Voltage balance control method and system in photovoltaic power distribution network based on DDPG algorithm

Abstract

The invention discloses a voltage balance control method and system in a photovoltaic power distribution network based on a DDPG algorithm, and relates to the technical field of power system automation and optimization control, and the method comprises the steps: collecting operation data, and building a distributed photovoltaic power distribution network voltage control model; establishing a reactive voltage optimization model of the power distribution network; and generating a voltage balance control strategy based on the reactive voltage optimization model of the power distribution network. According to the invention, the reactive output of the photovoltaic inverter is dynamically adjusted through the power distribution network reactive voltage optimization model based on the DDPG algorithm, so that the voltage of the power distribution network is controlled. A reactive voltage optimization control strategy is provided for solving the power grid safety problem caused by high-permeability photovoltaic grid connection. According to the strategy, the voltage out-of-limit problem of all nodes of the power distribution network system is controlled by adjusting the reactive power output of the grid-connected photovoltaic inverter, and the voltage on each node in the system is ensured to be maintained in a safe range, so that the operation safety of the power distribution network after photovoltaic grid connection is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system automation and optimization control, and in particular to a voltage balancing control method and system in a photovoltaic distribution network based on a DDPG algorithm. Background Art

[0002] With the large-scale access of distributed photovoltaic power sources to distribution networks, the randomness and volatility of their output have had a significant impact on the voltage stability of distribution networks.

[0003] If the large-scale grid connection of photovoltaic power sources is combined with load changes, it is easier to make the voltage of the distribution network unstable. For the problem of local voltage exceeding the limit of the power grid caused by the intermittent and volatile photovoltaic output, the traditional active and reactive control methods have become incapable of coping. For example, in the active control method, reducing the active power of some photovoltaic outputs will reduce the distribution network's ability to absorb photovoltaics, while the investment and maintenance costs of the method of using distributed energy storage for active regulation are too high; in the reactive control method, the voltage regulating equipment represented by capacitor banks (CBs) has a slow response speed and is difficult to deal with the voltage fluctuation problem caused by the rapid fluctuation of photovoltaic output, while the voltage regulating equipment represented by static synchronous compensator (SVG) has a fast response speed, but the investment cost is too high. In order to better solve a series of problems induced by high photovoltaic penetration, such as overvoltage faults, severe voltage fluctuations, and over-action of on-load tap changers, distributed photovoltaic inverters are allowed to use their own reactive power output to participate in the voltage control of the distribution network for the first time. Photovoltaic inverters have reactive power output capability and can provide reactive power compensation for distribution networks when necessary. Compared with other traditional reactive power compensation devices, photovoltaic inverters have faster response speeds and more flexible control methods, and can provide high-quality power services to the power grid in a long-term and stable manner. Therefore, it is of great significance to carry out research on voltage balancing control methods for distributed photovoltaic distribution networks based on deep learning. First, the ability of grid-connected inverters to output reactive power is analyzed for the grid security issues brought about by high-penetration photovoltaic grid-connected. Secondly, considering the large fluctuation of photovoltaic output at different times, the reactive power output of photovoltaic inverters is dynamically adjusted through deep reinforcement learning algorithms. The control of distribution network voltage is of great significance for improving the safety, stability and economy of distribution network operation. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is that due to the randomness and volatility of the output of distributed photovoltaic power sources, the voltage stability of the distribution network is affected, especially when photovoltaic power sources are connected to the grid on a large scale and load changes are superimposed, the problem of voltage instability is more prominent. The intermittent and volatile nature of photovoltaic output leads to local voltage exceeding the limit of the power grid, and traditional active and reactive power control methods are difficult to effectively deal with. Reducing the active power output of photovoltaics will reduce the absorption capacity of photovoltaics, and the method of using distributed energy storage for active power regulation is too costly.

[0006] In order to solve the above technical problems, the present invention provides the following technical solution: a voltage balancing control method in a photovoltaic distribution network based on a DDPG algorithm, which comprises the following steps:

[0007] Collect operation data and establish a distributed photovoltaic distribution network voltage control model. Establish a distribution network reactive power voltage optimization model. Generate a voltage balancing control strategy based on the distribution network reactive power voltage optimization model.

[0008] As a preferred solution of the voltage balancing control method in the photovoltaic distribution network of the DDPG algorithm described in the present invention, wherein: the operating data is the operating data of the grid-connected inverter in the distributed photovoltaic distribution network.

[0009] The distributed photovoltaic distribution network voltage control model combines power flow constraints in the distribution network and inverter reactive output constraints.

[0010] As a preferred solution of the voltage balancing control method in the photovoltaic distribution network of the DDPG algorithm described in the present invention, the objective function of the voltage control model of the distributed photovoltaic distribution network is expressed as:

[0011]

[0012] Among them, P loss It represents the active power loss of the distribution network system after the PV inverter is activated at time t.

[0013] The power flow constraint is expressed as,

[0014]

[0015] Among them, P PV,i and Q PV,i They represent the active and reactive power generated by the photovoltaic power source on the grid-connected node i, P L,i and Q L,i They represent the active and reactive power of the load on node i, U i and U j denote the voltage amplitudes at nodes i and j, respectively, ij Represents the voltage phase angle difference between two adjacent nodes, G ij and Bij They represent the conductance and susceptance of the line respectively.

[0016] The reactive output constraint is expressed as,

[0017]

[0018] Where S represents the rated capacity of the grid-connected inverter, P PV Indicates the active output of the photovoltaic power source, Q PV Indicates the reactive power compensation actually provided by the grid-connected inverter, Q min and Q min They respectively represent the minimum and maximum reactive power compensation provided by the grid-connected inverter.

[0019] As a preferred solution of the voltage balancing control method in the photovoltaic distribution network of the DDPG algorithm described in the present invention, the distribution network reactive voltage optimization model is based on the DDPG algorithm, the photovoltaic inverter in the distributed photovoltaic distribution network is integrated into the DDPG algorithm as an intelligent agent, the voltage of each node in the distributed photovoltaic distribution network is used as the state input of the DDPG algorithm, and the voltage of each node is used as the controlled object of the model to obtain the state space, and the state space is expressed as:

[0020] s=[U1,...U i , ...U N , P1, ...P i , ...P N , Q1, ...Q i ,...Q N ]

[0021] Among them, U i , P i , Q i They represent the voltage value, active power and reactive power of the i-th node respectively.

[0022] As a preferred solution of the voltage balancing control method in the photovoltaic distribution network of the DDPG algorithm described in the present invention, the DDPG algorithm includes an actor network and a critic network, wherein the actor network is used to generate the action of the agent, and the agent selects the optimal action value according to the shared observation results of the current state and the action at the next moment. The action space in the model is set to the reactive output value of each grid-connected inverter as follows,

[0023]

[0024] Among them, a i represents the action of the i-th photovoltaic inverter, A i represents the action search space of the ith agent, S irepresents the capacity of the i-th photovoltaic inverter, P i，PV represents the active output of the ith photovoltaic power source, a min and a max They respectively represent the minimum and maximum action values ​​actually provided by the grid-connected inverter agent.

[0025] As a preferred solution of the voltage balancing control method in the photovoltaic distribution network of the DDPG algorithm described in the present invention, the DDPG algorithm puts the voltage constraint and the network loss into the reward function together. If the node voltage is not within the specified safety range, a penalty term M is added to the reward function of the DDPG algorithm. The reward function is expressed as:

[0026] r t =-(M1P loss +M2ΔV t )

[0027] represents the reward value at time t, ΔV t It represents the voltage exceeding the limit of the node at time t, and M1 and M2 are penalty factors.

[0029] As a preferred solution of the voltage balancing control method in the photovoltaic distribution network of the DDPG algorithm described in the present invention, the generated voltage balancing control strategy includes dynamically adjusting the reactive output of the photovoltaic inverter through the reactive voltage optimization model of the distribution network based on the DDPG algorithm, and controlling the action of the grid-connected inverter through the DDPG algorithm after the photovoltaic power source is connected to the grid, so as to control the node voltage within a safe range.

[0030] Another object of the present invention is to provide a voltage balancing control system in a photovoltaic distribution network based on a DDPG algorithm, which can monitor the operating status of the distribution network in real time and dynamically adjust the reactive output of the photovoltaic inverter, thereby solving the problems of slow response speed and insufficient control flexibility of existing voltage control methods when dealing with the volatility and randomness of distributed photovoltaic power output.

[0031] To solve the above technical problems, the present invention provides the following technical solutions: a voltage balancing control system in a photovoltaic distribution network based on a DDPG algorithm, comprising a data acquisition module, a DDPG algorithm module and a voltage balancing control strategy generation module.

[0032] The data acquisition module is responsible for collecting the operating data of the grid-connected inverters in the distributed photovoltaic power distribution network.

[0033] The DDPG algorithm module selects the optimal action value based on the shared observation results of the current state and the action at the next moment.

[0034] The voltage balancing control strategy generation module is responsible for dynamically adjusting the reactive output of the photovoltaic inverter, controlling the action of the grid-connected inverter through the DDPG algorithm, and keeping the node voltage within a safe range.

[0035] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the voltage balancing control method in a photovoltaic distribution network of the DDPG algorithm as described above are implemented.

[0036] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the voltage balancing control method in a photovoltaic distribution network using the DDPG algorithm as described above.

[0037] Beneficial effects of the present invention: A voltage balancing control method in a distributed photovoltaic distribution network based on the DDPG algorithm of the present invention controls the voltage of the distribution network by dynamically adjusting the reactive output of the photovoltaic inverter based on the reactive voltage optimization model of the distribution network based on the DDPG algorithm. In view of the grid security problem brought about by high-penetration photovoltaic grid-connected, a reactive voltage optimization control strategy is proposed. This strategy controls the voltage over-limit problem of all nodes in the distribution network system by adjusting the reactive output of the grid-connected photovoltaic inverter, ensuring that the voltage at each node in the system is maintained within a safe range, thereby improving the safety of the distribution network operation after photovoltaic grid-connected. By constructing a system reactive voltage optimization model through a neural network, there is no need to rely on an accurate power flow model, and in the face of a complex environment in the distribution network, the information characteristics in the system can be deeply explored and learned, which reduces the complexity of distribution network modeling, solves the nonlinear and multi-constraint problems brought about by photovoltaic grid-connected, effectively improves the operating efficiency and reliability of the distribution network, and better realizes the reactive voltage optimization control of complex distribution network systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use 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 ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0039] Figure 1 This is an overall flow chart of a voltage balancing control method in a photovoltaic distribution network using a DDPG algorithm provided in a first embodiment of the present invention.

[0040] Figure 2 This is an overall framework diagram of a voltage balancing control system in a photovoltaic distribution network using a DDPG algorithm provided in a second embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0042] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides a voltage balancing control method in a photovoltaic distribution network using a DDPG algorithm, characterized in that:

[0043] S1: Collect operation data and establish a voltage control model for distributed photovoltaic distribution network.

[0044] The relevant operating data of the grid-connected inverters in the distributed photovoltaic distribution network are collected in real time. Based on the goal of minimizing the distribution network loss of the grid-connected inverters, combined with the power flow constraints in the distribution network and the inverter reactive output constraints, a distributed photovoltaic distribution network voltage control model with reactive output analysis is established.

[0045] In order to solve the voltage over-limit problem of distribution network caused by high photovoltaic penetration rate, an objective function of minimizing the network loss of distribution network is set with the premise of keeping the node voltage within a safe range (the range selected in this invention is 0.95-1.05pu). The objective function of the model can be expressed as:

[0046] Among them, P loss It represents the active power loss of the distribution network system after the PV inverter is activated at time t.

[0047] The power flow constraint is expressed as,

[0048]

[0049] Among them, P PV，i and Q PV，i They represent the active and reactive power generated by the photovoltaic power source on the grid-connected node i, P L,i and Q L，i They represent the active and reactive power of the load on node i, U i and U j denote the voltage amplitudes at nodes i and j, respectively, ij Represents the voltage phase angle difference between two adjacent nodes, G ij and B ij They represent the conductance and susceptance of the line respectively.

[0050] Reactive output constraint of photovoltaic inverter. Photovoltaic power source participates in reactive voltage regulation of distribution network through grid-connected inverter, thus avoiding voltage over-limit problem caused by photovoltaic fluctuation and ensuring safe operation of distribution network. When the voltage over-limit phenomenon occurs when photovoltaic power source is connected to the grid, the reactive compensation in the model is set to be provided only by photovoltaic inverter, and the reactive output constraint is expressed as,

[0051]

[0052] Where S represents the rated capacity of the grid-connected inverter, P PV Indicates the active output of the photovoltaic power source, Q PV Indicates the reactive power compensation actually provided by the grid-connected inverter, Q min and Q min They respectively represent the minimum and maximum reactive power compensation provided by the grid-connected inverter.

[0053] S2: Establish a reactive power and voltage optimization model for the distribution network.

[0054] The deep deterministic policy gradient (DDPG) algorithm combines the Actor-Critic framework with the deterministic policy gradient algorithm to solve the action control problem in continuous space. In the DDPG algorithm, there are two neural networks: one is the actor network, which is used to generate the actions of the agent and apply them to the environment to get the next state and reward; the other is the critic network, which is used to evaluate the quality of the actor network strategy and provide feedback.

[0055] The Actor network and Critic network of the intelligent agent of the present invention use the same network structure during training. Each network consists of an input layer, two hidden layers and an output layer. The number of inputs and outputs matches the state and action dimensions of the intelligent agent. Taking the Actor network as an example, the specific model is as follows Figure 1 As shown in the figure, the hidden layer adopts a fully connected structure. The number of neurons in the first hidden layer is 485, and the relu function is used as the activation function; the number of neurons in the second hidden layer is 300, and the tan h function is used as the activation function.

[0056] Each photovoltaic inverter in the distributed photovoltaic distribution network is integrated into the DDPG algorithm as an intelligent agent, and the distributed photovoltaic distribution network becomes the environment of the DDPG algorithm. The voltage of each node in the distribution network is taken as the controlled object of the DDPG algorithm. The intelligent agent will select the optimal action value based on the shared observation results of the current state and the action at the next moment, that is, the reactive output value of each grid-connected inverter is obtained. The reward value of the DDPG algorithm is calculated based on the reactive output value of each grid-connected inverter. The reward value is to minimize the network loss of the distribution network under the condition of satisfying voltage safety.

[0057] The reactive power voltage optimization model of the distribution network is based on the DDPG algorithm. The photovoltaic inverter in the distributed photovoltaic distribution network is integrated into the DDPG algorithm as an intelligent agent. The voltage of each node in the distributed photovoltaic distribution network is used as the state input of the DDPG algorithm. The voltage of each node is used as the controlled object of the model to obtain the state space. The state space is expressed as:

[0058] s=[U1,...U i , ...U N , P1, ...P i , ...P N , Q1, ...Q i ,...Q N ]

[0059] Among them, U i , P i , Q i They represent the voltage value, active power and reactive power of the i-th node respectively.

[0060] The DDPG algorithm includes an actor network and a critic network. The actor network is used to generate the actions of the agent. The agent will select the optimal action value based on the shared observation results of the current state and the action at the next moment. The action space in the model is set to the reactive output value of each grid-connected inverter as follows:

[0061]

[0062] Among them, a i represents the action of the ith photovoltaic inverter, Ai represents the action search space of the ith agent, S i represents the capacity of the i-th photovoltaic inverter, P i，PV represents the active output of the ith photovoltaic power source, a min and a max They respectively represent the minimum and maximum action values ​​actually provided by the grid-connected inverter agent.

[0063] The DDPG algorithm puts the voltage constraint and network loss into the reward function. If the node voltage is not within the specified safety range, a penalty term M is added to the reward function of the DDPG algorithm. The reward function is expressed as:

[0064] r t =-(M1P loss +M2ΔV t )

[0065]

[0066] Among them, P loss represents the active power loss of the distribution network system after the photovoltaic inverter is activated at time t, r t represents the reward value at time t, ΔV t It represents the voltage exceeding the limit of the node at time t, and M1 and M2 are penalty factors.

[0067] S3: Generate voltage balancing control strategy based on the reactive power voltage optimization model of the distribution network.

[0068] The distribution network voltage is controlled by dynamically adjusting the reactive output of the photovoltaic inverter through the distribution network reactive voltage optimization model based on the DDPG algorithm. After the photovoltaic power source is connected to the grid, the DDPG algorithm is used to control the action of the grid-connected inverter so that its output reactive power participates in voltage regulation and controls the node voltage within a safe range.

[0069] Example 2, reference Figure 2 , which is an embodiment of the present invention, provides a system for a voltage balancing control method in a photovoltaic distribution network using a DDPG algorithm, characterized in that it includes a data acquisition module 100, a DDPG algorithm module 200 and a voltage balancing control strategy generation module 300.

[0070] The data acquisition module 100 is responsible for collecting the operating data of the grid-connected inverters in the distributed photovoltaic power distribution network.

[0071] The DDPG algorithm module 200 selects the optimal action value based on the shared observation results of the current state and the action at the next moment.

[0072] The voltage balancing control strategy generation module 300 is responsible for dynamically adjusting the reactive output of the photovoltaic inverter, controlling the action of the grid-connected inverter through the DDPG algorithm, and keeping the node voltage within a safe range.

[0073] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0074] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0075] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0076] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0077] Example 3: In this example, in order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments. This example conducts experiments on the existing traditional method and the method of this example, as shown in Table 1.

[0078] Voltage stability is a key issue in distributed photovoltaic distribution networks. Traditional methods usually rely on fixed reactive compensation devices, such as capacitor banks or SVCs. Although these methods can provide certain voltage support, they have the disadvantages of slow response speed and insufficient control flexibility when dealing with photovoltaic output fluctuations. The present invention proposes a voltage balancing control method based on the DDPG algorithm, which can dynamically adjust the reactive output of the photovoltaic inverter to achieve faster and more flexible voltage control.

[0079] Table 1 Experimental effect comparison chart

[0080]

[0081]

[0082] Due to the use of the DDPG algorithm, the system can monitor the state of the power grid in real time and make control decisions quickly, which is due to the deep learning ability of the DDPG algorithm and the direct mapping of the environmental state. Compared with the traditional method, the response time of the method of the present invention is shorter, which means that the system can adapt to the changes in photovoltaic output more quickly and improve the real-time performance of the control.

[0083] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A voltage balancing control method in a photovoltaic distribution network based on a DDPG algorithm, characterized in that: include: Collect operation data and establish a voltage control model for distributed photovoltaic distribution networks; Establish reactive power and voltage optimization model of distribution network; The voltage balancing control strategy is generated based on the reactive power voltage optimization model of the distribution network.

2. The voltage balancing control method in a photovoltaic distribution network using the DDPG algorithm according to claim 1, characterized in that: The operating data is the operating data of the grid-connected inverter in the distributed photovoltaic distribution network; The distributed photovoltaic distribution network voltage control model combines power flow constraints in the distribution network and inverter reactive output constraints.

3. The voltage balancing control method in a photovoltaic distribution network using the DDPG algorithm as claimed in claim 2, characterized in that: The objective function of the distributed photovoltaic distribution network voltage control model is expressed as: Among them, P loss It represents the active power loss of the distribution network system after the photovoltaic inverter is activated at time t; The power flow constraint is expressed as, Among them, P PV,i and Q PV,i They represent the active and reactive power generated by the photovoltaic power source on the grid-connected node i, P L,i and Q L,i They represent the active and reactive power of the load on node i, U i and U j denote the voltage amplitudes at nodes i and j, respectively, ij Represents the voltage phase angle difference between two adjacent nodes, G ij and B ij They represent the conductance and susceptance of the line respectively; The reactive output constraint is expressed as, Where S represents the rated capacity of the grid-connected inverter, P PV Indicates the active output of the photovoltaic power source, Q PV Indicates the reactive power compensation actually provided by the grid-connected inverter, Q min and Q min They respectively represent the minimum and maximum reactive power compensation provided by the grid-connected inverter.

4. The voltage balancing control method in a photovoltaic distribution network using the DDPG algorithm as claimed in claim 3, characterized in that: The reactive voltage optimization model of the distribution network is based on the DDPG algorithm. The photovoltaic inverter in the distributed photovoltaic distribution network is integrated into the DDPG algorithm as an intelligent agent. The voltage of each node in the distributed photovoltaic distribution network is used as the state input of the DDPG algorithm. The voltage of each node is used as the controlled object of the model to obtain the state space. The state space is expressed as: s=[U1,…U i ,…U N ,P1,…P i ,…P N ,Q1,…Q i ,…Q N ] Among them, U i , P i , Q i They represent the voltage value, active power and reactive power of the i-th node respectively.

5. The voltage balancing control method in a photovoltaic distribution network using the DDPG algorithm as claimed in claim 4, characterized in that: The DDPG algorithm includes an actor network and a critic network, wherein the actor network is used to generate the actions of the agent. The agent will select the optimal action value based on the shared observation results of the current state and the action at the next moment. The action space in the model is set to the reactive output value of each grid-connected inverter as follows: Among them, a i represents the action of the ith photovoltaic inverter, A i represents the action search space of the ith agent, S i represents the capacity of the i-th photovoltaic inverter, P i,PV represents the active output of the ith photovoltaic power source, a min and a max They respectively represent the minimum and maximum action values ​​actually provided by the grid-connected inverter agent.

6. The voltage balancing control method in a photovoltaic distribution network using the DDPG algorithm as claimed in claim 5, characterized in that: The DDPG algorithm puts the voltage constraint and network loss into the reward function. If the node voltage is not within the specified safety range, a penalty term M is added to the reward function of the DDPG algorithm. The reward function is expressed as: r t =-(M1P loss +M2ΔV t ) represents the reward value at time t, ΔV t It represents the voltage exceeding the limit of the node at time t, and M1 and M2 are penalty factors.

7. The voltage balancing control method in a photovoltaic distribution network using the DDPG algorithm as claimed in claim 6, characterized in that: The generated voltage balancing control strategy includes dynamically adjusting the reactive output of the photovoltaic inverter through a distribution network reactive voltage optimization model based on the DDPG algorithm, and controlling the action of the grid-connected inverter through the DDPG algorithm after the photovoltaic power source is connected to the grid to control the node voltage within a safe range.

8. A system for voltage balancing control method in a photovoltaic distribution network using the DDPG algorithm as claimed in any one of claims 1 to 7, characterized in that: It comprises a data acquisition module (100), a DDPG algorithm module (200) and a voltage balancing control strategy generation module (300); The data acquisition module (100) is responsible for collecting the operation data of the grid-connected inverter in the distributed photovoltaic power distribution network; The DDPG algorithm module (200) selects the optimal action value according to the shared observation results of the current state and the action at the next moment; The voltage balancing control strategy generation module (300) is responsible for dynamically adjusting the reactive output of the photovoltaic inverter, controlling the action of the grid-connected inverter through the DDPG algorithm, and keeping the node voltage within a safe range.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the voltage balancing control method in the photovoltaic distribution network of the DDPG algorithm according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the voltage balancing control method in a photovoltaic distribution network of the DDPG algorithm described in any one of claims 1 to 7 are implemented.

Images