Microgrid group reactive voltage control method and system based on MADDPG algorithm

Abstract

The invention discloses a micro-grid group reactive voltage control method and system based on an MADDPG algorithm, and relates to the technical field of micro-grid group reactive voltage control, and the method comprises the following steps: determining a reactive voltage objective function and a constraint condition of a micro-grid group; introducing a prior knowledge base, and taking a reactive voltage objective function optimization result of the microgrid group as the prior knowledge base of an MADDPG algorithm; and designing a reactive voltage control model of the micro-grid group based on the prior knowledge base of the MADDPG algorithm, and applying the reactive voltage control model of the micro-grid group to reactive voltage control of an actual micro-grid group. According to the method, the convergence speed of the MADDPG algorithm is improved by introducing the priori knowledge base and microgrid group reactive power optimization control model conversion, the problem exploration time is shortened, and meanwhile reactive power output of all microgrids can be coordinated so as to maintain the voltage stability of the power grids.

Description

Technical Field

[0001] The present invention relates to the technical field of reactive voltage control of a microgrid group, and in particular to a reactive voltage control method and system of a microgrid group based on a MADDPG algorithm. Background Art

[0002] Microgrid groups contain a large number of distributed power sources with obvious randomness and intermittent characteristics, which brings challenges to the stability of microgrid groups. In the reactive voltage control of microgrid groups, it is necessary to ensure the stability and voltage quality of the system to avoid voltage collapse or power quality degradation caused by improper reactive control.

[0003] The traditional method of reactive power and voltage control in microgrid groups mainly relies on the droop control strategy, which distributes active and reactive power by adjusting the output frequency and voltage amplitude of the inverter. However, due to factors such as the line impedance mismatch of each distributed power source in the microgrid, the traditional droop control is difficult to achieve accurate distribution of reactive power, which may cause circulating reactive power between inverters, affecting power quality and system stability. Summary of the invention

[0004] In view of the existing problems in the reactive voltage control and system of microgrid groups based on the MADDPG algorithm, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that it is difficult to achieve accurate distribution of reactive power by conventional droop control, which may cause circulating reactive power between inverters, affecting power quality and system stability.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a microgrid group reactive voltage control method based on the MADDPG algorithm, which comprises the following steps:

[0008] Determine the reactive voltage objective function and constraints of the microgrid group;

[0009] The prior knowledge base is introduced, and the optimization results of the reactive power and voltage objective function of the microgrid group are used as the prior knowledge base of the MADDPG algorithm;

[0010] A reactive voltage control model of a microgrid group is designed based on the prior knowledge base of the MADDPG algorithm, and the reactive voltage control model of a microgrid group is applied to the reactive voltage control of an actual microgrid group.

[0011] As a preferred solution of the reactive voltage control method of a microgrid group based on the MADDPG algorithm of the present invention, wherein: the reactive voltage objective function of the microgrid group takes stability and economy as the goals;

[0012] Under the condition of minimizing voltage deviation, the reactive voltage objective function formula for stability is defined as:

[0013]

[0014] Where V j is the actual voltage of microgrid node j, V ref,j is the reference voltage of microgrid cluster node j;

[0015] The reactive voltage objective function formula for stability is defined by minimizing frequency fluctuations:

[0016]

[0017] In the formula, f j is the actual voltage of microgrid node j, f ref,j is the reference voltage of microgrid cluster node j;

[0018] Minimize the total harmonic distortion rate and define the reactive voltage objective function formula for stability:

[0019]

[0020] Where V n is the effective value of the nth harmonic, V1 is the fundamental voltage;

[0021] The objective function of the operation stability of the integrated microgrid group is:

[0022] min(αV+βf+γD);

[0023] Where V is the minimized voltage deviation of the microgrid group, f is the minimized frequency fluctuation of the microgrid group, D is the minimized total harmonic distortion rate of the microgrid group, and α, β, and γ are the weight coefficients of the microgrid group.

[0024] As a preferred solution of the reactive voltage control method of a microgrid group based on the MADDPG algorithm described in the present invention, the economic objective function of the microgrid group is:

[0025]

[0026] Where N is the number of power generation units in the microgrid group, c gen,i is the unit power generation cost of the i-th power generation unit, P gen,i is the power generation, M is the number of energy storage units in the microgrid group, c stor,j is the unit charging and discharging cost of the i-th energy storage unit, P ess,j is the charge and discharge power.

[0027] As a preferred solution of the reactive voltage control method of a microgrid group based on the MADDPG algorithm described in the present invention, the constraint condition is the active power and reactive power flow constraints of each node in the microgrid group, wherein the flow constraint equation is:

[0028]

[0029] In the formula, j∈i means that node j belongs to the set of nodes connected to node i, P i and Q i are the active power and reactive power injected at node i, V i and V j are the voltage amplitudes injected at node i and node j respectively, θ ij is the phase angle difference between node i and node j, G ij and B ij are the real and imaginary parts of the elements in the i-th row and j-th column of the node admittance matrix respectively;

[0030] The voltage amplitude constraint equation is:

[0031]

[0032] In the formula, are the minimum and maximum allowable voltages of node j respectively;

[0033] The reactive power balance constraint equation is:

[0034]

[0035] In the formula, and are the sum of reactive powers provided by the microgrid group’s power generation unit and energy storage unit, and Q loss They are the sum of reactive power demanded by the microgrid group loads and reactive power loss respectively.

[0036] As a preferred solution of the reactive voltage control method of a microgrid group based on the MADDPG algorithm of the present invention, wherein: the prior knowledge base includes human declarative knowledge and intelligent program-type knowledge;

[0037] The intelligent program-like knowledge is used to accelerate the speed of learning;

[0038] The human declarative knowledge is used to influence the actions of the intelligent agent.

[0039] As a preferred solution of the reactive voltage control method of a microgrid group based on the MADDPG algorithm of the present invention, the method for constructing the prior knowledge base of the MADDPG algorithm includes:

[0040] Get the state space of MADDPG at the current moment to generate the random function r uni ;

[0041] Get the decay function ε(l) of the current iteration number;

[0042] The random function r uni Compare with the decay function ε(l) to determine whether the improved MADDPG algorithm needs to use the prior knowledge base based on the optimal power flow to train MADDPG;

[0043] When the random function r uni When it is greater than the decay function ε(l), the reward value data obtained by the interaction between the MADDPG agent and the environment model is used to train the MADDPG algorithm;

[0044] When the random function r uni When it is less than or equal to the attenuation function ε(l), the MADDPG algorithm is trained using the prior knowledge in the prior knowledge base based on the optimal power flow;

[0045] Among them, the random function r uni The value range is: r uni =random[0,1];

[0046] The attenuation function is a quadratic function, and the specific expression is:

[0047] ε(l)=al 2 +bl+cl≥0andε≥0;

[0048] Where a, b, and c are the attenuation function parameters.

[0049] As a preferred solution of the reactive voltage control method of a microgrid group based on the MADDPG algorithm of the present invention, the state space of the reactive voltage control model of the microgrid group is divided into the corresponding states of the intelligent bodies such as the microgrid group and the sub-microgrid, which is specifically expressed as follows:

[0050]

[0051]

[0052]

[0053]

[0054] In the formula, S t Represented as the current state space, They represent the state subspaces of the current energy storage unit, generator set and reactive power regulation and compensation equipment respectively, and k represents the sensing range, that is, the number of nodes connected to a certain node;

[0055] Action execution performs reactive output actions of the agent according to the output characteristics of different agents. The formula is expressed as:

[0056]

[0057]

[0058]

[0059]

[0060] In the formula, and Respectively represent the current action execution, A t Represented as the current action set of each agent.

[0061] In a second aspect, an embodiment of the present invention provides a microgrid group reactive voltage control system based on the MADDPG algorithm, which includes a judgment module, an analysis module, and an output module;

[0062] The judgment module is used to judge the reactive voltage objective function and constraint conditions of the microgrid group;

[0063] The analysis module is used to use the optimization result of the reactive voltage objective function of the microgrid group as the prior knowledge base of the MADDPG algorithm;

[0064] The output module is used to construct a reactive voltage control model for a microgrid group.

[0065] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, it implements any step of the above-mentioned microgrid group reactive voltage control method based on the MADDPG algorithm.

[0066] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned microgrid group reactive voltage control method based on the MADDPG algorithm is implemented.

[0067] The beneficial effects of the present invention are as follows: by introducing a priori knowledge base and microgrid group reactive power optimization control model conversion, the convergence speed of the MADDPG algorithm is improved, the problem exploration time is reduced, and at the same time, the reactive power output of each microgrid can be coordinated to maintain the voltage stability of the power grid. Through the MADDPG algorithm, each microgrid can learn how to optimize the reactive output of various distributed power sources while maintaining the voltage stability of the power grid, which not only improves the operating efficiency of the power grid, but also enhances the reliability and flexibility of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0069] Figure 1 This is a flow chart of the reactive power and voltage control method of a microgrid group based on the MADDPG algorithm.

[0070] Figure 2 This is the changing trend of the attenuation function of the reactive power and voltage control method of the microgrid group based on the MADDPG algorithm with the number of iterations.

[0071] Figure 3 This is the conversion diagram of the reactive power and voltage control model of the microgrid group based on the MADDPG algorithm. DETAILED DESCRIPTION

[0072] 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.

[0073] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0074] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0075] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0076] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0077] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0078] Example 1

[0079] Reference Figure 1 to Figure 3 , which is the first embodiment of the present invention, and provides a microgrid group reactive voltage control method based on MADDPG algorithm, comprising the following steps:

[0080] S1. Determine the reactive voltage objective function and constraints of the microgrid group.

[0081] Due to the intermittent and random nature of the output of a large number of distributed energy sources in microgrid groups, it is necessary to ensure the stability of the system and maximize the economic benefits while ensuring stable operation of the system. Therefore, stability and economy are the common goals of the reactive voltage function of the microgrid group.

[0082] Poor power quality in the microgrid group will seriously affect the stable operation of the system. At this time, if all microgrids are connected to the grid at the same time, it will definitely bring shocks to the large grid. In order to ensure the coordination between the microgrid group and the main grid, the grid stability target is considered from the aspects of voltage quality, frequency stability and harmonic content.

[0083] The reactive voltage objective function of the microgrid group takes stability and economy as the goals;

[0084] Under the condition of minimizing voltage deviation, the reactive voltage objective function formula for stability is defined as:

[0085]

[0086] Where V j is the actual voltage of microgrid node j, V ref,j is the reference voltage of microgrid cluster node j;

[0087] The reactive voltage objective function formula for stability is defined by minimizing frequency fluctuations:

[0088]

[0089] In the formula, f j is the actual voltage of microgrid node j, f ref,j is the reference voltage of microgrid cluster node j;

[0090] Minimize the total harmonic distortion rate and define the reactive voltage objective function formula for stability:

[0091]

[0092] Where V n is the effective value of the nth harmonic, V1 is the fundamental voltage;

[0093] The objective function of the operation stability of the integrated microgrid group is:

[0094] min(αV+βf+γD);

[0095] Where V is the minimized voltage deviation of the microgrid group, f is the minimized frequency fluctuation of the microgrid group, D is the minimized total harmonic distortion rate of the microgrid group, and α, β, and γ are the weight coefficients of the microgrid group.

[0096] The economic objective function of the microgrid group is mainly used for the optimal operation of the microgrid group. It is usually considered from the perspective of power generation cost and energy storage system balancing cost to maximize benefits. The economic objective function of the microgrid group is:

[0097]

[0098] Where N is the number of power generation units in the microgrid group, c gen,i is the unit power generation cost of the i-th power generation unit, P gen,i is the power generation, M is the number of energy storage units in the microgrid group, c stor,j is the unit charging and discharging cost of the i-th energy storage unit, P ess,j is the charge and discharge power.

[0099] The constraints are the active power and reactive power flow constraints of each node in the microgrid group, where the flow constraint equation is:

[0100]

[0101] In the formula, j∈i means that node j belongs to the set of nodes connected to node i, P i and Q i are the active power and reactive power injected at node i, V i and V j are the voltage amplitudes injected at node i and node j respectively, θ ij is the phase angle difference between node i and node j, G ij and B ij are the real and imaginary parts of the elements in the i-th row and j-th column of the node admittance matrix respectively;

[0102] The voltage amplitude constraint equation is:

[0103]

[0104] In the formula, are the minimum and maximum allowable voltages of node j respectively;

[0105] The reactive power balance constraint equation is:

[0106]

[0107] In the formula, and are the sum of reactive powers provided by the microgrid group’s power generation unit and energy storage unit, and Q loss They are the sum of reactive power demanded by the microgrid group loads and reactive power loss respectively.

[0108] S2. Introduce a priori knowledge base and use the optimization results of the reactive voltage objective function of the microgrid group as the priori knowledge base of the MADDPG algorithm.

[0109] The prior knowledge base includes human declarative knowledge and intelligent program-like knowledge;

[0110] The intelligent program-like knowledge is used to accelerate the speed of learning;

[0111] The human declarative knowledge is used to influence the actions of the intelligent agent.

[0112] Human declarative knowledge is usually combined with intelligent algorithms in online and offline ways. Online combination of human declarative knowledge occurs in the process of interaction between the intelligent agent and the environmental model. One is that humans subjectively evaluate the actions performed by the intelligent agent, and the feedback information of the evaluation is used to optimize the policy parameters of the intelligent agent's action-evaluation, that is, the optimal action distribution under the state of judgment from the human perspective and applied to the policy gradient; the other is that humans make declarative suggestions based on historical experience for the actions that the intelligent agent should take, and the intelligent agent executes or optimizes the action according to this suggestion, that is, effective declarative suggestions will focus on constraining the intelligent agent to make relatively wrong actions and improve the speed of the intelligent agent to learn relatively correct strategies. Offline combination of human declarative knowledge occurs before the algorithm optimization, with the effective strategy set accumulated in other optimization tasks or the suggestions made before the intelligent agent interacts as the combined knowledge. The former requires excellent data in the effective strategy set as the basis, including specific rules such as action conversion performed by humans, which can realize the combination of the intelligent agent between self-learning and rules. The latter can use linear temporal logic to standardize action suggestions to ensure that the intelligent agent understands human declarative knowledge before interacting with the environmental model and realizes the influence on the intelligent agent's actions.

[0113] Intelligent program knowledge is the prior knowledge formed by adding heuristic algorithms to the intelligent agent algorithm to influence the action selection of the intelligent agent, thereby accelerating the convergence speed of the algorithm solution model. The key is to add heuristic functions in the early stage of algorithm agent learning to obtain intelligent heuristic knowledge, which transfers prior knowledge to the current task to achieve the effect of accelerated learning.

[0114] The methods for constructing the prior knowledge base of the MADDPG algorithm include:

[0115] Get the state space of MADDPG at the current moment to generate the random function r uni ;

[0116] Get the decay function ε(l) of the current iteration number;

[0117] The random function r uni Compare with the decay function ε(l) to determine whether the improved MADDPG algorithm needs to use the prior knowledge base based on the optimal power flow to train MADDPG;

[0118] When the random function r uni When it is greater than the decay function ε(l), the reward value data obtained by the interaction between the MADDPG agent and the environment model is used to train the MADDPG algorithm;

[0119] When the random function r uniWhen it is less than or equal to the attenuation function ε(l), the MADDPG algorithm is trained using the prior knowledge in the prior knowledge base based on the optimal power flow;

[0120] When the random function r uni Greater than the minimum attenuation function ε min At this time, it has gone through multiple rounds of iterations. Under the guidance of the prior knowledge base, the MADDPG algorithm has already had more feedback experience and can perform relatively good optimization exploration for optimal scheduling. Therefore, the prior knowledge base is closed and MADDPG continues to interact with the environmental model for training. If the prior knowledge base based on optimal power flow is continued to be used, the MADDPG algorithm will fall into the dilemma of local optimality, thereby limiting the MADDPG algorithm's advantage in exploring learning and searching for the global optimality.

[0121] Among them, the random function r uni The value range is: r uni =random[0,1];

[0122] Based on the above analysis of the application method of the prior knowledge base in the improved MADDPG algorithm, it can be seen that the design of the decay function ε(l) is the key to realizing the integration of the prior knowledge base and the MADDPG algorithm. In the early stage of optimization iteration, due to the instability of the global optimization of the MADDPG algorithm, the prior knowledge in the prior knowledge base should be used as much as possible to train the MADDPG algorithm to speed up the convergence speed. Therefore, the decay function should take a larger value. With the increase of optimization iterations, there are more states, actions, and reward sets in the feedback set. At this time, the guiding role of the prior knowledge base on the MADDPG algorithm is fully demonstrated. In order not to limit the advantage of the global optimization of the MADDPG algorithm, the decay function should take a smaller value. After 1500 iterations, the decay function is 0, and the prior knowledge base is no longer used.

[0123] The attenuation function is a quadratic function, and the specific expression is:

[0124] ε(l)=al 2 +bl+cl≥0andε≥0;

[0125] In the formula, a, b, and c are the attenuation function parameters respectively;

[0126] Among them, the variation trend of the attenuation function with the number of iterations is as follows: Figure 2 shown.

[0127] S3. Design a reactive voltage control model for a microgrid group based on the prior knowledge base of the MADDPG algorithm, and apply the reactive voltage control model for a microgrid group to the reactive voltage control of an actual microgrid group.

[0128] The MADDPG algorithm introduces two neural network modules, Actor and Critic, in each agent. The core of the algorithm is that the Critic module can observe and integrate the information of the entire system. By modifying the Critic objective function, the Actor module is guided to perform corresponding actions after receiving the state information of the environmental model. N agents act independently to improve the stability and robustness of the algorithm. At the same time, the system stores the environmental state, multi-agent actions, rewards received, and the resulting environmental state, which is convenient for each agent to sample and learn and update strategies.

[0129] First, each microgrid in the microgrid group is regarded as each intelligent agent. Each microgrid intelligent agent is in the microgrid group system together. By sensing environmental changes such as node voltage, reactive power changes and operating costs at a certain moment, it makes corresponding decision actions. Each intelligent agent has both independent decision actions and centralized interactions, which has the advantage of distributed expansion. When one of the regional intelligent agents changes, other intelligent agents can observe the relevant environmental changes and make adaptive actions through learning, which does not affect the overall task objectives and reflects good robustness. The coordinated optimization of each regional intelligent agent achieves the overall optimization goal of the reactive power voltage of the microgrid group.

[0130] According to the MADDPG "centralized learning-distributed action" method, corresponding to each intelligent agent, firstly, centralized learning is carried out in the microgrid group environment. According to the initial conditions and historical data of the microgrid group, the reactive power demand and node voltage of the current microgrid group are observed. The Actor module network of each intelligent agent takes action, that is, it decides how much power to output or absorb to the microgrid group system and then enters the observation state at the next moment. The Critic module network in the intelligent agent evaluates the value according to the action set taken by all intelligent agents, calculates the corresponding reward signal, and guides the Actor module network to take action in a continuous "trial and error" cycle until the end of training. Finally, the trained microgrid group reactive voltage control model is applied to the reactive voltage control of the actual microgrid group;

[0131] Among them, the reactive power and voltage control model of the microgrid group based on the MADDPG algorithm is transformed as follows: Figure 3 As shown in the figure, according to the algorithm elements of MADDPG, the reactive voltage control model of the microgrid group is designed from four aspects: state space, action execution, environment model and reward signal.

[0132] The state space of the reactive voltage control model of the microgrid group is divided into the corresponding states of the microgrid group, the sub-microgrid and other intelligent entities, which are specifically expressed as follows:

[0133]

[0134]

[0135]

[0136]

[0137] In the formula, S t Represented as the current state space, They represent the state subspaces of the current energy storage unit, generator set and reactive power regulation and compensation equipment respectively, and k represents the sensing range, that is, the number of nodes connected to a certain node;

[0138] Action execution performs reactive output actions of the agent according to the output characteristics of different agents. The formula is expressed as:

[0139]

[0140]

[0141]

[0142]

[0143] In the formula, and Respectively represent the current action execution, A t Represented as the current action set of each agent.

[0144] In the agent interaction process based on the MADDPG algorithm, the reactive voltage control model of the microgrid group maintains stability. After obtaining the action of the agent Actor module, the state is updated and the corresponding reward signal is given to the agent. After each agent performs the output action, it updates the elements of the state space through dynamic simulation while satisfying the constraints, and then calculates the reward signal and feeds it back to the agent, and this cycle is repeated until the end.

[0145] Designing reward signals is an important part of the MADDPG algorithm. The reward signal affects the value estimation of the Critic module of the microgrid group agent, which in turn determines whether the guided actions can be closer to the optimization goal and achieve the convergence of the entire model, that is:

[0146]

[0147] In the formula, R t n and r p Represented as the current reward signal and penalty value respectively.

[0148] In summary, by introducing the prior knowledge base and the conversion of the reactive power optimization control model of the microgrid group, the convergence speed of the MADDPG algorithm can be improved and the problem exploration time can be reduced. At the same time, the reactive power output of each microgrid can be coordinated to maintain the voltage stability of the power grid. Through the MADDPG algorithm, each microgrid can learn how to optimize the reactive output of various distributed power sources while maintaining the voltage stability of the power grid, which not only improves the operating efficiency of the power grid, but also enhances the reliability and flexibility of the system.

[0149] Example 2

[0150] On the basis of the first embodiment, this embodiment further provides a microgrid group reactive voltage control system based on the MADDPG algorithm, including a judgment module, an analysis module, and an output module;

[0151] The judgment module is used to judge the reactive voltage objective function and constraint conditions of the microgrid group;

[0152] The analysis module is used to use the optimization result of the reactive voltage objective function of the microgrid group as the prior knowledge base of the MADDPG algorithm;

[0153] The output module is used to construct a reactive voltage control model for a microgrid group.

[0154] This embodiment also provides a computer device, which is suitable for the case of a reactive voltage control method for a microgrid group based on the MADDPG algorithm, and includes a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the reactive voltage control method for a microgrid group based on the MADDPG algorithm as proposed in the above embodiment.

[0155] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0156] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for controlling reactive voltage of a microgrid group based on the MADDPG algorithm as proposed in the above embodiment is implemented.

[0157] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0158] 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 reactive voltage control method for a microgrid group based on the MADDPG algorithm, characterized in that: The following steps are included: Determine the reactive voltage objective function and constraints of the microgrid group; The prior knowledge base is introduced, and the optimization results of the reactive power and voltage objective function of the microgrid group are used as the prior knowledge base of the MADDPG algorithm; A reactive voltage control model of a microgrid group is designed based on the prior knowledge base of the MADDPG algorithm, and the reactive voltage control model of a microgrid group is applied to the reactive voltage control of an actual microgrid group.

2. The microgrid group reactive voltage control method based on the MADDPG algorithm as claimed in claim 1, characterized in that: The reactive voltage objective function of the microgrid group takes stability and economy as the goals; Under the condition of minimizing voltage deviation, the reactive voltage objective function formula for stability is defined as: Where V j is the actual voltage of microgrid node j, V ref,j is the reference voltage of microgrid cluster node j; The reactive voltage objective function formula for stability is defined by minimizing frequency fluctuations: In the formula, f j is the actual voltage of microgrid node j, f ref,j is the reference voltage of microgrid cluster node j; Minimize the total harmonic distortion rate and define the reactive voltage objective function formula for stability: Where V n is the effective value of the nth harmonic, V1 is the fundamental voltage; The objective function of the operation stability of the integrated microgrid group is: min(αV+βf+γD); Where V is the minimized voltage deviation of the microgrid group, f is the minimized frequency fluctuation of the microgrid group, D is the minimized total harmonic distortion rate of the microgrid group, and α, β, and γ are the weight coefficients of the microgrid group.

3. The microgrid group reactive voltage control method based on the MADDPG algorithm as claimed in claim 2, characterized in that: The economic objective function of the microgrid group is: Where N is the number of power generation units in the microgrid group, c gen,i is the unit power generation cost of the i-th power generation unit, P gen,i is the power generation, M is the number of energy storage units in the microgrid group, c stor,j is the unit charging and discharging cost of the i-th energy storage unit, P ess,j is the charge and discharge power.

4. The microgrid group reactive voltage control method based on the MADDPG algorithm as claimed in claim 3, characterized in that: The constraints are the active power and reactive power flow constraints of each node in the microgrid group, where the flow constraint equation is: In the formula, j∈i means that node j belongs to the set of nodes connected to node i, P i and Q i are the active power and reactive power injected at node i, V i and V j are the voltage amplitudes injected at node i and node j respectively, θ ij is the phase angle difference between node i and node j, G ij and B ij are the real and imaginary parts of the elements in the i-th row and j-th column of the node admittance matrix respectively; The voltage amplitude constraint equation is: In the formula, are the minimum and maximum allowable voltages of node j respectively; The reactive power balance constraint equation is: In the formula, and are the sum of reactive powers provided by the microgrid group’s power generation unit and energy storage unit, and Q loss They are the sum of reactive power demanded by the microgrid group loads and reactive power loss respectively.

5. The microgrid group reactive voltage control method based on the MADDPG algorithm as claimed in claim 4, characterized in that: The prior knowledge base includes human declarative knowledge and intelligent program-like knowledge; The intelligent program-like knowledge is used to accelerate the speed of learning; The human declarative knowledge is used to influence the actions of the intelligent agent.

6. The microgrid group reactive voltage control method based on MADDPG algorithm as claimed in claim 5, characterized in that: The method for constructing the prior knowledge base of the MADDPG algorithm includes: Get the state space of MADDPG at the current moment to generate the random function r uni ; Get the decay function ε(l) of the current iteration number; The random function r uni Compare with the decay function ε(l) to determine whether the improved MADDPG algorithm needs to use the prior knowledge base based on the optimal power flow to train MADDPG; When the random function r uni When it is greater than the decay function ε(l), the reward value data obtained by the interaction between the MADDPG agent and the environment model is used to train the MADDPG algorithm; When the random function r uni When it is less than or equal to the attenuation function ε(l), the MADDPG algorithm is trained using the prior knowledge in the prior knowledge base based on the optimal power flow; Among them, the random function r uni The value range is: r uni =random[0,1]; The attenuation function is a quadratic function, and the specific expression is: ε(l)=l 2 +bl+cl≥0andε≥0! Where a, b, and c are the attenuation function parameters.

7. The microgrid group reactive power voltage control method based on MADDPG algorithm as claimed in claim 6, characterized in that: The state space of the reactive voltage control model of the microgrid group is divided into the corresponding states of the microgrid group, the sub-microgrid and other intelligent entities, which are specifically expressed as follows: In the formula, S t Represented as the current state space, They represent the state subspaces of the current energy storage unit, generator set and reactive power regulation and compensation equipment respectively, and k represents the sensing range, that is, the number of nodes connected to a certain node; Action execution performs reactive output actions of the agent according to the output characteristics of different agents. The formula is expressed as: In the formula, and Respectively represent the current action execution, A t Represented as the current action set of each agent.

8. A microgrid group reactive voltage control system based on the MADDPG algorithm, based on the microgrid group reactive voltage control method based on the MADDPG algorithm according to any one of claims 1 to 7, characterized in that: It includes a judgment module, an analysis module, and an output module; The judgment module is used to judge the reactive voltage objective function and constraint conditions of the microgrid group; The analysis module is used to use the optimization result of the reactive voltage objective function of the microgrid group as the prior knowledge base of the MADDPG algorithm; The output module is used to construct a reactive voltage control model for a microgrid group.

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 microgrid reactive voltage control method based on the MADDPG algorithm described in 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 microgrid reactive voltage control method based on the MADDPG algorithm described in any one of claims 1 to 7 are implemented.

Images