A photovoltaic district optimization regulation method, system, device and medium

Abstract

The application discloses a kind of photovoltaic substation optimization regulation methods, systems, equipment and medium, belong to power distribution network technical field, including substation intelligent fusion terminal, fusion multi-source real-time data online diagnosis exception and execute hierarchical collaborative optimization regulation strategy;Energy storage device is used to store and release the substation electric energy according to regulation instruction;Photovoltaic grid-connected wisdom switch is used to control the rigidity of photovoltaic power supply and / or off-grid state;Real-time monitoring substation state;When exception occurs, energy storage device is adjusted preferentially;If energy storage is insufficient, then through photovoltaic information acquisition unit, flexible power regulation is carried out to photovoltaic inverter;If flexible regulation is invalid or situation is urgent, then through fusion terminal local pre-decision, fast rigid off-grid is executed by wisdom switch.The application realizes local fast response not dependent on main station by the collaborative regulation strategy of multiple intelligent devices, improves the safety, economy of photovoltaic substation operation and the accommodation capacity to new energy.

Description

Technical Field

[0001] This invention relates to the field of power distribution network technology, specifically to a method, system, equipment, and medium for optimizing and controlling photovoltaic power distribution areas. Background Technology

[0002] With strong national support for the development of new energy sources, distributed photovoltaic (PV) power generation has experienced rapid growth due to its advantages such as cleanliness, environmental friendliness, and local consumption, and has been widely connected to distribution network areas. However, under the policy of "connecting as much as possible," the randomness, volatility, and intermittent output characteristics of large-scale distributed PV pose serious challenges to the safe and stable operation of traditional distribution substations. The main problems include: power backflow is prone to occur during peak PV output and off-peak load periods, leading to reverse overload of substation transformers; concentrated PV output may cause voltage exceeding limits at the feeder end, affecting the power quality for users; and the substations face significant problems of "inability to see clearly and poor interaction" with distributed PV, lacking effective monitoring and refined control methods.

[0003] To address these challenges, existing technologies have been explored, but they generally suffer from deep-seated problems such as insufficient coordination and delayed response. For example, in some solutions, energy storage, intelligent distribution terminal, and smart switch devices often operate independently, forming functional silos; energy storage regulation is mostly based on preset strategies or historical data, failing to form a dynamic closed loop with the real-time sensing of distribution area conditions by the integrated terminal; at the same time, most regulation logic heavily relies on commands issued by the cloud master station, resulting in long communication links and high latency. Faced with rapidly changing operating conditions within the distribution area, it is difficult to achieve a rapid response within seconds, and it cannot achieve local self-healing in emergency situations, posing safety hazards. In addition, in terms of control methods, existing technologies lack a multi-resource integration framework; control methods are often simplistic and crude, either involving independent charging and discharging of energy storage or rigid "one-size-fits-all" shutdowns, lacking a refined strategy that combines flexibility and tiered progression. Currently, although the industry has recognized the importance of flexible power regulation of photovoltaic inverters, it still faces two major barriers in engineering practice: First, the inverter communication interface is often monopolized by the manufacturer's monitoring system, making it difficult for third-party devices to access it; second, the communication protocols of various brands of inverters have poor compatibility, making flexible regulation technically feasible but difficult to promote and apply on a large scale.

[0004] Therefore, there is an urgent need for a deep collaborative optimization and control scheme that can coordinate multiple adjustable resources within a photovoltaic distribution area, so as to achieve refined monitoring and flexible control of the photovoltaic distribution area, thereby maximizing grid security, improving photovoltaic absorption capacity, and effectively protecting user interests. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, this invention aims to provide a photovoltaic power grid optimization and control system and method based on the collaboration of integrated terminals, energy storage devices and smart switches. The core idea of ​​this invention is to construct a three-layer integrated collaborative control framework of "edge decision-making layer - execution adjustment layer - hardware control layer", and to achieve second-level precise management and optimized operation of photovoltaic power grid anomalies through multi-source data fusion and intelligent algorithms.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for optimizing and controlling a photovoltaic distribution area, comprising,

[0008] Through the intelligent integrated terminal of the photovoltaic power distribution area, the operation data of the photovoltaic power distribution area is collected in real time, including grid parameters, photovoltaic output data, and energy storage device status data. Based on the multi-source sensing data, the intelligent integrated terminal of the photovoltaic power distribution area determines whether there are preset operation anomalies in the photovoltaic power distribution area and generates control commands online at the edge side. If the operation anomaly is determined to exist, the intelligent integrated terminal of the photovoltaic power distribution area executes hierarchical collaborative control.

[0009] As a preferred embodiment of the photovoltaic power station area optimization and control method described in this invention, the hierarchical collaborative control includes: when a preset abnormal operation is detected in the photovoltaic power station area, the power is adjusted by controlling the energy storage device first; according to the generated control command, the energy storage device is adjusted by charging and discharging first to eliminate the abnormal operation for the first time; after the initial elimination, the intelligent fusion terminal of the power station area judges again whether there is an abnormal operation.

[0010] If the abnormal operation persists after the initial elimination or the energy storage device's adjustment capacity reaches the limit, the output of the photovoltaic power supply will be flexibly adjusted, and instructions will be sent to the corresponding photovoltaic inverter to adjust the output of active or reactive power, and the abnormality will be judged again to see if it has been eliminated.

[0011] If flexible regulation cannot be implemented, is ineffective, or the abnormal operation is urgent and not resolved, the photovoltaic grid-connected smart switch of the corresponding photovoltaic power source will be connected through local pre-decision control to execute rigid off-grid control of the photovoltaic power source.

[0012] As a preferred embodiment of the photovoltaic distribution area optimization and control method described in this invention, the flexible adjustment includes: if communication is established between the photovoltaic information acquisition unit and the target photovoltaic inverter, the distribution area intelligent fusion terminal generates a flexible adjustment command, which is then forwarded to the photovoltaic inverter through the photovoltaic information acquisition unit;

[0013] If the communication protocol of the photovoltaic inverter is inconsistent with the standard protocol, a protocol conversion will be performed before forwarding; if the communication interface of the photovoltaic inverter is occupied by other devices, communication will be carried out after the communication interface is shared through an interface adapter.

[0014] As a preferred embodiment of the photovoltaic power grid optimization and control method described in this invention, the energy storage device performs power regulation by: calculating the energy storage charging and discharging reference power based on a first-order low-pass filter algorithm according to the photovoltaic power output fluctuation characteristics, and performing feedback correction in combination with the current state of charge of the energy storage to smooth the photovoltaic grid-connected power.

[0015] Based on the planned output curve of the power distribution area, the charging and discharging of energy storage is controlled to compensate for the deviation between the actual output and the planned output of photovoltaic power.

[0016] As a preferred embodiment of the photovoltaic distribution area optimization and control method described in this invention, the hierarchical collaborative control further includes a hierarchical collaborative control logic-dependent optimization decision model, which is built into the distribution area fusion terminal; if there is an operational anomaly, a distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning is used to intelligently generate control commands.

[0017] Construct a distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning, including a state space. Action space The state transition function and reward function, wherein the state transition function is the function that, after performing action A in the current state S, the power grid operating environment transitions from the current state to a new state. The probability distribution under the preset control strategy Under the guidance of the algorithm, calculate the state transition probability. : The reward function takes reducing voltage deviation and minimizing active power loss as optimization objectives.

[0018]

[0019] Where U is the set of node voltages. Belongs to U, For the active power set of the load, For load reactive power collection, Let S be the set of photovoltaic active power output under the current state. Let S be the set of reactive power outputs of the photovoltaic inverter under the current state. For a set of charged states, Let S be the set of actions of the photovoltaic inverter under the current state. This is the set of charging and discharging power actions of the energy storage device under the current state S. and These are the penalty factors for voltage deviation and active power loss, respectively. This is the voltage reference value. For the system's active power loss, Here, n is the current voltage value, n is the total number of all nodes, and i is the variable index;

[0020] An improved dual-delay deep deterministic strategy gradient is used to solve the optimization model. At the same time, the algorithm is improved by using an orthogonal initialization strategy and a priority experience replay mechanism.

[0021] The orthogonal initialization strategy generates a random matrix M that is identical in shape to the target weight matrix W, and the elements in the matrix follow a standard normal distribution. Extract from M and perform singular value decomposition on M, i.e. , among which, U M V M Let U be an orthogonal matrix, T be the transpose matrix, and Σ be the singular value matrix. M and As an orthogonal basis, a gain factor g is introduced to adjust the output variance and update the weights. The initialization formula is: ;

[0022] High-value samples are sampled first through a priority experience replay mechanism, and a dual-objective Critic network is used to minimize the overestimated target Q value. The calculation expression is as follows:

[0023]

[0024] Where y is the target Q value, r is the immediate reward, and γ is the discount factor. Let A' be the objective Critic function, and A' be the action space in state S'. Calculate the deviation between the current predicted value and the objective value, and define the temporal difference error as... , The Critic function is defined after obtaining the time-series difference error of all empirical samples. Prioritization of experience for , For the first The empirical time-series difference error, Given a small normal number, calculate the probability of the l-th experience being selected based on priority. Represented as:

[0025]

[0026] Where α is a hyperparameter representing the degree of influence of the control priority on the sampling process, and s is the number of samples in the empirical replay region. For the first The weight of each piece of experience in the sampling probability, Let s be the weight of the s-th experience in the sampling probability; simultaneously, an importance sampling weight is introduced. The correction deviation is expressed as:

[0027]

[0028] Where C is the capacity of the experience pool, and β is the hyperparameter of the intensity of the control bias correction.

[0029] Another objective of this invention is to provide a photovoltaic power station area optimization and control system.

[0030] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a photovoltaic distribution area optimization and control system, comprising: a distribution area intelligent integration terminal, an energy storage device, and a photovoltaic grid-connected smart switch;

[0031] The intelligent integrated terminal for the photovoltaic power station connects to the energy storage device and is used to collect the operation data of the photovoltaic power station and the status data of the energy storage device. It makes edge decisions based on multi-source sensing data to diagnose abnormal operation and generates control commands according to the preset collaborative optimization and control strategy.

[0032] Energy storage devices are installed in photovoltaic power distribution areas to store and release electricity in the area according to control commands.

[0033] The photovoltaic grid-connected smart switch integrates advanced sensing, protection, metering, communication and control. It connects the photovoltaic power source and the power grid and communicates with the intelligent integration terminal of the distribution area to perform rigid control of the photovoltaic power source to be connected to or disconnected from the grid under the command of the intelligent integration terminal.

[0034] The intelligent converged terminal in the distribution area is configured to execute a hierarchical collaborative control strategy.

[0035] As a preferred embodiment of the photovoltaic distribution area optimization and control system described in this invention, the photovoltaic information acquisition unit communicates with the photovoltaic inverter and the distribution area intelligent fusion terminal, respectively, for collecting the operating data of the photovoltaic inverter and forwarding the flexible adjustment command generated by the distribution area intelligent fusion terminal to the corresponding photovoltaic inverter.

[0036] An interface adapter is configured at the communication interface of a photovoltaic inverter to share the current communication interface when it is already occupied by another device, enabling communication between the photovoltaic information acquisition unit and the photovoltaic inverter.

[0037] As a preferred embodiment of the photovoltaic distribution area optimization and control system described in this invention, the distribution area intelligent fusion terminal or the photovoltaic information acquisition unit integrates a protocol conversion module, which is used to convert the standard flexible adjustment command generated by the distribution area intelligent fusion terminal into a specific communication protocol format that can be recognized by the photovoltaic inverter.

[0038] The energy storage device includes an energy management system, a battery cluster, a battery management system, and an energy storage converter; the intelligent integrated terminal of the distribution area controls the charging and discharging behavior of the energy storage device by issuing commands to the energy management system.

[0039] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the photovoltaic area optimization and control method.

[0040] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of the photovoltaic area optimization and control method described above.

[0041] The beneficial effects of the present invention are as follows:

[0042] An integrated collaborative framework is constructed that integrates terminals, energy storage devices, and smart switches, forming a three-layer architecture of "edge decision-making layer - execution and regulation layer - hardware control layer". By combining multi-source data fusion and intelligent algorithms, the system can achieve second-level precise management of voltage overruns at photovoltaic power station nodes.

[0043] Strengthening transparent monitoring and reliable control of photovoltaic grid-connected points, and introducing smart switches as key sensing and execution nodes, enables refined data monitoring and rapid rigid control of photovoltaic branches, providing a solid foundation for local autonomous regulation.

[0044] Overcoming the difficulties in implementing flexible regulation of photovoltaic systems, the introduction of interface adapters and protocol conversion modules solves the equipment heterogeneity issues of communication interface occupation and protocol inconsistencies in practical applications, thus clearing obstacles for the widespread application of flexible regulation of photovoltaic systems.

[0045] Improve the speed of anomaly handling response by realizing local pre-decision at the intelligent converged terminal side of the distribution area. The smart switch can be driven to perform rigid control without relying on the master station command, which greatly shortens the fault handling time and enhances the self-healing capability of the power grid.

[0046] To improve the absorption capacity of photovoltaic power and the quality of power supply, the optimized operation of energy storage devices and the refined management of photovoltaic output can effectively smooth out power fluctuations, reduce reverse overload and voltage over-limit occurrences, and enhance the absorption capacity of distributed photovoltaic power in the distribution area and the quality of power supply. Attached Figure Description

[0047] To more clearly illustrate the technical solutions of 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.

[0048] Figure 1 The above is a flowchart of a photovoltaic power station optimization and control method according to an embodiment of the present invention.

[0049] Figure 2 This is a system structure diagram of a photovoltaic power station area optimization and control system provided in one embodiment of the present invention.

[0050] Figure 3 This is a schematic diagram of a typical energy storage device architecture for a photovoltaic power station optimization and control method provided in one embodiment of the present invention.

[0051] Figure 4 The diagram below shows the P / Q control block diagram of an energy storage converter for a photovoltaic power distribution area optimization and control method according to an embodiment of the present invention.

[0052] Figure 5 This is a smart switch hardware architecture diagram of a photovoltaic power distribution area optimization and control method provided in one embodiment of the present invention. Detailed Implementation

[0053] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. 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 protection scope of the present invention.

[0054] Example 1, referring to Figure 1 , Figures 3-5 This is one embodiment of the present invention, which provides a method for optimizing and controlling a photovoltaic power distribution area, comprising:

[0055] S100: Through the intelligent integrated terminal of the photovoltaic distribution area, real-time operation data of the photovoltaic distribution area is collected, including grid parameters, photovoltaic output data, and status data of energy storage devices.

[0056] Based on multi-source sensing data, the S200 and the intelligent integrated terminal of the photovoltaic area online at the edge side determine whether there are preset operational anomalies in the photovoltaic area and generate control commands.

[0057] S300 If an operational anomaly is detected, the intelligent converged terminal in the distribution area will perform hierarchical collaborative control.

[0058] It should be noted that existing technologies have several drawbacks. Inverter communication interfaces are often monopolized by the manufacturer's monitoring system, making it difficult for third-party devices to access them. Furthermore, the communication protocols of different brands of inverters have poor compatibility, which makes flexible regulation technically feasible but difficult to promote and apply on a large scale.

[0059] Therefore, in response to the aforementioned problems, through steps S100-S300, a specific optimization control system and method for distribution transformer areas with distributed photovoltaic access is provided, particularly a technical solution that integrates a distribution transformer area intelligent fusion terminal, energy storage device and smart switch, and realizes deep collaboration and local autonomous control of multiple intelligent devices.

[0060] Example 2, refer to Figure 1 , Figures 3-5 This is one embodiment of the present invention, which provides a method for optimizing and controlling a photovoltaic power distribution area, comprising:

[0061] This invention integrates a smart integrated terminal for a photovoltaic power distribution area, an energy storage device, and a smart switch. The photovoltaic power distribution area optimization and control system provided by this invention follows a three-layer architecture of "edge decision-making layer - execution regulation layer - hardware control layer". It mainly includes a smart integrated terminal for a photovoltaic power distribution area deployed in the distribution area, an energy storage device, and at least one photovoltaic grid-connected smart switch installed at the photovoltaic power grid connection point.

[0062] To achieve flexible adjustment of photovoltaic output, the system can also be equipped with a photovoltaic information acquisition unit, an interface adapter, and a built-in or external protocol converter.

[0063] In this embodiment of the invention, in S100, the operation data of the photovoltaic power distribution area is collected in real time through the intelligent integrated terminal of the distribution area, including grid parameters, photovoltaic output data, and status data of energy storage devices.

[0064] It should be noted that the intelligent converged terminal in the distribution area is the "brain" of the entire control system, undertaking the core functions of the edge decision-making layer. Its main functions include:

[0065] ① Data Acquisition and Monitoring: Real-time acquisition of transformer operation data in the distribution area, including voltage, current, power, power factor, etc., is achieved through HPLC communication. Output data of each photovoltaic access point is acquired through smart switches or photovoltaic information acquisition units, and operating status data of energy storage devices is acquired through EMS.

[0066] ② Online analysis of abnormal operation: The built-in distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning analyzes and judges in real time whether there are abnormal operations such as reverse heavy overload, voltage over-limit, and three-phase imbalance in the distribution area based on the collected multi-source data;

[0067] Among them, reverse overload is judged when the reverse power of the transformer exceeds the threshold and continues for a certain period of time.

[0068] Voltage exceeding limits is judged when the voltage of a critical node exceeds the allowable range, such as the upper limit of 1.05 pu or the lower limit of 0.95 pu.

[0069] ③ Collaborative optimization control decision-making and command issuance: If the integrated terminal detects abnormal situations, such as reverse overload or voltage exceeding limits, the built-in distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning is activated to intelligently generate control commands. According to the preset principle of "energy storage first, flexibility second, and rigid bottom line", control commands are issued to the corresponding equipment.

[0070] It should also be noted that, as a core device in the regulation layer, the typical architecture of energy storage devices is as follows: Figure 3 As shown, the system mainly consists of an Energy Management System (EMS), battery clusters, a Battery Management System (BMS), and a Power Conversion System (PCS). The EMS is the control core of the energy storage device. It receives dispatch commands from the smart converged terminal of the distribution area and, in conjunction with battery status information from the BMS, such as state of charge, battery health, temperature, and voltage, controls the PCS to achieve energy storage and release. The PCS can perform P / Q control, such as... Figure 4 As shown, based on active power and no merit The commands can be precisely output, or controlled under constant voltage, etc. The specific control strategy for energy storage devices is determined by the intelligent integrated terminal of the distribution network based on deep reinforcement learning-based collaborative optimization decision-making of power grid, source, load and storage.

[0071] It should also be noted that the photovoltaic grid-connected smart switch, as a core device of the hardware control layer, has the following hardware architecture: Figure 5 The diagram shows the "sensing-execution" terminal unit of the entire control system. Installed at the grid connection point of each photovoltaic power source, it integrates advanced sensing, protection, metering, communication, and control functions to form a closed-loop function of "data acquisition - command transmission - security protection - strategy execution." The specific collaborative logic and operation process are as follows:

[0072] ① Advanced sensing and precise measurement: providing a data foundation for system decision-making;

[0073] Specifically, the device has a built-in high-precision AC sampling module that can monitor key electrical quantities such as three-phase voltage, current, active / reactive power, power factor, and frequency at the grid connection point in real time; it also has power quality assessment functions such as harmonic analysis—this is the front-end detection for the entire control system to obtain the grid connection status of photovoltaics.

[0074] The collected branch-level data will be uploaded to the intelligent fusion terminal of the distribution area in real time through the device's built-in communication module, providing accurate and real-time data input for the online judgment and decision-making model (such as collaborative optimization decision) on the edge side.

[0075] ②Multiple protections and security safeguards: Security protection is triggered based on perceived data;

[0076] Specifically, supported by advanced sensing capabilities, the device can dynamically determine the operating status of the photovoltaic grid-connected system and the power grid based on real-time collected electrical quantity data.

[0077] On the one hand, it has conventional overload and short circuit protection functions, and automatically triggers the protection mechanism when an abnormal current is detected to exceed the standard;

[0078] On the other hand, considering the characteristics of photovoltaic grid connection, it integrates professional functions such as islanding protection, over / under voltage protection, over / under frequency protection, and reverse power protection. For example, when the sensing module detects that the grid voltage / frequency exceeds the safety threshold, the device can independently, quickly, and reliably disconnect the photovoltaic system from the grid without relying on external commands, ensuring equipment safety and grid stability.

[0079] ③ Reliable communication and rapid control: Establish a "command transmission channel" to achieve precise regulation;

[0080] Specifically, the device incorporates multi-mode communication modules such as HPLC and Ethernet to establish a stable and reliable two-way communication link with the intelligent fusion terminal in the distribution area.

[0081] On the one hand, it undertakes the data transmission needs of the sensing function, that is, it uploads the collected electrical quantity data to the terminal;

[0082] On the other hand, it receives remote control commands such as opening and closing commands from the terminal.

[0083] Once the terminal generates control decisions based on sensing data, it can send instructions to the smart switch via the communication link if it is necessary to limit the grid-connected power of photovoltaics or to urgently disconnect from the grid. With its millisecond-level response and action capabilities, the device can accurately execute rigid grid-connected or disconnected control, solving the pain points of traditional switches being uncontrollable and slow to respond, and ensuring the last line of defense for grid safety.

[0084] ④ Edge-to-device collaboration and local execution: completing the closed loop of strategy implementation;

[0085] Specifically, as a key execution end in the edge-end collaborative architecture, the smart switch is a physical extension of the decision-making capabilities of the smart converged terminal (edge ​​side) in the distribution area: when the converged terminal generates a local autonomous control strategy (such as emergency disconnection) through edge computing, it can directly convert the strategy into specific operation instructions through the communication module and send them to the smart switch without relying on the master station system; the device then converts the instructions into reliable physical actions (such as disconnecting the grid-connected circuit) based on its own control and execution capabilities, realizing localized rapid self-healing control.

[0086] The collaborative model of edge decision-making and terminal execution enhances the agility and reliability of the entire system in responding to emergencies, while reducing the communication pressure on the main station and ensuring that control commands are implemented locally and take effect quickly.

[0087] In this embodiment of the invention, the S200 central distribution area intelligent fusion terminal, based on multi-source sensing data, online analyzes at the edge side whether there are preset operational anomalies in the photovoltaic distribution area and generates control commands.

[0088] In an embodiment of the present invention, if an operational abnormality is determined in S300, the intelligent converged terminal of the distribution area performs hierarchical collaborative control.

[0089] S301. When a preset abnormal operation is detected in the photovoltaic area, the power is adjusted by controlling the energy storage device first. According to the generated control command, the energy storage device is adjusted for charging and discharging to eliminate the abnormal operation for the first time. After the initial elimination, the intelligent fusion terminal of the photovoltaic area will judge again whether there is an abnormal operation.

[0090] Specifically, the power regulation of energy storage devices includes calculating the energy storage charging and discharging reference power based on a first-order low-pass filter algorithm according to the photovoltaic power output fluctuation characteristics, and making feedback corrections based on the current state of charge of the energy storage to smooth the photovoltaic grid-connected power.

[0091] Based on the planned output curve of the power distribution area, the charging and discharging of energy storage is controlled to compensate for the deviation between the actual output and the planned output of photovoltaic power.

[0092] S302. If the abnormal operation still exists after the initial elimination or the adjustment capacity of the energy storage device reaches the limit, the output of the photovoltaic power supply is flexibly adjusted, and the command is issued to the corresponding photovoltaic inverter to adjust the output of active or reactive power, and then it is judged again whether the abnormality is eliminated.

[0093] Specifically, flexible regulation includes: if communication is made with the target photovoltaic inverter through the photovoltaic information acquisition unit, the intelligent integrated terminal of the distribution area generates a flexible regulation command, which is then forwarded to the photovoltaic inverter through the photovoltaic information acquisition unit;

[0094] If the communication protocol of the photovoltaic inverter is inconsistent with the standard protocol, a protocol conversion will be performed before forwarding; if the communication interface of the photovoltaic inverter is occupied by other devices, communication will be carried out after the communication interface is shared through an interface adapter.

[0095] S303. If flexible regulation cannot be implemented, is ineffective, or the abnormal operation is urgent and not resolved, the photovoltaic grid-connected smart switch of the corresponding photovoltaic power source will be connected through local pre-decision control to execute rigid off-grid control of the photovoltaic power source.

[0096] S304. The hierarchical collaborative control logic described in this invention relies on an optimization decision-making model, which is built into the distribution area fusion terminal to realize real-time analysis of the distribution area's operating status and intelligent generation of control commands. This invention constructs a distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning. This model is based on the physical laws of the distribution network and integrates an intelligent decision-making mechanism of reinforcement learning.

[0097] The hierarchical coordinated control logic relies on the optimization decision model, which is built into the distribution area fusion terminal. If there is an operational anomaly, the distribution network source-grid-load-storage coordinated optimization model based on deep reinforcement learning is used to intelligently generate control commands.

[0098] A collaborative optimization model for power generation, grid, load and storage based on deep reinforcement learning is constructed, which includes state space, action space, state transition function and reward function;

[0099] The state space S is used to observe the voltage of each node in the distribution network, the power of all loads, the output of photovoltaic power plants, and the state of charge of energy storage devices, specifically: ,

[0100] The action space is used to adjust the reactive power output of the photovoltaic power station and the charging and discharging power of the energy storage device, and its expression is: ,

[0101] The state transition function is the function that, after performing action A in the current state S, transitions the power grid operating environment from the current state to a new state. The probability distribution under the preset control strategy Under the guidance of the algorithm, calculate the state transition probability. : ;

[0102] The reward function is the core mechanism for the desired behavior in deep reinforcement learning, and its design needs to be deeply coupled with the specific optimization objective. This invention, to achieve hierarchical control, takes reducing voltage deviation and minimizing active power loss as optimization objectives.

[0103]

[0104] Where U is the set of node voltages. Belongs to U, For the active power set of the load, For load reactive power collection, Let S be the set of photovoltaic active power output under the current state. Let S be the set of reactive power outputs of the photovoltaic inverter under the current state. For a set of charged states, Let S be the set of actions of the photovoltaic inverter under the current state. This is the set of charging and discharging power actions of the energy storage device under the current state S. and These are the penalty factors for voltage deviation and active power loss, respectively. This is the voltage reference value. For the system's active power loss, Here, n is the current voltage value, n is the total number of all nodes, and i is the variable index;

[0105] Based on deep reinforcement learning, this invention proposes a collaborative optimization model for power generation, grid, load, and storage in a distribution network. The model is solved using an improved dual-delay deep deterministic strategy gradient, which enables efficient collaborative optimization of power generation, grid, load, and storage resources in the distribution network.

[0106] Furthermore, due to the problems of "slow training convergence" and "low sample utilization efficiency" of the traditional Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm in the power distribution network scenario, this invention introduces an orthogonal initialization strategy and a priority experience replay mechanism into the TD3 algorithm, and proposes an improved algorithm.

[0107] The orthogonal initialization strategy generates random orthogonal matrices through singular value decomposition to initialize the weight matrix. This strategy ensures that the variance and gradient magnitude of activation values ​​in each layer remain stable during forward and backward propagation in deep neural networks, effectively avoiding gradient vanishing or exploding phenomena and improving the convergence speed of neural network training by more than 30%. Its unique mathematical properties guarantee that the input signal is transmitted almost without distortion between network layers, providing a fundamental support for the stable training of complex networks. A method for generating random orthogonal matrices based on singular value decomposition.

[0108] An improved dual-delay deep deterministic strategy gradient is used to solve the optimization model. At the same time, the algorithm is improved by using an orthogonal initialization strategy and a priority experience replay mechanism.

[0109] The orthogonal initialization strategy generates a random matrix M that is identical in shape to the target weight matrix W, and the elements in the matrix follow a standard normal distribution. Extract from;

[0110] Perform singular value decomposition on M, i.e. , among which, U M V M Let Σ be an orthogonal matrix, T be the transpose matrix, and Σ be the singular value matrix;

[0111] Take U M and As an orthogonal basis, a gain factor g is introduced to adjust the output variance and update the weights. The initialization formula is: .

[0112] The priority experience replay mechanism is based on the temporal difference error (TD-error) to quantify the value of samples and prioritize the sampling of high-value samples, so that the neural network can focus more on learning from "unexpected" or "exceeding current cognition" high-value experiences.

[0113] High-value samples are sampled first through a priority experience replay mechanism, and a dual-objective Critic network is used to minimize the overestimated target Q value. The calculation expression is as follows:

[0114]

[0115] Where y is the target Q value, r is the immediate reward, and γ is the discount factor. Let A' be the objective Critic function, and A' be the action space under state S'.

[0116] Calculate the deviation between the current predicted value and the target value, and define the time-series difference error as... , For the Critic function;

[0117] After obtaining the time-series difference error of all empirical samples, the first... Prioritization of experience for , For the first The empirical time-series difference error, It is a tiny positive number;

[0118] Calculate the probability of the l-th experience being selected based on priority. Represented as:

[0119]

[0120] Where α is a hyperparameter representing the degree of influence of the control priority on the sampling process, and s is the number of samples in the empirical replay region. For the first The weight of each piece of experience in the sampling probability, Let be the weight of the s-th experience in the sampling probability;

[0121] However, prioritizing sampling can alter the distribution of the original empirical data, potentially leading to learning bias and affecting algorithm convergence. Therefore, importance sampling weights are introduced. The correction deviation is expressed as follows:

[0122]

[0123] Where C is the capacity of the experience pool, and β is the hyperparameter of the intensity of the control bias correction.

[0124] In summary, the process of the photovoltaic distribution area optimization and control method described in this invention is as follows: Figure 1 As shown, this can be understood as:

[0125] The smart converged terminal in the distribution area collects and monitors data regularly.

[0126] The intelligent integrated terminal in the distribution area analyzes the operating status of the distribution area online to determine whether there are any abnormal phenomena such as reverse overload or voltage exceeding limits. If no abnormality is found, the process ends; if an abnormality is found, the control command is intelligently generated using a distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning.

[0127] Based on the control commands generated by the optimization model built into the fusion terminal, and following the preset principle of "energy storage first, flexibility second, and rigidity as a safety net", the energy storage device is first activated to regulate charging and discharging.

[0128] The smart fusion terminal in the distribution area will again determine whether there are any abnormal phenomena such as reverse overload or voltage exceeding the limit. If the abnormality has been resolved, the process ends; if the abnormality has not been resolved but the energy storage device's regulation capacity has reached the limit (such as full / empty SOC, or power reaching the upper limit).

[0129] Then, the photovoltaic flexible regulation is initiated. Based on the output power of the photovoltaic inverter generated by the optimization model built into the fusion terminal, the fusion terminal selects a suitable photovoltaic inverter and sends a power regulation command to it through the photovoltaic information acquisition unit (and interface adapter, protocol conversion module).

[0130] The converged terminal makes a third judgment on whether the above-mentioned anomaly has been resolved. If the anomaly has been resolved, the process ends; if the anomaly has not been resolved and flexible adjustment measures cannot be implemented (such as inverter communication failure or lack of adjustment support).

[0131] Finally, the photovoltaic rigid control is activated, and the integrated terminal issues a command to one or more related photovoltaic grid-connected smart switches to disconnect the connected photovoltaic power source.

[0132] The exception has been handled and the process has ended.

[0133] Example 3 is an embodiment of the present invention, which provides a method for optimizing and controlling a photovoltaic power station area. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0134] An analysis was conducted on a photovoltaic (PV) distribution area configured within the IEEE-33 node system. On a sunny midday, the total PV output was 100kW, while the local load was only 20kW. The transformer's rated capacity was 100kVA, and the permissible reverse power threshold was 30% of its rated capacity (i.e., 30kW). At this time, the actual reverse power was 100kW - 20kW = 80kW, far exceeding the threshold. Simultaneously, the voltage at the feeder end was monitored to rise to 1.062pu, exceeding the upper limit of 1.05pu.

[0135] The intelligent fusion terminal in the distribution area detected reverse overload and voltage exceeding the limit abnormality.

[0136] Energy storage regulation: The smart integrated terminal in the distribution area detected that the current SOC of the 50kW / 100kWh energy storage device configured in the area was 30%, and instructed it to charge at full power (50kW). After charging, the net reverse power of the distribution area decreased to 80kW-50kW=30kW, and the terminal voltage decreased to 1.055pu. The reverse power reached the threshold, but the voltage was still slightly exceeded, and the energy storage was already operating at full power.

[0137] Photovoltaic flexible regulation: The intelligent integrated terminal of the distribution area determined that the energy storage regulation capacity was insufficient to completely solve the voltage problem. Therefore, several communicable photovoltaic inverters with a total capacity of 60kW in the distribution area were selected. Through the photovoltaic information acquisition unit, assuming that the compatibility problem had been solved through interface adapters and protocol conversion modules, instructions were sent to these inverters, requesting that their total output be reduced by 10kW.

[0138] After the inverter responds, the total photovoltaic output drops to 90kW, the energy storage continues charging at 50kW, and the local load is 20kW. Therefore, the net reverse power of the distribution area is 90kW - 50kW - 20kW = 20kW. At this point, the terminal voltage drops to 1.045pu, returning to normal. The anomaly is resolved.

[0139] If, during the control command process generated by the optimization model built into the smart distribution terminal, some inverters fail to communicate or refuse to execute the commands, resulting in the voltage still not meeting the requirements, and the energy storage SOC gradually increases and approaches full capacity, the distribution terminal may further instruct other adjustable photovoltaic units to reduce their output. If the anomaly remains severe after all flexible measures have been exhausted, for example, a sudden and significant drop in load causing a surge in reverse power, the smart distribution terminal will be activated for a third assessment to determine whether the anomaly has been resolved.

[0140] Photovoltaic rigid control: The intelligent integrated terminal of the distribution area will instruct the grid-connected smart switches of some non-critical users or photovoltaic power sources that have not responded to flexible adjustment to trip until the operating parameters of the distribution area return to normal.

[0141] Example 4, refer to Figure 2 This is an embodiment of the present invention, and the above is an illustrative scheme of a photovoltaic power station optimization and control method. It should be noted that the technical solution of a photovoltaic power station optimization and control system and the technical solution of the photovoltaic power station optimization and control method described above belong to the same concept. Details not described in detail in the technical solution of the photovoltaic power station optimization and control system in this embodiment can be found in the description of the technical solution of the photovoltaic power station optimization and control method described above.

[0142] This embodiment provides a photovoltaic distribution area optimization and control system, including: a distribution area intelligent integration terminal, an energy storage device, and a photovoltaic grid-connected smart switch;

[0143] The intelligent integrated terminal for the photovoltaic power station connects to the energy storage device and is used to collect the operation data of the photovoltaic power station and the status data of the energy storage device. It makes edge decisions based on multi-source sensing data to diagnose abnormal operation and generates control commands according to the preset collaborative optimization and control strategy.

[0144] Energy storage devices are installed in photovoltaic power distribution areas to store and release electricity in the area according to control commands.

[0145] The photovoltaic grid-connected smart switch integrates advanced sensing, protection, metering, communication and control. It connects the photovoltaic power source and the power grid and communicates with the intelligent integration terminal of the distribution area to perform rigid control of the photovoltaic power source to be connected to or disconnected from the grid under the command of the intelligent integration terminal.

[0146] The intelligent converged terminal in the distribution area is configured to execute a hierarchical collaborative control strategy.

[0147] It also includes a photovoltaic information acquisition unit, which communicates with the photovoltaic inverter and the intelligent integrated terminal of the distribution area, to collect the operating data of the photovoltaic inverter and forward the flexible adjustment command generated by the intelligent integrated terminal of the distribution area to the corresponding photovoltaic inverter.

[0148] An interface adapter is configured at the communication interface of a photovoltaic inverter to share the current communication interface when it is already occupied by another device, enabling communication between the photovoltaic information acquisition unit and the photovoltaic inverter.

[0149] The intelligent integrated terminal of the distribution area or the photovoltaic information acquisition unit integrates a protocol conversion module, which is used to convert the standard flexible adjustment command generated by the intelligent integrated terminal of the distribution area into a specific communication protocol format that can be recognized by the photovoltaic inverter.

[0150] The energy storage device includes an energy management system, a battery cluster, a battery management system, and an energy storage converter; the intelligent integrated terminal of the distribution area controls the charging and discharging behavior of the energy storage device by issuing commands to the energy management system.

[0151] This embodiment also provides an electronic device applicable to a photovoltaic area optimization and control method, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the photovoltaic area optimization and control method proposed in the above embodiment.

[0152] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements a photovoltaic area optimization and control method as proposed in the above embodiments.

[0153] The storage medium proposed in this embodiment belongs to the same inventive concept as the photovoltaic area optimization and control method proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0154] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0155] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for optimizing and controlling a photovoltaic distribution area, characterized in that: include, Through the intelligent integrated terminal of the photovoltaic distribution area, real-time operation data of the photovoltaic distribution area is collected, including grid parameters, photovoltaic output data, and energy storage device status data; Based on multi-source sensing data, the intelligent integrated terminal in the photovoltaic area can online analyze whether there are preset operational anomalies in the photovoltaic area and generate control commands at the edge. If an operational anomaly is detected, the intelligent converged terminal in the distribution area will perform hierarchical collaborative control. The hierarchical collaborative control includes a hierarchical collaborative control logic that depends on an optimization decision model, which is built into the transformer area fusion terminal. If an operational anomaly occurs, a distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning is used to intelligently generate control commands. A distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning is constructed, which includes a state space, action space, state transition function and reward function. The state transition function is the probability distribution of transitioning from the current state to a new state. Under the guidance of a preset control strategy, the state transition probability is calculated. An improved dual-delay deep deterministic strategy gradient is used to solve the optimization model. At the same time, the algorithm is improved by using an orthogonal initialization strategy and a priority experience replay mechanism. The orthogonal initialization strategy generates a random matrix and performs singular value decomposition on the random matrix to update the weights; The priority experience replay mechanism uses a dual-objective Critic network to minimize the target Q value; it defines the priority of experiences, calculates the probability of an experience being selected based on the priority, and introduces importance sampling weights to correct the bias.

2. The photovoltaic distribution area optimization and control method as described in claim 1, characterized in that: The hierarchical coordinated control includes, when a preset abnormal operation is detected in the photovoltaic area, the power is adjusted by controlling the energy storage device first. According to the generated control command, the energy storage device is adjusted by charging and discharging first to eliminate the abnormal operation. After the initial elimination, the intelligent fusion terminal of the photovoltaic area will judge again whether there is an abnormal operation. If the abnormal operation persists after the initial elimination or the energy storage device's adjustment capacity reaches the limit, the output of the photovoltaic power supply will be flexibly adjusted, and instructions will be sent to the corresponding photovoltaic inverter to adjust the output of active or reactive power, and the abnormality will be judged again to see if it has been eliminated. If flexible regulation cannot be implemented, is ineffective, or the abnormal operation is urgent and not resolved, the photovoltaic grid-connected smart switch of the corresponding photovoltaic power source will be connected through local pre-decision control to execute rigid off-grid control of the photovoltaic power source.

3. The photovoltaic distribution area optimization and control method as described in claim 2, characterized in that: The flexible adjustment includes, if communication is made with the target photovoltaic inverter through the photovoltaic information acquisition unit, the intelligent fusion terminal of the distribution area generates a flexible adjustment command, which is forwarded to the photovoltaic inverter through the photovoltaic information acquisition unit; If the communication protocol of the photovoltaic inverter is inconsistent with the standard protocol, a protocol conversion will be performed before forwarding; if the communication interface of the photovoltaic inverter is occupied by other devices, communication will be carried out after the communication interface is shared through an interface adapter.

4. The photovoltaic distribution area optimization and control method as described in claim 3, characterized in that: The energy storage device performs power regulation by calculating the energy storage charging and discharging reference power based on a first-order low-pass filter algorithm according to the photovoltaic power output fluctuation characteristics, and performing feedback correction in combination with the current state of charge of the energy storage to smooth the photovoltaic grid-connected power. Based on the planned output curve of the power distribution area, the charging and discharging of energy storage is controlled to compensate for the deviation between the actual output and the planned output of photovoltaic power.

5. The photovoltaic distribution area optimization and control method as described in claim 4, characterized in that: The construction of the distribution network source-grid-load-storage collaborative optimization model based on deep reinforcement learning includes the state space. Action space State transition function and reward function; The state transition function is the function that, after performing action A in the current state S, the power grid operating environment transitions from the current state to a new state. The probability distribution under the preset control strategy Under the guidance of the algorithm, calculate the state transition probability. : The reward function takes reducing voltage deviation and minimizing active power loss as its optimization objectives, R: Where U is the set of node voltages. Belongs to U For the active power set of the load, For load reactive power collection, Let S be the set of photovoltaic active power output under the current state. Let S be the set of reactive power outputs of the photovoltaic inverter under the current state. For a set of charged states, Let S be the set of actions of the photovoltaic inverter under the current state. This is the set of charging and discharging power actions of the energy storage device under the current state S. and These are the penalty factors for voltage deviation and active power loss, respectively. This is the voltage reference value. For the system's active power loss, Here, n is the current voltage value, n is the total number of all nodes, and i is the variable index; The orthogonal initialization strategy generates a random matrix M that is identical in shape to the target weight matrix W, and the elements in the matrix follow a standard normal distribution. Extract from M and perform singular value decomposition on M, i.e. , among which, U M V M Let U be an orthogonal matrix, T be the transpose matrix, and Σ be the singular value matrix. M and As an orthogonal basis, a gain factor g is introduced to adjust the output variance and update the weights. The initialization formula is: ; High-value samples are sampled first through a priority experience replay mechanism, and a dual-objective Critic network is used to minimize the overestimated target Q value. The calculation expression is as follows: Where y is the target Q value, r is the immediate reward, and γ is the discount factor. Let A' be the objective Critic function, and A' be the action space in state S'. Calculate the deviation between the current predicted value and the objective value, and define the temporal difference error as... , For the Critic function, after obtaining the time-series difference errors of all empirical samples, the priority of the l-th empirical rule is defined. for , The time-series difference error of the lth empirical rule is... Given a small normal number, calculate the probability of the l-th experience being selected based on priority. Represented as: Where α is a hyperparameter representing the degree of influence of the control priority on the sampling process, and s is the number of samples in the empirical replay area; Let l be the weight of the l-th experience in the sampling probability. Let s be the weight of the s-th experience in the sampling probability; the introduction of importance sampling weights The correction deviation is expressed as: Where C is the capacity of the experience pool, and β is the hyperparameter of the intensity of the control bias correction.

6. A photovoltaic power distribution area optimization and control system, employing the photovoltaic power distribution area optimization and control method as described in any one of claims 1 to 5, characterized in that, include: Smart integrated terminals for distribution areas, energy storage devices, and smart switches for grid-connected photovoltaic systems; The intelligent integrated terminal for the photovoltaic power station connects to the energy storage device and is used to collect the operation data of the photovoltaic power station and the status data of the energy storage device. It makes edge decisions based on multi-source sensing data to diagnose abnormal operation and generates control commands according to the preset collaborative optimization and control strategy. Energy storage devices are installed in photovoltaic power distribution areas to store and release electricity in the area according to control commands. The photovoltaic grid-connected smart switch integrates advanced sensing, protection, metering, communication and control. It connects the photovoltaic power source and the power grid and communicates with the intelligent integration terminal of the distribution area. It can rigidly control the grid connection or off-grid of the photovoltaic power source according to the instructions of the intelligent integration terminal of the distribution area. The intelligent converged terminal in the distribution area is configured to execute a hierarchical collaborative control strategy.

7. A photovoltaic distribution area optimization and control system as described in claim 6, characterized in that, It also includes a photovoltaic information acquisition unit, which communicates with the photovoltaic inverter and the intelligent integrated terminal of the distribution area, to collect the operating data of the photovoltaic inverter and forward the flexible adjustment command generated by the intelligent integrated terminal of the distribution area to the corresponding photovoltaic inverter. An interface adapter is configured at the communication interface of a photovoltaic inverter to share the current communication interface when it is already occupied by another device, enabling communication between the photovoltaic information acquisition unit and the photovoltaic inverter.

8. A photovoltaic distribution area optimization and control system as described in claim 7, characterized in that: It also includes a protocol conversion module integrated in the intelligent integrated terminal of the distribution area or the photovoltaic information acquisition unit, which is used to convert the standard flexible adjustment command generated by the intelligent integrated terminal of the distribution area into a specific communication protocol format that can be recognized by the photovoltaic inverter. The energy storage device includes an energy management system, a battery cluster, a battery management system, and an energy storage converter; the intelligent integrated terminal of the distribution area controls the charging and discharging behavior of the energy storage device by issuing commands to the energy management system.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the photovoltaic area optimization and control method according to any one of claims 1 to 5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the photovoltaic area optimization and control method according to any one of claims 1 to 5.

Images