by inefficient energy use. Finally, the dual-entropy ordered energy efficiency is obtained by weighting and summing the global energy efficiency and global disorder, which is the frequency performance index.

[0061] In this embodiment, by introducing information entropy and energy entropy, the disorder of power output distribution and energy efficiency are quantified, the rationality of power output allocation of each photovoltaic unit is accurately measured, and the power output balance of the cluster is optimized.

[0062] In this embodiment, optionally, the operating data includes the energy storage parameters of the energy storage unit and the photovoltaic-storage operating data; determining the energy storage performance index based on the operating data includes: determining the total energy storage loss data and the photovoltaic-storage synergistic benefit data based on the energy storage parameters and the photovoltaic-storage operating data; and determining the energy storage performance index based on the total energy storage loss data and the photovoltaic-storage synergistic benefit data.

[0063] Here, energy storage parameters refer to relevant parameters of the energy storage unit. In this embodiment, energy storage parameters may include rated capacity, current remaining power ratio, and remaining lifetime parameters. Photovoltaic-energy storage operation data refer to related operating condition data of photovoltaic and energy storage collaborative operation, such as photovoltaic output fluctuation data, energy storage charge and discharge commands, joint peak shaving duration, and power interaction data. In this embodiment, photovoltaic-energy storage operation data may include energy storage charge and discharge power and time-of-use electricity price parameters. Total energy storage loss data refers to the quantified value of the total loss generated by the energy storage unit during charge and discharge. Photovoltaic-energy storage synergy benefit data refers to the quantified value of the comprehensive benefits brought about by the coordinated operation of photovoltaic and energy storage to improve new energy consumption, smooth power fluctuations, and reduce grid dispatch costs.

[0064] In this embodiment, the operating parameters of the energy storage device and the operating data of the photovoltaic-storage system are obtained, and the operating loss characteristics of energy storage and the economic benefit characteristics of photovoltaic-storage synergy are quantified respectively. The optimal trade-off solution between the two is solved based on the multi-objective optimization method, and the comprehensive gain performance index of photovoltaic-storage operation, i.e., the energy storage performance index, is obtained by fusion calculation.

[0065] Specifically, in this embodiment, energy storage unit parameters and photovoltaic-storage operation data are collected in real time. The energy storage unit parameters include rated capacity, current remaining power ratio (SOC), and remaining lifetime. The data on photovoltaic and energy storage operation includes energy storage charging and discharging power and time-of-use electricity pricing. The rainflow counting method is used to quantify the lifespan loss of a single charge-discharge cycle. A complete charge-discharge cycle is identified based on the change in the percentage of remaining charge. It should be noted that a cycle is defined as the state of charge (SOC) decreasing from a to b and then increasing back to a. The maximum percentage of remaining charge within a complete charge-discharge cycle is obtained. and minimum remaining power ratio This is used to calculate the depth of charge-discharge cycles. The calculation method can be as follows: ; Where j is the energy storage unit number, and the single-cycle lifetime loss is obtained based on the depth of charge / discharge and remaining lifetime. The calculation method can be as follows: Where kc is the cycle lifetime coefficient, the current cycle number is extracted, and the energy storage cycle lifetime loss is obtained based on the cycle number and the single cycle lifetime loss; the synergistic benefits of photovoltaic and energy storage are obtained through energy storage charging and discharging power and time-of-use pricing. The calculation method can be as follows: ; Where t is the sampling time number, and T is the total sampling time point number. This represents the total discharge power of the energy storage system. The total charging power of energy storage. The time interval between two adjacent sampling times.

[0066] Then, by using the NSGA-Ⅲ algorithm, several sets of energy storage charging and discharging strategies are randomly generated as Pareto non-dominated solution sets. Each solution corresponds to a set of photovoltaic-storage synergistic benefits and energy storage cycle life loss. The loss-benefit comprehensive index of each non-dominated solution is calculated, and the solution with the smallest value is selected as the optimal solution under the current operating conditions. The corresponding optimal energy storage cycle life loss and optimal photovoltaic-storage synergistic benefits are extracted. The optimal energy storage cycle life loss and optimal photovoltaic-storage synergistic benefits are integrated according to the balance formula to obtain the quantitative value of photovoltaic-storage balanced gain, that is, the energy storage performance index.

[0067] It should be noted that a lower value for the quantification of the photovoltaic-storage synergy gain indicates a better synergy effect, achieving Pareto optimality between returns and lifespan degradation. This means maximizing the synergistic gains of photovoltaic and energy storage while controlling the lifespan degradation of energy storage at an optimal level, resulting in the highest cost-effectiveness over the entire lifecycle of photovoltaic-storage synergy. Conversely, a higher value indicates a worse synergistic effect, either due to frequent deep charging and discharging of energy storage in pursuit of short-term gains, leading to excessive battery lifespan degradation. This embodiment, through such a setting, can take into account both energy storage loss and synergistic benefits, quantify energy storage degradation and operating costs, and extend the service life of energy storage, so as to improve the economic efficiency and adaptability of photovoltaic-storage joint operation.

[0068] Furthermore, in this embodiment, the specific method for determining the global optimization index based on frequency performance index, energy performance index, and energy storage performance index can be to use a reinforcement learning algorithm to dynamically output the weights corresponding to inertia-coordinated frequency stability, dual-entropy ordered energy efficiency, and photovoltaic-storage balanced gain of the policy network. The calculation method can be as follows: Where e is the weight number and Norm is the normalization function. For the boundary constraint function, the dynamic weights are clipped to [0.05, 0.4]. For the dynamic strategy network, KL is a set of real-time state vectors, including instantaneous frequency deviation, real-time output, and remaining lifetime. Then, based on the dynamic weights, the global optimization index is obtained by weighted summation of inertia cooperative frequency stability, dual-entropy ordered energy efficiency, and optical-storage balanced gain, and the corresponding weights.

[0069] S250. Based on the global optimization index, construct the photovoltaic objective function and the energy storage objective function respectively. Combine the fractional optimization method and the multi-objective genetic algorithm to jointly solve the photovoltaic objective function and the energy storage objective function to obtain the global control sequence.

[0070] In this embodiment, the specific method for determining the photovoltaic objective function based on the global optimization index can be as follows: ; Where K1 is the theoretical maximum value of the power grid inertial support function. To provide real-time output for the previous moment.

[0071] In this embodiment, the specific method for determining the energy storage objective function based on the global optimization index can be as follows: ; in, This is the quantization value for the optical-storage equalization gain. The proportion of energy storage output to total output. The overall efficiency of energy storage charging and discharging.

[0072] In this embodiment, optionally, the photovoltaic objective function and the energy storage objective function are jointly solved by combining fractional optimization method and multi-objective genetic algorithm to obtain a global control sequence, including: using fractional optimization method to reconstruct the photovoltaic objective function and the energy storage objective function in fractional order to obtain the photovoltaic fractional objective function and the energy storage fractional objective function; and using multi-objective genetic algorithm to perform collaborative optimization iteration on the photovoltaic fractional objective function and the energy storage fractional objective function to obtain a global control sequence.

[0073] The photovoltaic objective function can be a basic evaluation function with optimization directions including stable photovoltaic output, power generation absorption efficiency, and grid connection constraints. The energy storage objective function can be a basic evaluation function with optimization directions including energy storage charge and discharge losses, lifetime degradation, and operational constraints. Fractional-order reconstruction can be a process of structural modification and dynamic feature fitting of the original integer-order objective function using fractional-order operators. The photovoltaic fractional-order objective function can be an optimization objective function adapted to the dynamic fluctuation characteristics of photovoltaics after fractional-order modification. The energy storage fractional-order objective function can be an optimization objective function adapted to the nonlinear loss characteristics of energy storage after fractional-order modification. In this embodiment, the multi-objective genetic algorithm can also be called a non-dominated sorting genetic algorithm.

[0074] In this embodiment, fractional calculus theory (Caputo fractional derivative) can be introduced as an optimization tool to perform fractional-order mapping and reconstruction correction on the constraint terms, error terms, and dynamic adjustment terms of the photovoltaic objective function and the energy storage objective function. This establishes a fractional-order mathematical model that takes into account the nonlinear variation law of the system, resulting in fractional-order objective functions for both photovoltaic and energy storage. Then, using the two sets of fractional-order objective functions as dual optimization objectives, a multi-objective genetic algorithm is used to generate an initial candidate solution set. Combining fractional-order fitness evaluation, fractional-order crowding constraints, and fractional-order iterative update rules, multi-objective evolutionary optimization iteration is continuously carried out. After satisfying the preset convergence conditions, the optimal solution set for multi-objective equilibrium is selected and mapped analytically into power regulation, output allocation, and charge / discharge control commands for each photovoltaic unit and energy storage unit, integrating them into a unified global control sequence.

[0075] In this embodiment, optionally, a multi-objective genetic algorithm is used to perform collaborative optimization iteration on the photovoltaic fractional-order objective function and the energy storage fractional-order objective function to obtain a global control sequence. This includes: determining the candidate solution set corresponding to the photovoltaic fractional-order objective function and the energy storage fractional-order objective function through the multi-objective genetic algorithm; performing optimization iteration based on the candidate solution set and the fractional-order iteration relationship, combined with the fractional-order crowding constraint; and outputting the optimal solution set and mapping it to generate a global control sequence after satisfying the preset convergence condition.

[0076] Here, the candidate solution set can refer to a sample of solutions consisting of multiple sets of candidate control parameters and adjustment strategies. The fractional-order iterative relationship can refer to the iterative update logic relationship established based on fractional-order operation rules. The fractional-order crowding constraint can be a solution set constraint condition set for fractional-order optimization scenarios, which can be used to quantitatively measure the density of the solution set distribution. The preset convergence condition can refer to a pre-defined criterion for determining iteration termination. For example, the preset convergence condition can be a condition setting corresponding to the number of iterations, the objective function error threshold, and the fitness stability threshold, which can be set according to actual needs. The optimal solution set can refer to a set of multi-objective non-dominated optimal control schemes that satisfy the dual fractional-order objective equilibrium requirements of photovoltaics and energy storage while taking into account multiple operational constraints.

[0077] In this embodiment, based on the population initialization mechanism of the multi-objective genetic algorithm, initial population individuals covering the control parameters of each photovoltaic and energy storage unit are randomly generated under the constraint boundaries of the photovoltaic fractional objective function and the energy storage fractional objective function. Each population individual is treated as a group of control schemes to be optimized, and they are uniformly classified and integrated to form a candidate solution set that simultaneously meets the needs of solving the dual fractional objective functions. Then, based on the candidate solution set, a fractional iterative relationship adapted to the dynamic characteristics of the photovoltaic and energy storage system is introduced to complete the dynamic update and deduction of the solution set. At the same time, a fractional congestion constraint is introduced to screen and suppress repeated solutions and homogeneous solution sets during the iteration process. When the preset convergence condition is reached, the iteration loop is terminated. From the iterative solution set, a non-dominated optimal solution set that takes into account both photovoltaic operating efficiency and energy storage loss benefits is selected. The optimization parameters in the optimal solution set are format converted and logically mapped to be parsed into control instructions that can be executed by each photovoltaic unit and energy storage unit, and integrated to form a global control sequence.

[0078] Specifically, in this embodiment, an initial solution population is randomly generated based on the photovoltaic objective function and the energy storage objective function, and then the Caputo fractional derivative is embedded into the initial solution population. The construction logic of the Caputo fractional derivative can be as follows: Where D is the Caputo fractional derivative, and fm is the objective term, including inertia-coordinated frequency stability, bientropy ordered energy efficiency, and optical-storage balanced gain. For the Gamma function, For fractional order, Let be the integer derivative of the objective term; obtain the fractional objective function of photovoltaic and the fractional objective function of energy storage, solve the fractional objective function of photovoltaic and the fractional objective function of energy storage to obtain the corresponding fractional fitness, sort them from high to low based on the fractional fitness, and retain the top k high-quality solutions.

[0079] Then, the fractional crowding degree is calculated based on the Caputo fractional derivative, which can be done as follows: ,in, For different high-quality solutions, The distance is an integer Euclidean distance. If the fractional congestion of a high-quality solution is less than the preset congestion threshold, it is discarded, and several local fractional non-dominated solution sets are retained. Each photovoltaic unit and energy storage unit sends the local fractional non-dominated solution set to the neighboring unit and receives the fractional non-dominated solution from the neighbor. The local solution and the neighboring solution are merged to generate a temporary solution set. Finally, the local solution set is updated based on the temporary solution set and the fractional-order iterative formula. The calculation method is as follows: ,in, The fractional gradient step size, For fractional inertial weights, The optimal solution for the neighbors. This is the local optimal solution. Let s be the fractional fitness and s be the iteration number. The fractional fitness of the updated solution is recalculated, and a new set of local non-dominated solutions is selected. The fractional fitness of all photovoltaic units and energy storage units at the next time step is subtracted from the fractional fitness of the current time step to obtain the fitness difference value. If the fitness difference value is less than the preset fitness difference threshold, the preset convergence condition is considered to be met, and the iteration terminates. Each unit selects the minimum solution from the final set of local fractional non-dominated solutions as the local optimal solution through the entropy weight fractional fitness weighting method, which yields the global control sequence.

[0080] S260. Verify the global control sequence based on the set verification method to obtain the target control sequence, and convert the target control sequence into target control instructions to perform coordinated control of the target photovoltaic cluster.

[0081] In this embodiment, the optimal solution sequence is verified by a digital twin-driven instruction verification algorithm, and the execution link of the verified result is compensated by a sliding mode control algorithm to obtain the corresponding target control sequence. Specifically, a high-precision digital twin can be constructed for the photovoltaic unit and the energy storage unit. The local optimal solution is used as the pre-execution input of the twin. The twin loads the current real-time operating data, simulates the instruction execution process at the next moment, and outputs the predicted data sequence. It is then determined whether the predicted data sequence meets the preset constraints. If it does, the optimal solution that has passed the verification is directly output. Otherwise, a correction equation is constructed based on the twin sensitivity matrix SD.

[0082] In this embodiment, the twin sensitivity matrix can be calculated as follows: ,in, Output for twins, The correction equation for the optimal solution vector is: ,in, To obtain the correction value, the gradient descent method is used to solve the correction value equation. The correction value is then embedded into the optimal solution to obtain the corrected solution, thereby obtaining the target control sequence. In this embodiment, the corrected solution can be converted into hardware control commands that can be directly executed by the physical units and sent to each unit to adjust the power, photovoltaic output, and grid status.

[0083] It should be noted that the preset constraints are that photovoltaic units must meet grid security constraints and hardware operation constraints, and energy storage units must meet safety operation constraints or lifespan protection constraints. All constraints are quantified into specific numerical thresholds as the criteria for judging twin prediction data sequences. For example, grid security constraints are that the grid connection point frequency is between 49.5 and 50.5 Hz and the voltage is within ±5% of the rated range.

[0084] This embodiment comprehensively considers multiple objectives, including inertia-coordinated frequency stabilization, dual-entropy ordered energy efficiency, and photovoltaic-storage balanced gain. Reinforcement learning is used to dynamically allocate weights, avoiding the limitations of fixed weights and single objectives. This allows for adaptive adaptation to changing operating conditions, resulting in more comprehensive and dynamically adaptable optimization objectives and improved global optimization rationality. Furthermore, a fractional-order optimization algorithm is embedded in the local controller to characterize dynamic characteristics such as photovoltaic power output fluctuations and energy storage nonlinear losses. Combined with a non-dominated sorting genetic algorithm for iterative solving, the solution accuracy is improved, enhancing the reliability of the optimal solution. Digital twin verification proactively avoids constraint violation risks, and sliding mode control compensates for hardware nonlinear deviations, ensuring safe and reliable instruction execution.

[0085] The technical solution of this invention involves the following steps: When a target photovoltaic (PV) unit is detected to be connected to the distribution network, a routing algorithm is used to determine multiple PV units within a set range, and a communication link is established between the target PV unit and the multiple PV units. Based on the communication link, the original parameter set of the target PV unit is sent to the gateway node. Based on the original parameter set and the cluster baseline parameters, the fit between the target PV unit and the gateway node is determined. Based on the fit, the original PV cluster topology is updated to obtain the target PV cluster. The operating data of the target PV cluster is acquired, and frequency performance indicators, energy performance indicators, and energy storage performance indicators are determined based on the operating data. A global optimization index is then determined based on the frequency performance indicators, energy performance indicators, and energy storage performance indicators. Based on the global optimization index, PV objective functions and energy storage objective functions are constructed respectively. The PV objective functions and energy storage objective functions are jointly solved using a fractional-order optimization method and a multi-objective genetic algorithm to obtain a global control sequence. The global control sequence is verified using a set verification method to obtain the target control sequence. The target control sequence is then converted into target control commands for coordinated control of the target PV cluster. This technical solution constructs a global optimization target through multi-dimensional indicators, and then determines the target control command, realizing the coordinated regulation of photovoltaic and energy storage in the photovoltaic cluster, improving the rationality of global optimization, and enhancing the efficiency and stability of the overall operation of the cluster.

[0086] Example 3 Figure 3 This is a schematic diagram of a photovoltaic cluster control device based on distributed optimization according to Embodiment 3 of the present invention. Figure 3 As shown, the device includes: The cluster update module 310 is used to determine the adaptability of the target photovoltaic unit when the target photovoltaic unit is detected to be connected to the distribution network, and update the original photovoltaic cluster topology based on the adaptability to obtain the target photovoltaic cluster. The indicator determination module 320 is used to acquire the operating data of the target photovoltaic cluster, determine the frequency performance indicators, energy performance indicators and energy storage performance indicators based on the operating data, and determine the global optimization indicators based on the frequency performance indicators, energy performance indicators and energy storage performance indicators. The global control sequence determination module 330 is used to construct photovoltaic objective functions and energy storage objective functions based on global optimization indices, and to jointly solve the photovoltaic objective functions and energy storage objective functions by combining fractional optimization method and multi-objective genetic algorithm to obtain the global control sequence; The collaborative control module 340 is used to verify the global control sequence based on a set verification method, obtain the target control sequence, and convert the target control sequence into target control commands to perform collaborative control on the target photovoltaic cluster.

[0087] Optionally, the cluster update module 310 is specifically used for: when a target photovoltaic unit is detected to be connected to the distribution network, determining multiple photovoltaic units within a set range through a routing algorithm, and establishing a communication link between the target photovoltaic unit and the multiple photovoltaic units; sending the original parameter set of the target photovoltaic unit to the gateway node based on the communication link, and determining the compatibility between the target photovoltaic unit and the gateway node based on the original parameter set and the cluster baseline parameters.

[0088] Optionally, the cluster update module 310 is specifically used for: if the fit is greater than or equal to a preset threshold, updating the original photovoltaic cluster topology based on the target photovoltaic unit to obtain the target photovoltaic cluster; if the fit is less than the preset threshold, determining the parameter adjustment amount of the target photovoltaic unit, adjusting the parameters of the target photovoltaic unit based on the parameter adjustment amount to obtain the adjusted photovoltaic unit, and updating the original photovoltaic cluster topology based on the adjusted photovoltaic unit to obtain the target photovoltaic cluster.

[0089] Optionally, the operating data includes the rated frequency data of the photovoltaic unit and the virtual inertia reference coefficient; the index determination module 320 is specifically used for: acquiring the real-time frequency data and inertia adjustment coefficient of the grid connection point; determining the instantaneous frequency deviation based on the real-time frequency data and rated frequency data; determining the virtual inertia coefficient corresponding to each photovoltaic unit based on the instantaneous frequency deviation, inertia adjustment coefficient and virtual inertia reference coefficient; and determining the frequency performance index based on the virtual inertia coefficient and instantaneous frequency deviation.

[0090] Optionally, the operating data includes real-time output data and theoretical power generation data of the photovoltaic units.

[0091] The index determination module 320 is specifically used for: determining the output ratio data of each photovoltaic unit based on real-time output data and theoretical power generation data; determining the global output distribution disorder based on the output ratio data and the information entropy relationship; determining the global energy efficiency based on the photoelectric conversion efficiency of each photovoltaic unit and the energy entropy relationship; and determining the energy performance index based on the global output distribution disorder and the global energy efficiency.

[0092] Optionally, the operational data includes the energy storage parameters of the energy storage unit and the photovoltaic-storage operational data.

[0093] The indicator determination module 320 is specifically used to: determine the total energy storage loss data and the photovoltaic-storage synergy benefit data based on energy storage parameters and photovoltaic-storage operation data; and determine the energy storage performance indicators based on the total energy storage loss data and the photovoltaic-storage synergy benefit data.

[0094] Optionally, the global control sequence determination module 330 includes: a reconstruction unit, used to perform fractional-order reconstruction of the photovoltaic objective function and the energy storage objective function using a fractional-order optimization method to obtain the photovoltaic fractional-order objective function and the energy storage fractional-order objective function; and an optimization iteration unit, used to perform collaborative optimization iteration of the photovoltaic fractional-order objective function and the energy storage fractional-order objective function using a multi-objective genetic algorithm to obtain the global control sequence.

[0095] Optionally, the optimization iteration unit is specifically used to: determine the candidate solution set corresponding to the photovoltaic fractional objective function and the energy storage fractional objective function through a multi-objective genetic algorithm; perform optimization iteration based on the candidate solution set and the fractional iteration relationship, combined with the fractional crowding constraint; and output the optimal solution set and map it to generate a global control sequence after satisfying the preset convergence condition.

[0096] The photovoltaic cluster control device based on distributed optimization provided in this embodiment of the invention can execute the photovoltaic cluster control method based on distributed optimization provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0097] Example 4 Figure 4This is a schematic diagram of an electronic device according to Embodiment 4 of the present invention. The electronic device 410 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0098] like Figure 4 As shown, the electronic device 410 includes at least one processor 411 and a memory, such as a read-only memory (ROM) 412 or a random access memory (RAM) 413, communicatively connected to the at least one processor 411. The memory stores computer programs executable by the at least one processor. The processor 411 can perform various appropriate actions and processes based on the computer program stored in the ROM 412 or loaded from storage unit 418 into the RAM 413. The RAM 413 may also store various programs and data required for the operation of the electronic device 410. The processor 411, ROM 412, and RAM 413 are interconnected via a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.

[0099] Multiple components in electronic device 410 are connected to I / O interface 415, including: input unit 416, such as keyboard, mouse, etc.; output unit 417, such as various types of displays, speakers, etc.; storage unit 418, such as disk, optical disk, etc.; and communication unit 419, such as network card, modem, wireless transceiver, etc. Communication unit 419 allows electronic device 410 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0100] Processor 411 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 411 performs the various methods and processes described above, such as a photovoltaic cluster control method based on distributed optimization.

[0101] In some embodiments, the distributed optimization-based photovoltaic cluster control method can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 418. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 410 via ROM 412 and / or communication unit 419. When the computer program is loaded into RAM 413 and executed by processor 411, one or more steps of the distributed optimization-based photovoltaic cluster control method described above can be performed. Alternatively, in other embodiments, processor 411 can be configured to perform the distributed optimization-based photovoltaic cluster control method by any other suitable means (e.g., by means of firmware).

[0102] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0103] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0104] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0105] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0106] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0107] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0108] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0109] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A photovoltaic cluster control method based on distributed optimization, characterized in that, include: When a target photovoltaic unit is detected to be connected to the distribution network, the adaptability of the target photovoltaic unit is determined, and the original photovoltaic cluster topology is updated based on the adaptability to obtain the target photovoltaic cluster. The operation data of the target photovoltaic cluster is obtained, and the frequency performance index, energy performance index and energy storage performance index are determined based on the operation data. The global optimization index is determined based on the frequency performance index, the energy performance index and the energy storage performance index. Based on the global optimization index, photovoltaic objective functions and energy storage objective functions are constructed respectively. The photovoltaic objective functions and energy storage objective functions are jointly solved by combining fractional optimization method and multi-objective genetic algorithm to obtain global control sequence. The global control sequence is verified based on the established verification method to obtain the target control sequence, and the target control sequence is converted into target control instructions to perform coordinated control of the target photovoltaic cluster.

2. The method according to claim 1, characterized in that, When a target photovoltaic unit is detected to be connected to the distribution network, determining the suitability of the target photovoltaic unit includes: When a target photovoltaic unit is detected to be connected to the power distribution network, multiple photovoltaic units within a set range are determined through a routing algorithm, and a communication link is established between the target photovoltaic unit and the multiple photovoltaic units. Based on the communication link, the original parameter set of the target photovoltaic unit is sent to the gateway node, and the compatibility between the target photovoltaic unit and the gateway node is determined based on the original parameter set and the cluster baseline parameters.

3. The method according to claim 1, characterized in that, The process of updating the original photovoltaic cluster topology based on the adaptability to obtain the target photovoltaic cluster includes: If the adaptability is greater than or equal to a preset threshold, the original photovoltaic cluster topology is updated based on the target photovoltaic unit to obtain the target photovoltaic cluster. If the fit is less than the preset threshold, the parameter adjustment amount of the target photovoltaic unit is determined, and the parameters of the target photovoltaic unit are adjusted based on the parameter adjustment amount to obtain the adjusted photovoltaic unit. The original photovoltaic cluster topology is updated based on the adjusted photovoltaic unit to obtain the target photovoltaic cluster.

4. The method according to claim 1, characterized in that, The operating data includes the rated frequency data and virtual inertia reference coefficient of the photovoltaic unit; Determining the frequency performance index based on the operational data includes: Acquire real-time frequency data and inertia adjustment coefficient at the grid connection point; The instantaneous frequency deviation is determined based on the real-time frequency data and the rated frequency data; The virtual inertia coefficient for each photovoltaic unit is determined based on the instantaneous frequency deviation, the inertia adjustment coefficient, and the virtual inertia reference coefficient. The frequency performance index is determined based on the virtual inertia coefficient and the instantaneous frequency deviation.

5. The method according to claim 1, characterized in that, The operational data includes the real-time output data and theoretical power generation data of the photovoltaic unit; The energy performance indicators are determined based on the operational data, including: The output ratio of each photovoltaic unit is determined based on the real-time output data and the theoretical power generation data. The global power distribution disorder is determined based on the power output ratio data and the information entropy relationship. The global energy efficiency is determined based on the photoelectric conversion efficiency of each photovoltaic unit and the energy entropy relationship. The energy performance index is determined based on the global power output distribution disorder and the global energy efficiency.

6. The method according to claim 1, characterized in that, The operational data includes the energy storage parameters of the energy storage unit and the photovoltaic-storage operational data; Determining the energy storage performance indicators based on the operational data includes: Based on the energy storage parameters and the photovoltaic-storage operation data, determine the total energy storage loss data and the photovoltaic-storage synergistic benefit data; The energy storage performance indicators are determined based on the total energy storage loss data and the photovoltaic-energy storage synergy benefit data.

7. The method according to claim 1, characterized in that, By combining fractional-order optimization methods and multi-objective genetic algorithms to jointly solve the photovoltaic objective function and the energy storage objective function, a global control sequence is obtained, including: The fractional-order optimization method is used to reconstruct the photovoltaic objective function and the energy storage objective function in fractional order, resulting in the photovoltaic fractional-order objective function and the energy storage fractional-order objective function. A global control sequence is obtained by performing collaborative optimization iterations on the photovoltaic fractional objective function and the energy storage fractional objective function using a multi-objective genetic algorithm.

8. The method according to claim 7, characterized in that, A global control sequence is obtained by iteratively optimizing the fractional-order objective functions of photovoltaic and energy storage using a multi-objective genetic algorithm, including: The candidate solution sets corresponding to the photovoltaic fractional objective function and the energy storage fractional objective function are determined by the multi-objective genetic algorithm. Based on the relationship between the candidate solution set and the fractional-order iteration, and combined with the fractional-order crowding constraint, optimization iteration is performed. After satisfying the preset convergence condition, the optimal solution set is output and mapped to generate a global control sequence.

9. A photovoltaic cluster control device based on distributed optimization, characterized in that, include: The cluster update module is used to determine the adaptability of the target photovoltaic unit when it is detected that the target photovoltaic unit is connected to the distribution network, and update the original photovoltaic cluster topology based on the adaptability to obtain the target photovoltaic cluster. The indicator determination module is used to acquire the operating data of the target photovoltaic cluster, determine the frequency performance indicator, energy performance indicator and energy storage performance indicator based on the operating data, and determine the global optimization indicator based on the frequency performance indicator, the energy performance indicator and the energy storage performance indicator. The global control sequence determination module is used to construct photovoltaic objective functions and energy storage objective functions based on the global optimization index, and to jointly solve the photovoltaic objective functions and the energy storage objective functions by combining fractional optimization method and multi-objective genetic algorithm to obtain the global control sequence; The collaborative control module is used to verify the global control sequence based on a set verification method to obtain the target control sequence, and to convert the target control sequence into target control commands to perform collaborative control on the target photovoltaic cluster.

10. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the photovoltaic cluster control method based on distributed optimization as described in any one of claims 1-8.

Images