Power distribution network voltage control scheduling method and device based on static var generator, electronic equipment and storage medium

Abstract

The invention discloses a power distribution network voltage control scheduling method and device based on a static var generator, electronic equipment and a storage medium, and belongs to the field of intelligent power grids, and the method comprises the steps: obtaining a topological structure of a power distribution network to be scheduled, hardware parameters of the static var generator, and a photovoltaic inverter group agent; according to the topological structure and the hardware parameters, with the purpose of minimizing the weighted sum of the comprehensive operation cost of the static var generator and the voltage deviation of the power distribution network, constructing a collaborative decision-making voltage model and constraint conditions of the collaborative decision-making voltage model; under the constraint condition, performing iteration solution of a preset round on the collaborative decision voltage model to generate a final scheduling strategy of the static var generator; and scheduling the static var generator in the to-be-scheduled power distribution network according to the final scheduling strategy. By implementing the method, the problem of inaccurate voltage control scheduling of the power distribution network due to the fact that the dynamic response behavior of the photovoltaic inverter group is not considered in the prior art can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart grids, and in particular to a method, device, electronic equipment and storage medium for controlling and dispatching voltage in a distribution network based on a static VAR generator. Background Art

[0002] With the increasing penetration of distributed renewable energy, especially photovoltaic power generation, in distribution networks, grid operation faces challenges such as intensified voltage fluctuations and increased risk of over-limit, which puts higher demands on traditional voltage control methods. Static VAR generator (SVG), as an advanced fast dynamic reactive power compensation device, can flexibly and accurately adjust reactive power, which plays a key role in maintaining voltage stability and improving power quality in distribution networks. Therefore, how to effectively control and dispatch the SVG in the distribution network to adapt to the new trend of increasing proportion of new energy and diversified development of loads, and to ensure the safe, high-quality and economical operation of the power grid, has become an important issue that needs to be solved in the current distribution network field.

[0003] When designing voltage dispatch strategies based on SVG, existing technologies often ignore the potential role of distributed photovoltaic inverter groups in voltage regulation and fail to include them as coordinated control resources in the dispatch model for unified consideration. This isolated dispatching method requires SVG to independently handle all voltage regulation tasks, which not only increases its operating burden, but also weakens the overall response capability of the system, thereby affecting the accuracy and adaptability of the dispatching strategy. Especially when the photovoltaic output fluctuates violently or the proportion of distributed access is high, this dispatching method that does not consider the dynamic response behavior of the photovoltaic inverter group will affect the reliability of the voltage control dispatch of the distribution network. Summary of the invention

[0004] The embodiments of the present invention provide a distribution network voltage control scheduling method, device, electronic device and storage medium based on a static VAR generator, which can solve the problem of inaccurate distribution network voltage control scheduling caused by not considering the dynamic response behavior of a photovoltaic inverter group in the prior art.

[0005] An embodiment of the present invention provides a distribution network voltage control scheduling method based on a static VAR generator, comprising:

[0006] Acquire the topological structure of the distribution network to be dispatched, the hardware parameters of the static VAR generator in the distribution network to be dispatched, and the photovoltaic inverter group agent; wherein the photovoltaic inverter group agent is obtained through interactive training through the actor-discriminator multi-agent algorithm based on a preset power flow calculation model and the electrical parameters of the photovoltaic inverter group in the distribution network to be dispatched;

[0007] Based on the topological structure and the hardware parameters, with the goal of minimizing the weighted sum of the comprehensive operating cost of the static var generator and the voltage deviation of the distribution network, and taking the scheduling strategy of the static var generator as the decision variable, a collaborative decision-making voltage model and the constraint conditions of the collaborative decision-making voltage model are constructed; the constraint conditions include reactive power over-limit constraint, node voltage constraint, and branch current constraint;

[0008] Under the constraint conditions, perform iterative solution of the collaborative decision-making voltage model for a preset number of rounds to generate the final scheduling strategy of the static var generator; among them, in the process of solving the collaborative decision-making voltage model in each round, based on the power flow calculation model, determine the autonomous power response generated by the photovoltaic inverter cluster agent after the scheduling strategy of the candidate static var generator is injected into the distribution network; according to the autonomous power response and the candidate static var generator scheduling strategy, based on the power flow calculation model, calculate and generate the objective function value of the collaborative decision-making voltage model in the current round; take the candidate static var generator scheduling strategy with the minimum objective function value in the preset number of rounds as the final scheduling strategy of the static var generator;

[0009] According to the final scheduling strategy, schedule the static var generators in the distribution network to be scheduled.

[0010] Furthermore, the photovoltaic inverter cluster agent is obtained through the following method:

[0011] According to the power flow calculation model, construct a dynamic operation simulation environment for the distribution network; among them, the power flow calculation model is generated by the topological structure and line impedance parameters of the distribution network to be scheduled;

[0012] In the dynamic operation simulation environment of the distribution network, define the state space, action space, reward function, and training objective of the agent based on the Markov decision process framework, and construct an actor network, a discriminator network, and an experience replay buffer;

[0013] Based on the actor-discriminator multi-agent algorithm, regard the photovoltaic inverters in the photovoltaic inverter cluster of the distribution network to be scheduled as agents, and through the interaction between the agents and the dynamic operation simulation environment of the distribution network, train the agents until the preset number of training times is reached to generate the trained actor network and the trained discriminator network; among them, in the training process, the agents generate state transition experience samples through interaction with the simulation environment and store them in the experience replay buffer; optimize the network parameters of the actor network and the discriminator network by sampling the experiences in the experience replay buffer;

[0014] Generate a photovoltaic inverter cluster agent according to the trained actor network and the trained discriminator network.

[0015] Further, the collaborative decision-making voltage model is specifically as follows:

[0016]

[0017] Among them, F1 is the objective function value of the collaborative decision-making voltage model; ω1 is the weight coefficient of the operating cost item of the static var generator; T is the total number of time periods in the optimization cycle; N SVG is the total number of static var generators participating in the optimization in the power grid; α k is the unit reactive power regulation cost coefficient of the kth static var generator; Q SVG，k (t) is the reactive power output of the kth static var generator at time period t; ω2 is the weight coefficient of the voltage deviation item; N is the total number of nodes included in the voltage deviation calculation in the distribution network; V i，t is the actual voltage value of node i at time period t; is the rated voltage value of node i at time period t.

[0018] Further, the reactive power over-limit constraint is specifically as follows:

[0019]

[0020] In the formula, is the minimum reactive power output limit of the kth static var generator; is the maximum reactive power output limit of the kth static var generator;

[0021] The node voltage constraint is specifically as follows:

[0022]

[0023] In the formula, υ i，t is the square of the voltage amplitude of the ith node at time period t; V i，t，min is the minimum allowable voltage amplitude of the ith node at time period t; V i，t，max is the maximum allowable voltage amplitude of the ith node at time period t;

[0024] The branch current constraint is specifically as follows:

[0025]

[0026] In the formula, l ij，t is the square of the current amplitude of the branch connecting node i and node j at time period t; I ij，t，min is the minimum allowable current amplitude of branch ij at time period t; I ij，t，max is the maximum allowable current amplitude of branch ij at time period t.

[0027] Further, the state space of the agent includes the current voltage measurement value at the grid connection point of the PV inverter, the current active power output value of the PV inverter, and the current reactive power output value of the PV inverter.

[0028] Further, the action space of the agent includes the current reactive power output target value and the active power output target value of the PV inverter.

[0029] Based on the above method item embodiments, the present invention correspondingly provides device item embodiments.

[0030] An embodiment of the present invention provides a distribution network voltage control and scheduling device based on a static var generator, including: a basic data acquisition module, a collaborative decision-making voltage model construction module, a collaborative decision-making voltage model solution module, and a static var generator scheduling module;

[0031] The basic data acquisition module is used to acquire the topological structure of the distribution network to be scheduled, the hardware parameters of the static var generators in the distribution network to be scheduled, and the PV inverter group agents; wherein, the PV inverter group agents are obtained through interactive training by an actor-critic multi-agent algorithm based on a preset power flow calculation model according to the electrical parameters of the PV inverter group in the distribution network to be scheduled;

[0032] The collaborative decision-making voltage model construction module is used to construct a collaborative decision-making voltage model and the constraint conditions of the collaborative decision-making voltage model with the goal of minimizing the weighted sum of the comprehensive operating cost of the static var generator and the voltage deviation of the distribution network, and using the static var generator scheduling strategy as the decision variable; the constraint conditions include reactive power over-limit constraints, node voltage constraints, and branch current constraints;

[0033] The collaborative decision-making voltage model solution module is used to iteratively solve the collaborative decision-making voltage model for a preset number of rounds under the constraint conditions to generate the final scheduling strategy of the static var generator; wherein, in the process of solving the collaborative decision-making voltage model in each round, based on the power flow calculation model, determine the autonomous power response generated by the PV inverter group agents after the scheduling strategy of the candidate static var generator is injected into the distribution network; according to the autonomous power response and the candidate static var generator scheduling strategy, calculate and generate the objective function value of the collaborative decision-making voltage model in the current round based on the power flow calculation model; take the candidate static var generator scheduling strategy with the minimum objective function value in the preset number of rounds as the final scheduling strategy of the static var generator;

[0034] The static var generator scheduling module is used to schedule the static var generators in the distribution network to be scheduled according to the final scheduling strategy.

[0035] Further, for the distribution network voltage control and scheduling device based on a static var generator, a collaborative decision-making voltage model construction module, the collaborative decision-making voltage model is specifically as follows:

[0036]

[0037] Among them, F1 is the objective function value of the collaborative decision-making voltage model; ω1 is the weight coefficient of the operating cost item of the static var generator; T is the total number of time periods in the optimization cycle; N SVG is the total number of static var generators participating in the optimization in the power grid; α k is the unit reactive power regulation cost coefficient of the kth static var generator; Q SVG，k (t) is the reactive power output of the kth static var generator at time period t; ω2 is the weight coefficient of the voltage deviation item; N is the total number of nodes included in the voltage deviation calculation in the distribution network; V i，t is the actual voltage value of node i at the time period; is the rated voltage value of node i at time period t.

[0038] Based on the above method item embodiments, the present invention correspondingly provides electronic device item embodiments.

[0039] An embodiment of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the distribution network voltage control and scheduling method based on a static var generator according to any one of the above method item embodiments.

[0040] Based on the above method item embodiments, the present invention correspondingly provides storage medium item embodiments.

[0041] An embodiment of the present invention provides a storage medium, on which a computer program is stored. Among them, when the computer program runs, it controls the device where the storage medium is located to execute the distribution network voltage control and scheduling method based on a static var generator according to any one of the above method item embodiments.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] An embodiment of the present invention provides a method, device, electronic device, and storage medium for voltage control and scheduling of a distribution network based on a static var generator. The method first obtains the topological structure of the distribution network, the hardware parameters of the static var generator, and a photovoltaic inverter group agent trained based on a power flow calculation model and an actor-critic multi-agent algorithm; then, with the goal of minimizing the weighted sum of the operating cost of the static var generator and the voltage deviation, a collaborative decision-making voltage model including constraints such as power, voltage, and current is constructed; in each round of model iteration, the objective function value is calculated by combining the candidate reactive power scheduling strategy with the autonomous response of the photovoltaic inverter group, and finally an optimal reactive power scheduling strategy is generated to guide the actual scheduling of the static var generator.

[0044] The present invention fully exploits the potential role of distributed photovoltaic inverters in voltage regulation by introducing a photovoltaic inverter group agent and considering its dynamic power response behavior in the scheduling model. Specifically, in the solution process of the collaborative decision-making voltage model in each round, the response of the photovoltaic inverter group agent to the candidate static var generator scheduling strategy is dynamically evaluated based on the power flow calculation model, so as to realize the coordinated control of the SVG and the photovoltaic inverter group. Compared with the traditional method that only relies on the independent regulation of the SVG, this method can more accurately reflect the characteristics of multi-source collaborative regulation in the actual distribution network, improve the adaptability and regulation accuracy of the scheduling strategy to system state changes, and thus effectively enhance the reliability of the voltage control and scheduling of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 FIG. is a schematic flowchart of a method for voltage control and scheduling of a distribution network based on a static var generator provided by an embodiment of the present invention.

[0046] Figure 2 FIG. is a schematic structural diagram of a device for voltage control and scheduling of a distribution network based on a static var generator provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0048] As Figure 1 shown, to solve the problem of inaccurate voltage control and scheduling of the distribution network caused by not considering the dynamic response behavior of the photovoltaic inverter group in the prior art, an embodiment of the present invention provides a method for voltage control and scheduling of a distribution network based on a static var generator, which at least includes the following steps:

[0049] Step S1: Obtain the topological structure of the power distribution network to be scheduled, the hardware parameters of the static var generators in the power distribution network to be scheduled, and the photovoltaic inverter group agents; wherein, the photovoltaic inverter group agents are obtained through interactive training by an actor-critic multi-agent algorithm based on a preset power flow calculation model and according to the electrical parameters of the photovoltaic inverter group in the power distribution network to be scheduled.

[0050] Specifically, this step involves obtaining several key data related to the power distribution network to be scheduled. This includes detailed topological structure information of the power distribution network, such as the connection relationships of each node and branch, and the model specifications of the lines. These information are the basis for subsequent power grid analysis and power flow calculation. At the same time, it is also necessary to obtain the hardware parameters of the static var generators (SVG) installed in the power distribution network to be scheduled, such as the upper and lower limits of their rated reactive power compensation capacity, the allowed adjustment rate, and possible operating cost characteristics. These parameters limit the operable range and economic considerations of the SVG during the scheduling process.

[0051] In addition to the above physical parameters and device characteristics, this step also needs to obtain the trained photovoltaic inverter group agents. These agents are the key to the present invention for realizing the coordinated scheduling of SVG and distributed photovoltaics. It should be noted that the acquisition of the photovoltaic inverter group agents is not generated out of thin air, but obtained through a careful training process. Specifically, the photovoltaic inverter group agents are based on a preset or constructed power flow calculation model that can accurately reflect the physical characteristics of the power distribution network to be scheduled, and fully according to the actual electrical parameters of each photovoltaic inverter or its group in the power distribution network to be scheduled (such as inverter capacity, power factor adjustment range, response characteristics, etc.), and further through a multi-agent reinforcement learning algorithm with an actor-critic architecture, after sufficient interactive training and learning. This training process enables each photovoltaic agent to learn the strategy of autonomously adjusting its power output to support the stability of the grid voltage under different grid states.

[0052] It should be noted here that for the common radial or weakly looped topological structures of the power distribution network, the DistFlow model framework is usually preferably used as the basis for its mathematical expression in the power flow calculation model; its core technical feature lies in the use of the second-order cone relaxation (SOCP) convex optimization technology, which effectively transforms the non-convex equality constraints inherent in the original power flow equation and difficult to directly and efficiently solve into a set of convex second-order cone inequality constraints. This transformation process aims to ensure that when performing subsequent power grid state analysis based on this model or embedding optimization calculations in this model, it can effectively avoid falling into local optimal solutions and significantly improve the global convergence and calculation efficiency of the solution; the relaxation process of the second-order cone relaxation is as follows:

[0053]

[0054] where υ i，t is the square of the voltage amplitude of the i-th node at time period t; l ij，t is the square of the current amplitude of the branch connecting node i and node j at time period t; P ij，t is the active power on the branch flowing from node i to node j at time period t; Q ij，t is the reactive power on the branch flowing from node i to node j at time period t.

[0055] In a preferred embodiment, the photovoltaic inverter group agent is obtained in the following manner:

[0056] According to the power flow calculation model, a dynamic operation simulation environment of the distribution network is constructed; wherein, the power flow calculation model is generated by the topological structure and line impedance parameters of the distribution network to be scheduled;

[0057] In the dynamic operation simulation environment of the distribution network, based on the Markov decision process framework, the state space, action space, reward function and training objective of the agent are defined, and an actor network, a discriminator network and an experience replay buffer are constructed;

[0058] Based on the actor-discriminator multi-agent algorithm, the photovoltaic inverters in the photovoltaic inverter group of the distribution network to be scheduled are regarded as agents, and the agents interact with the dynamic operation simulation environment of the distribution network, and the agents are trained until a preset number of training times is reached, generating a trained actor network and a trained discriminator network; wherein, during the training process, the agents generate state transition experience samples through interaction with the simulation environment and store them in the experience replay buffer; the network parameters of the actor network and the discriminator network are optimized by sampling the experiences in the experience replay buffer;

[0059] According to the trained actor network and the trained discriminator network, a photovoltaic inverter group agent is generated.

[0060] Specifically, first, according to the power flow calculation model, a dynamic operation simulation environment for agent training is constructed. It should be emphasized that the power flow calculation model is carefully generated in advance by obtaining and processing the detailed topological structure data of the distribution network to be scheduled and the accurate impedance parameters (including resistance and reactance) of each line. It constitutes the physical basis of the simulation environment and can simulate the power flow distribution and voltage response of the power grid under different disturbances or control actions. This simulation environment can respond to the control actions output by the agents and feedback the corresponding changes in the power grid state, providing the necessary interaction conditions for the learning of the agents.

[0061] Subsequently, in the constructed dynamic operation simulation environment of the distribution network, the learning problem of the agent is regulated following the basic framework of the Markov decision process (MDP). This includes clearly defining for each PV inverter agent the state space, action space, reward function for evaluating the quality of its behavior, and the training objective to be achieved, based on which it makes decisions.

[0062] In a preferred embodiment, the state space of the agent includes the current voltage measurement value at the connection point of the PV inverter, the current active power output value of the PV inverter, and the current reactive power output value of the PV inverter. This state information ensures that the agent can perceive the key features of its local power grid environment and its own operating conditions.

[0063] In a preferred embodiment, the action space of the agent includes setting the target value of the reactive power output and the target value of the active power output of the PV inverter. Through these actions, the agent can directly affect its power injection into the power grid, thus participating in voltage regulation.

[0064] The reward function is designed to guide the agent to learn the desired voltage support behavior. For example, in a preferred embodiment, the reward function can be designed as:

[0065] R t = R voltage，t + R reactive，t + R active，t + R terminal，t

[0066] In the formula, R t is the immediate reward obtained by the agent at time t; R voltage，t is the voltage deviation reward term; R reactive，t is the reactive power regulation cost term; R active，t is the active power curtailment penalty term; R terminal，t is the iteration termination penalty term.

[0067] In one embodiment, the voltage deviation reward term R voltage，t , specifically, is:

[0068] R υoltage，t = -w υ · (V PCC，t - V N ) 2

[0069] In the formula, V PCC，t is the actual voltage value at the connection point of the agent at time t; V N is the rated voltage value of the agent's connection point; w υ is the penalty weight coefficient of the voltage deviation;

[0070] In one embodiment, the reactive power regulation cost item R reactive，t , specifically:

[0071] R reactive，t =-w q ·(Q out，t ) 2

[0072] In the formula, Q out，t is the reactive power value output by the agent at time t; w q is the penalty weight coefficient for reactive power output;

[0073] In one embodiment, the active power curtailment penalty item R active，t , specifically:

[0074] R active，t =-w p ·(P avail，t -P out，t )

[0075] In the formula, P avail，t is the maximum active power value that the agent can generate at time t; P out，t is the actual active power value output by the agent at time t; w p is the penalty weight coefficient for active power curtailment.

[0076] In one embodiment, the iteration termination penalty item R terminal，t , specifically:

[0077] If the solution fails at time t, then

[0078] R terminal，t =-C failure +c duration ·t episode_duration

[0079] If the solution does not fail, then R terminal，t =0.

[0080] Meanwhile, in order to support value- and policy-based reinforcement learning, it is also necessary to construct an internal Actor Network and Critic Network for each agent, and establish an experience replay cache mechanism to store and utilize the valuable experience data generated during the training process.

[0081] The training process itself is carried out based on an actor-critic multi-agent reinforcement learning algorithm. Under the framework of this algorithm, each (or each group) of photovoltaic inverters in the photovoltaic inverter cluster of the power distribution network to be scheduled is abstracted and modeled as an independent agent. These agents learn and optimize their control strategies through a large number of continuous interactions with the dynamic operation simulation environment of the power distribution network. In each round of interaction, the agent will output an action (i.e., a power regulation instruction) according to its current state and the actor network, and the simulation environment will update the grid state according to this action and return a reward signal.

[0082] The training objective is to drive each agent to learn an optimal action policy π * . In a preferred embodiment, this optimal policy π * aims to maximize the expected cumulative discounted reward starting from any state s, usually expressed as the state value function

[0083]

[0084] or the action value function

[0085]

[0086] where γ is the discount factor (0 < γ ≤ 1), which is used to balance the importance of immediate reward and future reward. By optimizing this objective, the agent can learn to take power regulation actions that are most beneficial to long-term voltage stability and economic operation in various grid states.

[0087] This process will train the agent until its policy performance reaches the preset convergence criterion or completes the predetermined number of training times / rounds, and finally generate a trained actor network and a trained critic network with optimized parameters. During the entire training process, the agent will generate a large number of state transition experience samples (usually containing information such as the current state, executed action, obtained reward, next state, etc.) through interaction with the simulation environment, and these samples are effectively stored in the experience replay buffer. The learning algorithm will periodically sample historical experience data from the experience replay buffer to calculate the loss function and use optimization algorithms such as gradient descent to update and adjust the internal network parameters of its actor network and critic network, so as to gradually improve the effectiveness of the policy.

[0088] Finally, based on the actor network (mainly reflecting the policy) and the critic network (mainly used for value function evaluation and auxiliary policy learning) that have completed training and carry optimized network parameters, a trained photovoltaic inverter cluster agent that can autonomously respond and participate in the coordinated control of the grid voltage is formed.

[0089] Step S2: Based on the topological structure and the hardware parameters, with the goal of minimizing the weighted sum of the comprehensive operating cost of the static var generator and the voltage deviation of the distribution network, and using the scheduling strategy of the static var generator as the decision variable, construct a collaborative decision-making voltage model and the constraint conditions of the collaborative decision-making voltage model; the constraint conditions include reactive power over-limit constraint, node voltage constraint, and branch current constraint.

[0090] Specifically, this step is based on the detailed topological structure information of the previously obtained distribution network and the hardware parameters of the static var generator (such as the rated reactive power capacity of each SVG, the allowable adjustment range, and the relevant operating cost coefficients, etc.), with the goal of minimizing the weighted sum of the comprehensive operating cost of the static var generator and the voltage deviation of the entire distribution network as the core optimization objective, and constructs a collaborative decision-making voltage model through mathematical programming. In this model, the core decision variable is set as the specific scheduling strategy of each static var generator in one or more future scheduling periods, that is, the reactive power value that each SVG should generate or absorb in each period.

[0091] In order to ensure that the collaborative decision-making voltage model can generate a safe and effective scheduling strategy that meets the actual grid operation requirements when solving, a series of strict constraint conditions also need to be configured for this model. These constraint conditions define the feasible region for model solution. The constraint conditions mainly include: First is the reactive power over-limit constraint, which stipulates that the reactive power output of each static var generator must be within the maximum and minimum capacity ranges rated on its nameplate and cannot exceed its physical adjustment ability. Second is the node voltage constraint, which requires that the voltage amplitudes of all key nodes in the distribution network must be maintained within the allowable fluctuation range specified by national standards or industry regulations (such as ±7% or ±10% of the rated voltage) to ensure the normal operation of user equipment and power quality. Finally, it also includes the branch current constraint, that is, the current flowing through each transmission line or transformer branch in the distribution network cannot exceed its long-term allowable maximum current carrying capacity or thermal stability limit to prevent the line or equipment from being damaged due to overcurrent. These constraints together constitute the boundary conditions that must be strictly observed when solving the collaborative decision-making voltage model.

[0092] Through the constructed collaborative decision-making voltage model and its constraint conditions, it provides a clear mathematical framework and boundary for subsequently obtaining the optimal SVG scheduling strategy through iterative solution by an optimization algorithm.

[0093] In a preferred embodiment, the collaborative decision-making voltage model is specifically:

[0094]

[0095] Among them, F1 is the objective function value of the collaborative decision-making voltage model; ω1 is the weight coefficient of the operating cost item of the static var generator; T is the total number of time periods in the optimization cycle; N SVG is the total number of static var generators participating in the optimization in the power grid; α k is the unit reactive power regulation cost coefficient of the kth static var generator; Q SVG，k (t) is the reactive power output of the kth static var generator at time period t; ω2 is the weight coefficient of the voltage deviation item; N is the total number of nodes included in the voltage deviation calculation in the distribution network; V i，t is the actual voltage value of node i at time period t; is the rated voltage value of node i at time period t.

[0096] In a preferred embodiment, the reactive power over-limit constraint is specifically:

[0097]

[0098] In the formula, is the minimum reactive power output limit of the kth static var generator; is the maximum reactive power output limit of the kth static var generator;

[0099] The node voltage constraint is specifically:

[0100]

[0101] In the formula, υ i，t is the square of the voltage amplitude of the ith node at time period t; V i，t，min is the minimum allowable voltage amplitude of the ith node at time period t; V i，t，max is the maximum allowable voltage amplitude of the ith node at time period t;

[0102] The branch current constraint is specifically:

[0103]

[0104] In the formula, l ij，t is the square of the current amplitude of the branch connecting node i and node j at time period t; I ij，t，min is the minimum allowable current amplitude of branch ij at time period t; I ij，t，max is the maximum allowable current amplitude of branch ij at time period t.

[0105] Step S3: Under the above - mentioned constraint conditions, perform iterative solution on the collaborative decision - making voltage model for a preset number of rounds to generate the final scheduling strategy of the static var generator; wherein, in the process of solving the collaborative decision - making voltage model in each round, based on the power flow calculation model, determine the autonomous power response generated by the photovoltaic inverter group agent after the scheduling strategy of the candidate static var generator is injected into the distribution network; according to the autonomous power response and the candidate static var generator scheduling strategy, calculate and generate the objective function value of the collaborative decision - making voltage model in the current round based on the power flow calculation model; take the candidate static var generator scheduling strategy with the minimum objective function value in the preset number of rounds as the final scheduling strategy of the static var generator.

[0106] After constructing the collaborative decision - making voltage model and clarifying its constraint conditions, the embodiments of the present invention then perform iterative solution on the collaborative decision - making voltage model under the strict limitation of the constraint conditions. This process aims to generate the final scheduling strategy that can guide the optimal operation of the static var generator (SVG) through systematic search and evaluation. This iterative solution is usually driven by a suitable optimization algorithm (such as a gradient - based algorithm, a heuristic algorithm, or an algorithm built into a commercial mathematical programming solver) and is executed for a preset number of rounds or until a specific convergence criterion is met.

[0107] Among them, in the specific process of solving the collaborative decision - making voltage model in each round, the core is to evaluate the pros and cons of the currently considered candidate SVG scheduling strategy. This evaluation process first accurately determines, based on the preset power flow calculation model, the autonomous power response that the trained photovoltaic inverter group agent will generate after a candidate static var generator scheduling strategy (i.e., a set of assumed SVG reactive power output values) is injected into the distribution network. This step is crucial because it dynamically incorporates the intelligent behavior of distributed photovoltaics into the evaluation of SVG scheduling decisions. The power flow calculation model is used here to analyze the initial impact of the candidate SVG strategy on the grid state, thereby providing a decision - making basis for the photovoltaic agent.

[0108] After obtaining the autonomous power response of the photovoltaic agent, based on this autonomous power response data and the candidate static var generator scheduling strategy being evaluated currently, and again based on the power flow calculation model, a comprehensive calculation of the grid operation state is performed. The purpose of this calculation is to obtain detailed information such as the actual voltage distribution and line power flow of the distribution network under the combined action of the SVG candidate strategy and the photovoltaic autonomous response, and calculate and generate the objective function value of the collaborative decision - making voltage model in the current iteration round (i.e., the weighted sum of the SVG operation cost and the voltage deviation). This objective function value quantifies the comprehensive performance of the current candidate SVG strategy in terms of economy and voltage quality. By repeating this evaluation process, the optimization algorithm can compare the pros and cons of different candidate strategies.

[0109] Finally, after all the preset rounds of iteration are completed (or when other convergence conditions are met), the candidate static var generator scheduling strategy with the minimum objective function value will be selected as the final scheduling strategy for guiding the actual operation of the static var generator.

[0110] Through this iterative solution mechanism that takes into account the dynamic response of photovoltaic power, it can ensure the adaptability and overall optimization effect of the generated final scheduling strategy of SVG in a complex power grid environment.

[0111] Step S4: Schedule the static var generator in the distribution network to be scheduled according to the final scheduling strategy.

[0112] Specifically, according to the generated final scheduling strategy, perform real-time or pre-set time-section scheduling control on one or more static var generators in the distribution network to be scheduled.

[0113] This final scheduling strategy usually contains a series of instructions, indicating the specific values of the reactive power that each relevant SVG device should emit or absorb, or the target set points that its operating state should reach, in one or more future scheduling periods. The process of executing this scheduling instruction may involve sending corresponding control signals to the SVG device or its superior control system. For example, through the remote control terminal (RTU) or the energy management system (EMS / DMS) interface, the calculated reactive power setting values are transmitted to the local controllers of each SVG. After receiving the instruction, the local controller of the SVG will, based on its own control logic and the fast response ability of the power electronic converter, accurately adjust its reactive power output within the allowed adjustment rate and capacity range to match the requirements of the final scheduling strategy. During this process, it is also necessary to continuously monitor the actual operating state of the power grid to ensure the correct execution of the scheduling instruction and the safe and stable operation of the power grid.

[0114] By effectively converting the final scheduling strategy obtained from the optimization calculation into actual control actions for the static var generator, this method can implement the optimization benefits achieved at the model level in the operation of the physical power grid.

[0115] Based on the above method item embodiments, the present invention correspondingly provides device item embodiments.

[0116] As Figure 2 shown, an embodiment of the present invention provides a distribution network voltage control scheduling device based on a static var generator, including: a basic data acquisition module, a collaborative decision-making voltage model construction module, a collaborative decision-making voltage model solution module, and a static var generator scheduling module;

[0117] The basic data acquisition module is used to acquire the topological structure of the power distribution network to be scheduled, the hardware parameters of the static var generators in the power distribution network to be scheduled, and the photovoltaic inverter cluster agents; wherein, the photovoltaic inverter cluster agents are obtained through interactive training by an actor-critic multi-agent algorithm based on a preset power flow calculation model and according to the electrical parameters of the photovoltaic inverter cluster in the power distribution network to be scheduled.

[0118] The collaborative decision-making voltage model construction module is used to construct a collaborative decision-making voltage model and the constraint conditions of the collaborative decision-making voltage model with the goal of minimizing the weighted sum of the comprehensive operating cost of the static var generator and the voltage deviation of the power distribution network, taking the scheduling strategy of the static var generator as the decision variable; the constraint conditions include reactive power over-limit constraint, node voltage constraint, and branch current constraint.

[0119] The collaborative decision-making voltage model solving module is used to perform iterative solving of the collaborative decision-making voltage model for a preset number of rounds under the constraint conditions to generate the final scheduling strategy of the static var generator; wherein, in the process of solving the collaborative decision-making voltage model in each round, based on the power flow calculation model, determine the autonomous power response generated by the photovoltaic inverter cluster agents after the scheduling strategy of the candidate static var generator is injected into the power distribution network; according to the autonomous power response and the candidate static var generator scheduling strategy, calculate and generate the objective function value of the collaborative decision-making voltage model in the current round based on the power flow calculation model; take the candidate static var generator scheduling strategy with the minimum objective function value in the preset number of rounds as the final scheduling strategy of the static var generator.

[0120] The static var generator scheduling module is used to schedule the static var generators in the power distribution network to be scheduled according to the final scheduling strategy.

[0121] In a preferred embodiment, for the power distribution network voltage control scheduling device based on a static var generator, the collaborative decision-making voltage model construction module, the collaborative decision-making voltage model is specifically:

[0122]

[0123] Among them, F1 is the objective function value of the collaborative decision-making voltage model; ω1 is the weight coefficient of the operating cost item of the static var generator; T is the total number of time periods in the optimization period; N SVG is the total number of static var generators participating in the optimization in the power grid; α k is the unit reactive power regulation cost coefficient of the kth static var generator; Q SVG，k$Q_{k,t}$ is the reactive power output of the $k$-th static var generator at time $t$; $\omega_2$ is the weight coefficient of the voltage deviation term; $N$ is the total number of nodes in the distribution network included in the voltage deviation calculation; $V_i^t$ i，t is the actual voltage value of node $i$ at time $t$; $V_{i,rated}$ is the rated voltage value of node $i$ at time $t$.

[0124] It should be noted that the embodiments of the devices described above correspond to the above embodiments of the present invention, and can implement the voltage control and scheduling method of the distribution network based on static var generators described in any one of the above of the present invention. In addition, the embodiments of the above devices are only illustrative. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the drawings of the device embodiments provided by the present invention, the connection relationships between the modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement without creative efforts.

[0125] Based on the above method embodiments of the present invention, an embodiment of an electronic device is correspondingly provided.

[0126] An embodiment of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the voltage control and scheduling method of the distribution network based on static var generators described in any one of the present invention is implemented, or when the processor executes the computer program, the functions of each module in the above device embodiments are implemented.

[0127] Exemplarily, the computer program can be divided into one or more modules. The one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0128] The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0129] The so-called processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device, connecting various parts of the entire terminal device through various interfaces and lines.

[0130] The memory can be used to store the computer program and / or module. By running or executing the computer program and / or module stored in the memory, and by invoking the data stored in the memory, the processor realizes various functions of the terminal device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store the operating system, application programs required for at least one function, etc.; the data storage area can store the data created according to the use of the mobile phone, etc. In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0131] Based on the above method item embodiments, the present invention correspondingly provides storage medium item embodiments;

[0132] Another embodiment of the present invention provides a storage medium. The storage medium includes a stored computer program. When the computer program runs, it controls the device where the storage medium is located to execute any one of the above-mentioned distribution network voltage control and scheduling methods based on a static var generator of the present invention.

[0133] Among them, the above storage medium is a computer-readable storage medium, and the computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0134] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0135] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art in this technical field, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A distribution network voltage control and scheduling method based on a static var generator, characterized in that, Including: Obtain the topological structure of the to-be-scheduled distribution network, the hardware parameters of the static var generators in the to-be-scheduled distribution network, and the photovoltaic inverter cluster agent; wherein, the photovoltaic inverter cluster agent is obtained through interactive training of the actor-discriminator multi-agent algorithm based on the preset power flow calculation model and according to the electrical parameters of the photovoltaic inverter cluster in the to-be-scheduled distribution network; According to the topological structure and the hardware parameters, with the goal of minimizing the weighted sum of the comprehensive operating cost of the static var generator and the voltage deviation of the distribution network, and taking the scheduling strategy of the static var generator as the decision variable, construct a collaborative decision-making voltage model and the constraint conditions of the collaborative decision-making voltage model; the constraint conditions include reactive power over-limit constraint, node voltage constraint, and branch current constraint; Under the constraint conditions, perform iterative solution of the collaborative decision-making voltage model for a preset number of rounds to generate the final scheduling strategy of the static var generator; wherein, in the process of solving the collaborative decision-making voltage model in each round, based on the power flow calculation model, determine the autonomous power response generated by the photovoltaic inverter cluster agent after the scheduling strategy of the candidate static var generator is injected into the distribution network; according to the autonomous power response and the candidate static var generator scheduling strategy, based on the power flow calculation model, calculate and generate the objective function value of the collaborative decision-making voltage model in the current round; take the candidate static var generator scheduling strategy with the minimum objective function value in the preset number of rounds as the final scheduling strategy of the static var generator; According to the final scheduling strategy, schedule the static var generators in the to-be-scheduled distribution network.

2. The distribution network voltage control and scheduling method based on a static var generator according to claim 1, wherein Obtain the photovoltaic inverter cluster agent in the following way: According to the power flow calculation model, construct a dynamic operation simulation environment for the distribution network; wherein, the power flow calculation model is generated through the topological structure and line impedance parameters of the to-be-scheduled distribution network; In the dynamic operation simulation environment of the distribution network, define the state space, action space, reward function, and training objective of the agent based on the Markov decision process framework, and construct an actor network, a discriminator network, and an experience replay buffer; Based on the actor-discriminator multi-agent algorithm, regard the photovoltaic inverters in the photovoltaic inverter cluster of the to-be-scheduled distribution network as agents, and interact with the dynamic operation simulation environment of the distribution network through the agents to train the agents until the preset number of training times is reached, generating a trained actor network and a trained discriminator network; wherein, in the training process, the agents generate state transition experience samples through interaction with the simulation environment and store them in the experience replay buffer; optimize the network parameters of the actor network and the discriminator network by sampling the experiences in the experience replay buffer; Generate the photovoltaic inverter cluster agent according to the trained actor network and the trained discriminator network.

3. The distribution network voltage control and scheduling method based on a static var generator according to claim 2, characterized in that, The collaborative decision-making voltage model is specifically: Among them, F1 is the objective function value of the collaborative decision-making voltage model; ω1 is the weight coefficient of the operating cost item of the static var generator; T is the total number of time periods in the optimization cycle; N SVG is the total number of static var generators participating in the optimization in the power grid; α k is the unit reactive power regulation cost coefficient of the kth static var generator; Q SVG，k (t) is the reactive power output of the kth static var generator at time period t; ω2 is the weight coefficient of the voltage deviation item; N is the total number of nodes included in the voltage deviation calculation in the distribution network; V i，t is the actual voltage value of node i at time period t; is the rated voltage value of node i at time period t.

4. The distribution network voltage control and scheduling method based on a static var generator according to claim 3, characterized in that, The reactive power over-limit constraint is specifically: wherein is the minimum reactive power output limit of the k-th static var generator; is the maximum reactive power output limit of the k-th static var generator; The node voltage constraint is specifically: where, v i，t is the square of the voltage amplitude of the i-th node at time period t; V i，t，min is the minimum allowable voltage amplitude of the i-th node at time period t; V i，t，max is the maximum allowable voltage amplitude of the i-th node at time period t; The branch current constraint is specifically: where \(l\) ij，t is the square of the current amplitude of the branch connecting node \(i\) and node \(j\) at time period \(t\); I ij，t，min is the minimum current amplitude allowed for branch ij during time period t; I ij，t，max is the maximum current amplitude allowed for branch ij during time period t.

5. The distribution network voltage control and scheduling method based on a static var generator according to claim 4, wherein The state space of the agent includes the current voltage measurement value at the grid connection point of the PV inverter, the current active power output value of the PV inverter, and the current reactive power output value of the PV inverter.

6. The distribution network voltage control and scheduling method based on a static var generator according to claim 5, characterized in that, The action space of the agent includes the current reactive power output target value and the active power output target value of the PV inverter.

7. A distribution network voltage control and scheduling device based on a static var generator, characterized in that, It includes: A basic data acquisition module, a collaborative decision-making voltage model construction module, a collaborative decision-making voltage model solution module, and a static var generator scheduling module; The basic data acquisition module is used to obtain the topological structure of the power distribution network to be scheduled, the hardware parameters of the static var generators in the power distribution network to be scheduled, and the PV inverter cluster agents; among them, the PV inverter cluster agents are obtained through interactive training by the actor-critic multi-agent algorithm based on the preset power flow calculation model and the electrical parameters of the PV inverter cluster in the power distribution network to be scheduled. The collaborative decision-making voltage model construction module is used to construct a collaborative decision-making voltage model and its constraint conditions with the goal of minimizing the weighted sum of the comprehensive operating cost of the static var generator and the voltage deviation of the power distribution network, taking the static var generator scheduling strategy as the decision variable; the constraint conditions include reactive power over-limit constraint, node voltage constraint, and branch current constraint. The collaborative decision-making voltage model solution module is used to iteratively solve the collaborative decision-making voltage model for a preset number of rounds under the constraint conditions to generate the final scheduling strategy of the static var generator; among them, in the process of solving the collaborative decision-making voltage model in each round, based on the power flow calculation model, determine the autonomous power response generated by the PV inverter cluster agents after the scheduling strategy of the candidate static var generator is injected into the power distribution network; according to the autonomous power response and the candidate static var generator scheduling strategy, calculate and generate the objective function value of the collaborative decision-making voltage model in the current round based on the power flow calculation model; take the candidate static var generator scheduling strategy with the minimum objective function value in the preset number of rounds as the final scheduling strategy of the static var generator. The static var generator scheduling module is used to schedule the static var generators in the power distribution network to be scheduled according to the final scheduling strategy.

8. The distribution network voltage control and scheduling device based on a static var generator according to claim 7, wherein, The collaborative decision-making voltage model construction module, the collaborative decision-making voltage model is specifically: Among them, F1 is the objective function value of the collaborative decision-making voltage model; ω1 is the weight coefficient of the operating cost item of the static var generator; T is the total number of time periods in the optimization cycle; N SVG is the total number of static var generators participating in the optimization in the power grid; α k is the unit reactive power regulation cost coefficient of the kth static var generator; Q SVG，k (t) is the reactive power output of the kth static var generator at time period t; ω2 is the weight coefficient of the voltage deviation item; N is the total number of nodes included in the voltage deviation calculation in the distribution network; V i，t is the actual voltage value of node i at time period t; is the rated voltage value of node i at time period t.

9. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the static var generator-based power distribution network voltage control and scheduling method according to any one of claims 1 to 6.

10. A storage medium, characterized in that, The storage medium includes a stored computer program, wherein when the computer program runs, it controls the device where the storage medium is located to execute the static var generator-based power distribution network voltage control and scheduling method according to any one of claims 1 to 6.

Images