Distribution network maximum access microgrid capacity dynamic optimization solution method considering new energy output uncertainty

Abstract

The invention provides a distribution network maximum access microgrid capacity dynamic optimization solution method considering new energy output uncertainty, and the method comprises the steps: carrying out the modeling of schedulable resources in a distribution network, and obtaining a day-ahead resource model; constructing a power distribution network-microgrid group maximum accessible capacity simplified evaluation model; optimizing the power distribution network-microgrid group maximum accessible capacity simplified evaluation model based on opportunity constraint to obtain a power distribution network-microgrid group maximum accessible capacity optimization model under a day-ahead-day scale; and constructing a power prediction model based on reinforcement learning, and coupling and solving the power prediction model based on reinforcement learning and the maximum access capacity optimization model of the power distribution network-microgrid group under the day-day-intra-day scale based on the real-time output power of the new energy and the day-day resource model to obtain a real-time intra-day scheduling scheme.

Description

Technical Field

[0001] This invention belongs to the field of power grid resource optimization and scheduling, and specifically relates to a dynamic optimization solution method for the maximum access microgrid capacity of the distribution network considering the uncertainty of new energy output. Background Technology

[0002] Integrating and scheduling distributed resources in the form of microgrids represents a new structural form for controlling high-proportion distributed resources in the future. The access of a large number of distributed resources alters the power flow distribution of the distribution network, increasing the risk of voltage exceeding limits. Furthermore, microgrids can profit by participating in the distribution network ancillary services market using their adjustable capacity. Therefore, it is urgent to conduct real-time optimization research on the maximum accessible microgrid capacity of the distribution network under uncertainties and considering constraints from different operating scenarios.

[0003] Regarding the optimization and evaluation model for the maximum accessible microgrid group capacity of distribution networks, current domestic and international research mainly focuses on the carrying capacity of microgrids for their internal distributed resources, with less research on optimization and evaluation models for the maximum accessible capacity of microgrid groups within distribution networks. A data-driven carrying capacity evaluation model combining situational awareness methods has not been established, and how to efficiently utilize measurement data to achieve operational scenario prediction and rapid response remains to be studied. Methods such as optimized scheduling, reliability assessment, and multi-scenario technologies can effectively improve the carrying capacity of distribution networks for new energy sources and achieve collaborative evaluation of multi-level distribution networks. However, current engineering simplification does not fully investigate the evolution of distribution network carrying capacity under different security domains. Simultaneously, using digital twin technology to intuitively present the maximum accessible microgrid group capacity helps improve the efficiency of operators in actual production. However, optimization involves only a single entity, and research on the carrying capacity of new distribution systems under multi-entity interaction is still limited, and the carrying capacity evaluation of distribution networks involving microgrid groups is not considered. Summary of the Invention

[0004] Based on the above analysis, the problems and defects of the existing technology are as follows: (1) Most existing studies focus on the maximum accessible capacity of a single type of distributed resource, and there is little research on the maximum accessible capacity of microgrids within the distribution network after the integration of distributed resources; (2) Currently, the research on the capacity optimization and evaluation model of microgrids mainly focuses on the optimization of accessible capacity under deterministic new energy output and typical output scenario library, without considering the impact of its uncertainty on actual scheduling; (3) Currently, existing studies mainly focus on the capacity optimization of microgrids under the daytime scale, without considering the fluctuation of new energy output of the distribution network under the intraday time scale, and cannot fully explore the carrying potential of the distribution network for microgrids. To overcome the shortcomings of the existing technologies, this invention provides a dynamic optimization solution method for the maximum access capacity of distribution networks and microgrids that considers the uncertainty of renewable energy output. This method constructs a simplified evaluation model of the maximum accessible capacity of the distribution network-microgrid cluster, taking into account application scenario limitations. Then, through opportunity constraint optimization, an optimization model for the maximum access capacity of the distribution network-microgrid cluster at the day-ahead and intraday scales is obtained. At the day-ahead scale, discrete variables such as distribution network topology and energy storage charging / discharging are determined. At the intraday scale, continuous variables such as distribution network energy storage output and photovoltaic output are continuously optimized. This finds the optimal balance between safety and economy, achieving dynamic optimization scheduling of the maximum access capacity of the distribution network and microgrid clusters.

[0005] According to one aspect of the present invention, a method for dynamically optimizing the maximum connected microgrid capacity of a distribution network considering the uncertainty of new energy output is provided, comprising:

[0006] The dispatchable resources in the distribution network are modeled to obtain the day-ahead resource model;

[0007] A simplified evaluation model for the maximum accessible capacity of the distribution network-microgrid cluster is constructed, including: considering the constraints of the distribution network's active reconfiguration behavior and the coordinated scheduling of wind, solar, load and storage resources within the microgrid cluster; and an objective function that aims to maximize the accessible capacity of the distribution network to the microgrid cluster and minimize the scheduling cost of the distribution network-microgrid cluster.

[0008] Based on the opportunity-constrained optimization of the simplified evaluation model of the maximum accessible capacity of the distribution network-microgrid, we obtain the optimization model of the maximum accessible capacity of the distribution network-microgrid at the day-ahead and intraday scales.

[0009] A power prediction model based on reinforcement learning is constructed. Based on the real-time power output of new energy sources and the day-ahead resource model, the power prediction model based on reinforcement learning is coupled with the maximum access capacity optimization model of distribution network-microgrid group at the day-ahead-intraday scale to obtain the real-time intraday scheduling scheme.

[0010] As a further technical solution, the current resource model is represented by constraints on multiple resources, including: energy storage-related constraints, flexible load-related constraints, new energy output-related constraints, capacitor-related constraints, and SVC-related constraints.

[0011] As a further technical solution, the constraints of the distribution network’s active reconfiguration behavior and the multi-resource coordinated scheduling of wind, solar, load and storage within the microgrid are considered, including: power balance constraints within the microgrid, constraints related to the distribution network’s active reconfiguration, power flow constraints, distribution network state constraints, and total power balance constraints of the distribution network.

[0012] As a further technical solution, a power prediction model based on reinforcement learning is proposed. The real-time output power of the new energy source is defined as the environmental state, and the predicted output power of the new energy source during the prediction period is defined as the action. Based on the actor-commentator method, the predicted output power of the new energy source during the prediction period is output based on the real-time output power of the new energy source.

[0013] As a further technical solution, the maximum access capacity optimization model for distribution network-microgrid groups at the day-ahead-intraday scale includes: a risk-stochastic optimization day-ahead scheduling model and an intraday scheduling model;

[0014] The process of constructing the objective function of the risk stochastic optimization day-ahead scheduling model is as follows:

[0015] Based on opportunity constraints, the output power of new energy sources is transformed into random variables that follow a specific empirical probability distribution. According to the objective function of the simplified evaluation model of the maximum accessible capacity of the distribution network-microgrid, the expected value function of the day-ahead dispatch cost of the distribution network-microgrid under the influence of random variables is calculated.

[0016] Based on the value at risk function, calculate the day-ahead operational risk cost measurement function under the influence of random variables;

[0017] By combining the day-ahead operational risk cost metric function and the expected value function of the distribution network-microgrid group scheduling cost, the objective function of the day-ahead scheduling model with risk stochastic optimization is obtained.

[0018] As a further technical solution, the process of coupling the power prediction model based on reinforcement learning with the distribution network-microgrid group maximum access capacity optimization model at the day-ahead-intraday scale includes:

[0019] Substitute the day-ahead resource model into the day-ahead scheduling model with risk stochastic optimization to obtain the day-ahead optimal scheduling scheme;

[0020] Using a power prediction model based on reinforcement learning, and based on real-time monitoring of renewable energy power output and the day-ahead optimal scheduling scheme, the predicted power output of renewable energy during the prediction period is output.

[0021] Based on the predicted power output of new energy sources within the forecast period, the intraday scheduling model is solved to obtain the real-time intraday scheduling scheme.

[0022] As a further technical solution, the process of solving the intraday dispatch model based on the predicted output power of new energy sources within the forecast period includes: taking the forecast period as the optimization period, taking the predicted output power of new energy sources within the forecast period as the input value, and taking the maximum capacity of the distribution network that can be connected to the microgrid group and the minimum dispatch cost of the distribution network-microgrid group as the objective, to obtain the real-time intraday dispatch scheme.

[0023] According to another aspect of the present invention, a dynamic optimization solution system for the maximum connected microgrid capacity of the distribution network considering the uncertainty of new energy output is provided, comprising:

[0024] The schedulable resource modeling module is used to model schedulable resources in the distribution network to obtain the day-ahead resource model;

[0025] A simplified evaluation model construction module is used to construct a simplified evaluation model for the maximum accessible capacity of the distribution network-microgrid cluster. This includes: constraints considering the active reconfiguration behavior of the distribution network and the coordinated scheduling of multiple resources such as wind, solar, load and storage within the microgrid cluster; and an objective function that aims to maximize the accessible capacity of the distribution network to the microgrid cluster and minimize the scheduling cost of the distribution network-microgrid cluster.

[0026] A multi-scale optimization model construction module is used to optimize a simplified evaluation model of the maximum accessible capacity of the distribution network-microgrid group based on opportunity constraints, and to obtain the optimization model of the maximum accessible capacity of the distribution network-microgrid group at the day-ahead and intraday scales.

[0027] The real-time intraday scheduling scheme acquisition module is used to construct a power prediction model based on reinforcement learning. Based on the real-time output power of new energy sources and the day-ahead resource model, the power prediction model based on reinforcement learning is coupled with the maximum access capacity optimization model of distribution network-microgrid group at the day-ahead-intraday scale to obtain the real-time intraday scheduling scheme.

[0028] According to another aspect of this specification, an electronic device is provided, including a memory and a processor, wherein the memory stores program instructions that are executed by the processor, and the processor invokes the program instructions to execute a dynamic optimization solution method for the maximum connected microgrid capacity of the distribution network considering the uncertainty of new energy output.

[0029] According to another aspect of this specification, a non-transitory computer-readable storage medium is provided, the non-transitory computer-readable storage medium storing computer instructions that cause the computer to execute a dynamic optimization solution method for the maximum access microgrid capacity of the distribution network considering the uncertainty of new energy output.

[0030] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention constructs a simplified evaluation model of the maximum accessible capacity of the distribution network-microgrid group considering the limitations of application scenarios, and then obtains an optimization model of the maximum accessible capacity of the distribution network-microgrid group at the day-ahead and intraday scales through chance constraint optimization. At the day-ahead scale, discrete variables such as energy storage charging and discharging are determined, and at the intraday scale, continuous variables such as distribution network energy storage output and photovoltaic output are continuously optimized. The optimal balance between safety and economy is found, and dynamic optimization scheduling of the maximum accessible microgrid group capacity of the distribution network is realized.

[0031] This invention incorporates the uncertainty of renewable energy output into the optimization method for the maximum capacity of microgrids that can be connected to the distribution network. This allows for a dynamic balance between safety and cost, ensuring grid safety margins under extreme output conditions. Simultaneously, based on minimizing the expected operating cost for application scenarios, this invention proposes a multi-timescale optimization method ("day-ahead to intraday") to accurately quantify the cost of mitigating operational risks for different capacity access schemes. This avoids conservative planning that restricts renewable energy development while mitigating safety risks through dynamic margins, thus promoting the transformation of the distribution network to a "observable, measurable, and controllable" state under high-proportion renewable energy scenarios. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 A flowchart illustrating a method for dynamically optimizing the maximum microgrid capacity of the distribution network considering the uncertainty of new energy output, provided in an embodiment of the present invention;

[0034] Figure 2 A schematic diagram of a dynamic optimization solution system for the maximum access microgrid capacity of the distribution network considering the uncertainty of new energy output is provided in an embodiment of the present invention.

[0035] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0036] It should be noted that:

[0037] The terms “comprising” and “having”, and any variations thereof, in the specification, claims, and accompanying drawings of this invention are intended to cover a non-exclusive inclusion, such as a process, method, system, product, or apparatus that includes a series of steps or units, not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0038] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices. The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be decomposed, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined to form new technical solutions. Such combinations are not bound by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0040] like Figure 1 As shown, a dynamic optimization method for determining the maximum microgrid capacity of a distribution network considering the uncertainty of new energy output includes:

[0041] Step 1: Model the dispatchable resources in the distribution network to obtain the day-ahead resource model;

[0042] Step 2: Construct a simplified evaluation model for the maximum accessible capacity of the distribution network-microgrid cluster, including: considering the constraints of the distribution network's active reconfiguration behavior and the coordinated scheduling of wind-solar-load-storage resources within the microgrid cluster, and the objective function that aims to maximize the accessible capacity of the distribution network to the microgrid cluster and minimize the scheduling cost of the distribution network-microgrid cluster.

[0043] Step 3: Optimize the simplified evaluation model of the maximum accessible capacity of the distribution network-microgrid group based on opportunity constraints to obtain the optimization model of the maximum accessible capacity of the distribution network-microgrid group at the day-ahead and intraday scale;

[0044] Step 4: Construct a power prediction model based on reinforcement learning. Based on the real-time power output of new energy sources and the day-ahead resource model, couple the power prediction model based on reinforcement learning with the maximum access capacity optimization model of the distribution network-microgrid group at the day-ahead-intraday scale to obtain the real-time intraday scheduling scheme.

[0045] In step 1, the day-ahead resource model is represented by constraints on multiple resources, including: energy storage-related constraints, flexible load-related constraints, new energy output-related constraints, capacitor-related constraints, and SVC (Static Var Compensator)-related constraints.

[0046] Specifically, in this application, the distribution network is assumed to have new energy sources connected to it. The power output of these new energy sources is limited by the actual engineering scenario; in this application, new energy sources are assumed to be wind power and photovoltaic power. Optionally, the types of new energy sources can be increased or decreased according to the actual engineering scenario.

[0047] The current resource model essentially reflects the actual engineering scenario constraints of all dispatchable resources in the distribution network. Optionally, it can be expressed as the constraints of multi-resource dispatching of new energy output, load, and energy storage, including:

[0048] A1-1. Energy storage-related constraints:

[0049] (a) Charge and discharge power constraints:

[0050]

[0051] In the formula, , , , These are the real-time charging and discharging power of energy storage and the maximum charging and discharging power of energy storage, respectively. , These are the energy storage charging and discharging indicators. and Time-based energy storage and charging, and Energy is stored and discharged at any time.

[0052] (b) Apparent power constraint:

[0053]

[0054] In the formula, For real-time energy storage, This refers to the rated capacity of the energy storage.

[0055] (c) Charge state constraints:

[0056]

[0057] In the formula, , These are the real-time SOC state of energy storage and the initial SOC state of energy storage, respectively. , These represent the maximum and minimum SOC values ​​for energy storage, For energy storage charging and discharging efficiency, The unit is the scheduling cycle.

[0058] A1-2, Flexible Load Related Constraints:

[0059]

[0060] in, , These represent the real-time active power and typical active power of flexible loads within the microgrid group, respectively. Maximum adjustable power for flexible loads; The change in electricity consumption of flexible loads within a specified time period. , These are indicators of flexible load increase / decrease, and they are discrete variables. and This indicates that the flexible load increases power. and This indicates that the flexible load reduces power. and This indicates that the flexible load will not be adjusted.

[0061] A1-3. Constraints Related to New Energy Output:

[0062] (a) Power upper and lower limit constraints:

[0063]

[0064] In the formula, , These are the real-time active power of wind power and solar power, respectively. , This represents the maximum active power of wind / solar power.

[0065] (b) Inverter capacity constraints:

[0066]

[0067] In the formula, , These are the real-time reactive power values ​​for wind power and solar power, respectively. , These are the apparent power of wind power / photovoltaic inverters, respectively.

[0068] A1-4. Constraints related to reactive power compensation equipment:

[0069] (a) Capacitor-related constraints:

[0070]

[0071] In the formula, , These represent the total reactive power of the capacitor in real time and the reactive power that a single capacitor can provide, respectively. , These refer to the number of capacitors to be installed and the maximum number that can be installed, respectively.

[0072] (b) SVC related constraints:

[0073]

[0074] In the formula, , These are the real-time reactive power and maximum reactive power of the SVC, respectively.

[0075] In step 2, the optimization model for the maximum accessible capacity of the distribution network to the microgrid includes: considering the constraints of the distribution network's active reconfiguration behavior and the coordinated scheduling of wind-solar-load-storage resources within the microgrid group, as well as the objective function that aims to maximize the accessible capacity of the distribution network to the microgrid group and minimize the scheduling cost of the distribution network to the microgrid group.

[0076] Among them, the constraints of the active reconfiguration behavior of the distribution network and the multi-resource coordinated scheduling of new energy output, load and energy storage in the microgrid group include: internal power balance constraints of the microgrid group, active reconfiguration related constraints of the distribution network, power flow constraints, distribution network state constraints and total power balance constraints of the distribution network.

[0077] Specifically, this patent adopts active reconfiguration of the distribution network and multi-resource coordinated scheduling of new energy output, load, and energy storage within the microgrid group. Different requirements exist in engineering applications such as peak shaving and valley filling, and power fluctuation smoothing. The relevant constraints are as follows:

[0078] A2-1. Power balance constraints within a microgrid:

[0079]

[0080] In the formula, , These represent the energy storage charging and discharging power at time t, respectively. The active power output of flexible loads within microgrid i. , These represent the active power output of photovoltaic and wind power, respectively. This represents the maximum active power output from microgrid i to the distribution network. The reactive power output of flexible loads within microgrid i. , These represent the reactive power output of photovoltaic and wind power, respectively. This represents the reactive power output from microgrid i to the distribution network.

[0081] A2-2. Constraints related to active reconfiguration of distribution networks:

[0082]

[0083] In the formula, , These are the main line set and the transformer line set, Let mn be the connectivity of line mn at time t. This indicates that line mn is connected at time t. This indicates that line mn is disconnected at time t. , These represent the total number of nodes in the distribution network and the number of nodes connected to substations, respectively. , These represent the disconnection / connection indication values ​​of transformer line mn at time t. =1 indicates that the transformer line mn is disconnected at time t. This indicates that the transformer line mn is closed at time t.

[0084] A2-3, Power Flow Constraints:

[0085]

[0086] In the formula, , Let m and n be the active power and reactive power on line mn at time t, respectively. , Let be the node voltages of node m and node n at time t, respectively. , These are the conductance and susceptance of line mn, respectively. Let m be the phase angle difference of line mn at time t.

[0087] A2-4. Constraints on Distribution Network Status Quantities:

[0088]

[0089] In the formula, , These are the maximum and minimum allowable voltages for the distribution network. Let be the current flowing through line mn at time t. , These represent the maximum and minimum allowable currents for the line, respectively.

[0090] A2-5. Total power balance constraints of the distribution network:

[0091]

[0092] In the formula, , These refer to the power purchased and sold by the distribution network to the upper-level power grid. , Let t represent the active and reactive loads at node i in the distribution network at time t. Let m be the reactive power transmitted on line mn at time t. , , , , These represent the total number of distribution network loads, nodes, microgrids, capacitors, and SVC devices, respectively.

[0093] In step 2, the simplified evaluation model for the maximum accessible capacity of the distribution network-microgrid group aims to maximize the accessible capacity of the distribution network-microgrid group while minimizing the scheduling cost of the distribution network-microgrid group.

[0094] Specifically, the simplified evaluation model for the maximum accessible capacity of the distribution network-microgrid group in this application adopts a two-layer scheduling method based on the objective cascading method. The upper-layer model schedules the distribution network with the objective of maximizing the accessible capacity of the microgrid group, while the lower-layer model schedules the distribution network-microgrid group with the objective of minimizing the scheduling cost. The upper and lower-layer models are solved iteratively, and the convergence of the model is accelerated by the objective cascading method.

[0095] The objective functions for the upper and lower layers are shown in the following equations:

[0096]

[0097] In the formula, F up F represents the upper-level objective function; bottom This represents the lower-level objective function; , , These are the unit operating cost of energy storage, the unit adjustment cost of flexible load, and the unit electricity purchase cost of the distribution network. The total scheduling cycle.

[0098] Step 3, the optimization model for the maximum access capacity of the distribution network-microgrid group at the day-ahead and intraday scales, includes: a risk-stochastic optimization day-ahead scheduling model and an intraday scheduling model. During the day-ahead period, given the determined output of new energy sources, the decision on whether to use energy storage is made. During the intraday period, the predicted output of new energy sources within the forecast period is considered to determine the output size of energy storage.

[0099] Furthermore, the objective function of the risk stochastic optimization day-ahead scheduling model is constructed as follows:

[0100] Based on opportunity constraints, the output power of new energy sources is transformed into random variables that follow a specific empirical probability distribution. According to the objective function of the simplified evaluation model of the maximum accessible capacity of the distribution network-microgrid, the expected value function of the day-ahead dispatch cost of the distribution network-microgrid under the influence of random variables is calculated.

[0101] Based on the value at risk function, calculate the day-ahead operational risk cost measurement function under the influence of random variables;

[0102] By combining the day-ahead operational risk cost metric function and the expected value function of the distribution network-microgrid group scheduling cost, the objective function of the day-ahead scheduling model with risk stochastic optimization is obtained.

[0103] Specifically, the core idea of ​​the risk stochastic optimization day-ahead scheduling model is to transform traditional hard constraints into probabilistic flexible constraints that hold at a certain confidence level. That is, this method treats random variables such as photovoltaic output as probability distributions (e.g., normal or empirical distributions), proposes key constraints that consider uncertainties (e.g., power balance, voltage safety limits), and establishes a quantitative trade-off mechanism between system safety and economy. Essentially, it is a chance-constrained uncertainty modeling of new energy sources, studying the optimization of the maximum access capacity of distribution networks and microgrids at the day-ahead scale. The derivation process is as follows:

[0104] The new distribution network integrates a large number of distributed power sources, including new energy power generation units (wind turbines and photovoltaic power plants in this application), gas turbines, and energy storage systems. Due to the random fluctuations in wind speed and solar intensity, the active power output of wind turbines and photovoltaic power plants in the new distribution network exhibits strong uncertainty. Actual dispatching requires providing predicted active power outputs for wind and photovoltaic power; however, prediction technologies are limited.

[0105] This application employs chance-constrained stochastic optimization, assuming that the active power output of wind power and photovoltaic power are random variables and follow a specific empirical probability distribution. The expected value of the day-ahead dispatching scheme for the maximum access capacity of the distribution network-microgrid group under the influence of these random variables is calculated. Minimizing this expected value yields a day-ahead optimization dispatching model based on stochastic optimization, mathematically expressed as:

[0106]

[0107] In the formula, Calculate the expected value; For the decision and state variables in the day-ahead optimization scheduling model, during the day-ahead optimization process, the decision variables are discrete variables such as energy storage activation and deactivation indicators; To optimize the random variables describing uncertain fluctuations in the scheduling model; , These are the equality constraints and inequality constraints in the model, consistent with the simplified evaluation model for the maximum accessible capacity of the distribution network-microgrid group.

[0108] The objective function of the aforementioned stochastic day-ahead optimization scheduling model only considers the expected value of operating costs, while the actual operating costs in the future may deviate significantly from the expected value. To mitigate this high-cost risk as much as possible, a suitable risk metric is needed to describe it. Value at Risk (VaR) can calculate a specific operating risk cost such that the probability that the possible future operating costs will not exceed the risk cost satisfies a given confidence level.

[0109] To address the issue that VaR cannot measure the risk cost of the right-hand tail of probability, this invention uses the CvaR metric, calculated as follows:

[0110]

[0111] In the formula, CVaR represents the conditional value-at-risk function; inf represents the infimum; r is the minimum upper bound to be determined, i.e., the VaR value; and α is the confidence level. Indicates when At that time ,when The time is 0.

[0112] After constructing an uncertainty model of the system operation risk measurement cost using the CvaR metric, the system operation risk measurement cost and expected value cost are weighted and summed to construct the objective function of the day-ahead scheduling model with risk stochastic optimization. The day-ahead scheduling model with risk stochastic optimization is shown in the following equation:

[0113]

[0114] In the formula, This is the risk weighting coefficient.

[0115] In step 4, the power prediction model based on reinforcement learning defines the real-time output power of the new energy source as the environmental state and the predicted output power of the new energy source in the next prediction period as the action. Based on the actor-commentator method, the predicted output power of the new energy source in the prediction period is output based on the real-time output power of the new energy source.

[0116] Specifically, reinforcement learning focuses on intelligent agents or decision-making agents. These agents perceive the state of the environment and perform actions based on a specific strategy. The environment then provides feedback to the agent regarding the next state and reward, initiating the next stage of the interaction process, which is repeated. The training process for a power prediction model based on reinforcement learning includes:

[0117] A4-0-1. Construct a reinforcement learning training environment model.

[0118] Specifically, assuming each microgrid dispatch center is the decision-making entity and the microgrid system containing renewable energy sources is the environment, then the reinforcement learning-based renewable energy output prediction model can be transformed into a typical multi-step decision problem. The variables in the decision-making process are defined as follows:

[0119] (1) The state space S is the real-time power output of new energy sources that can be monitored by each microgrid dispatch center.

[0120] (2) Action space A is defined as the predicted output power of new energy obtained by each microgrid dispatch center through calculation in the next instruction cycle.

[0121] (3) Return space R, the instantaneous return in the target return space is defined as the negative number of the difference between the photovoltaic predicted power and the typical photovoltaic power curve in this period.

[0122] (4) State transition relationship P: Since the environment is the actual operating power distribution network, the state transition relationship naturally satisfies the timing relationship during the operation of the system.

[0123] (5) Report discount rate A higher return discount rate indicates a greater focus on the long-term return impact. In this application, we take... It is a fixed value of 0.9.

[0124] A4-0-2. Train the constructed reinforcement learning training environment model and output the trained model as a power prediction model based on reinforcement learning.

[0125] Specifically, the model's input is the state of the Markov decision process of the renewable energy output power of the microgrid at each moment. However, its matrix has high dimensionality and low information density, and directly using it may lead to too many parameters and make training difficult. This application adopts a deep voltage regulation network, using a 3-layer convolutional neural network to extract key features and concatenate them with variables in the action space A to obtain the input matrix.

[0126] After receiving the input matrix, the actor network begins reinforcement learning. During reinforcement learning, it uses the policy gradient method to optimize the action policy when interacting with the environment. After collecting the effective interactions between the actor network and the environment, the critic network updates the actor network parameters according to the defined loss function and optimization objective, thus completing the model training.

[0127] As a supplement, considering the economics of system operation and the adjustment costs of adjustable resources within the microgrid, the objective function for the maximum capacity of the distribution network connected to the microgrid group consists of the loss function of the commentator network and the optimization objective.

[0128] Furthermore, there are currently three commonly used reinforcement learning algorithms: value function-based methods, policy-based methods, and actor-to-critic (A2C) methods. Given the advantages of A2C methods such as high convergence and high data utilization, this application adopts the A2C method to design a deep reinforcement learning model for real-time prediction of renewable energy power output.

[0129] Specifically, based on real-time monitoring data, the algorithm steps for real-time prediction of new energy power output using a reinforcement learning-based power prediction model are as follows:

[0130] (1) Strategy evaluation: Use the current strategy and collected data to update the actor network (Q-network).

[0131] (2) Strategy improvement: Update the actor network based on the critics’ value estimates to generate better strategies.

[0132] (3) Value network update: Update the critic network (V-network) to make its estimated state value more accurate.

[0133] Step 5, the process of coupling the reinforcement learning-based power prediction model with the distribution network-microgrid maximum access capacity optimization model at the day-ahead-intraday scale includes:

[0134] Substitute the day-ahead resource model into the day-ahead scheduling model with risk stochastic optimization to obtain the day-ahead optimal scheduling scheme;

[0135] Using a power prediction model based on reinforcement learning, and based on real-time monitoring of renewable energy power output and the day-ahead optimal scheduling scheme, the predicted power output of renewable energy during the prediction period is output.

[0136] Based on the predicted power output of new energy sources within the forecast period, the intraday scheduling model is solved to obtain the real-time intraday scheduling scheme.

[0137] Furthermore, the process of solving the intraday scheduling model based on the predicted output power of new energy sources within the prediction period includes: taking the prediction period as the optimization period, taking the predicted output power of new energy sources within the prediction period as the input value, and taking the maximum capacity of the distribution network that can be connected to the microgrid group and the minimum scheduling cost of the distribution network-microgrid group as the objective, to solve and obtain the real-time intraday scheduling scheme.

[0138] Specifically, the intraday scale mainly considers the impact of the uncertainty of microgrid renewable energy output on the distribution network voltage stability and the capacity of microgrids that can be connected. Therefore, at this scale, with the goal of maximizing voltage stability and the capacity that can be connected, a certain time window is constructed. Within each time window, the predicted renewable energy output is used for rolling optimization to achieve the operability of the maximum connected capacity. The decision variables include continuous variables such as energy storage charging and discharging power and the interaction power between the distribution network and the microgrid. The constraints are the same as those in the simplified evaluation model of the maximum connected capacity of the distribution network-microgrid group. The objective function is shown in the following equation:

[0139]

[0140] In the formula, , These are the objective functions for the distribution network layer and the microgrid layer on the intraday time scale, respectively. , These represent the initial time point and the length of the i-th time window, respectively.

[0141] Due to uncertainties such as photovoltaic power output, the maximum access capacity of the distribution network-microgrid cluster at the day-ahead scale cannot be accurately described by day-ahead dispatch. This invention determines discrete variables such as energy storage charging and discharging at the day-ahead scale, and continuously optimizes continuous variables such as energy storage output and photovoltaic output of the distribution network cluster at the intraday scale, finding the optimal balance between safety and economy.

[0142] The implementation of the various embodiments of this invention is based on programmed processing through a system with processor functionality. Therefore, in practical engineering, the technical solutions and functions of the various embodiments of this invention are encapsulated into various modules. Based on this reality, and building upon the above embodiments, the embodiments of this invention provide a dynamic optimization solution system for the maximum connected microgrid capacity of the distribution network considering the uncertainty of renewable energy output. This system is used to execute a dynamic optimization solution method for the maximum connected microgrid capacity of the distribution network considering the uncertainty of renewable energy output from the above method embodiments.

[0143] See Figure 2 The system includes:

[0144] The system comprises the following modules: a dispatchable resource modeling module for modeling dispatchable resources in the distribution network to obtain a day-ahead resource model; a simplified evaluation model construction module for constructing a simplified evaluation model for the maximum accessible capacity of the distribution network-microgrid cluster, including: constraints considering the active reconfiguration behavior of the distribution network and the coordinated scheduling of wind-solar-load-storage resources within the microgrid cluster, and an objective function aiming to maximize the accessible capacity of the distribution network to the microgrid cluster and minimize the scheduling cost of the distribution network-microgrid cluster; a multi-scale optimization model construction module for optimizing the simplified evaluation model for the maximum accessible capacity of the distribution network-microgrid cluster based on chance constraints, obtaining an optimized model for the maximum accessible capacity of the distribution network-microgrid cluster at the day-ahead and intraday scales; and a real-time intraday scheduling scheme acquisition module for constructing a power prediction model based on reinforcement learning. Based on the real-time output power of new energy sources and the day-ahead resource model, the power prediction model based on reinforcement learning is coupled with the optimized model for the maximum accessible capacity of the distribution network-microgrid cluster at the day-ahead and intraday scales to obtain a real-time intraday scheduling scheme.

[0145] It should be noted that the system embodiments provided by the present invention are used not only to implement the methods in the above method embodiments, but also to implement the methods in other method embodiments provided by the present invention. The only difference is that corresponding functional modules are set. The principle is basically the same as that of the above system embodiments provided by the present invention. As long as those skilled in the art can improve the modules in the above system embodiments by referring to the specific technical solutions in other method embodiments and combining technical features to obtain corresponding technical means and technical solutions composed of these technical means, on the basis of the above system embodiments, and on the premise of ensuring the practicality of the technical solutions, they can obtain corresponding system-like embodiments for implementing the methods in other method-like embodiments.

[0146] The method in this embodiment of the invention is implemented using an electronic device; therefore, it is necessary to introduce the relevant electronic device. For this purpose, embodiments of the present invention provide an electronic device, such as... Figure 3 As shown, the electronic device includes: at least one processor, a communication interface, at least one memory, and a communication bus, wherein the at least one processor, the communication interface, and the at least one memory communicate with each other via the communication bus. The at least one processor invokes logical instructions stored in the at least one memory to execute all or part of the steps of the methods provided in the foregoing method embodiments.

[0147] Furthermore, when the logical instructions in at least one of the aforementioned memories are implemented as software functional units and sold or used as independent products, they are stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, is embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (a personal computer, server, or network device) to execute all or part of the steps of the methods described in the various method embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks—various media for storing program code.

[0148] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, located in one place, or distributed across multiple network units. The purpose of this embodiment is achieved by selecting some or all of the modules according to actual needs. Those skilled in the art will understand and implement this without any inventive effort.

[0149] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0150] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0151] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0152] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0153] Based on the same technical concept as the foregoing embodiments, the present invention provides a non-transitory computer-readable storage medium that stores computer instructions that cause the computer to execute a dynamic optimization solution method for the maximum access microgrid capacity of the distribution network considering the uncertainty of new energy output.

[0154] In summary, this invention provides a dynamic optimization method for the maximum microgrid capacity of a distribution network considering the uncertainty of renewable energy output. This method includes: studying an optimization model for the maximum accessible capacity of the distribution network to the microgrid under typical operating scenarios; proposing a renewable energy output prediction method based on reinforcement learning; and proposing a multi-timescale optimization method for the maximum accessible capacity of the distribution network-microgrid group based on reinforcement learning. By incorporating the uncertainty of renewable energy output into the optimization method for the maximum accessible capacity of the distribution network to the microgrid group, this invention establishes a dynamic balance between safety and cost, ensuring the safety margin of the grid under extreme output conditions. Simultaneously, based on minimizing the expected cost of the scenario, this invention proposes a multi-timescale optimization method based on the day-to-day and intra-day scenarios, accurately quantifying the compensation cost of different capacity schemes for operational risks. This avoids conservative planning that restricts the development of renewable energy and reduces safety risks by reserving dynamic margins, promoting the transformation of the distribution network to achieve the goal of being "observable, measurable, and controllable" in scenarios with a high proportion of renewable energy.

[0155] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic optimization solution method for maximum access micro-grid capacity of distribution network considering new energy output uncertainty, characterized in that, The application relates to a power grid-microgrid group maximum accessible capacity optimization method. The application comprises the following steps: a day-ahead resource model is established by modeling the dispatchable resources in a power grid; a power grid-microgrid group maximum accessible capacity simplified evaluation model is constructed, which comprises: constraint conditions considering the active reconstruction behavior of the power grid and the collaborative scheduling of wind, light, load and storage multi-resources in the microgrid group, and an objective function taking the maximum microgrid group accessible capacity of the power grid and the minimum scheduling cost of the power grid-microgrid group as the target; the power grid-microgrid group maximum accessible capacity simplified evaluation model is optimized based on the chance constraint, and a power grid-microgrid group maximum accessible capacity optimization model in the day-ahead-intra-day scale is obtained; 2. The method of claim 1, wherein the method further comprises: a power prediction model based on reinforcement learning is constructed, the power prediction model based on reinforcement learning is coupled with the power grid-microgrid group maximum accessible capacity optimization model in the day-ahead-intra-day scale based on the real-time output power of the new energy and the day-ahead resource model, and the real-time intra-day scheduling scheme is obtained.

3. The method of claim 1, wherein the method further comprises: The day-ahead resource model is represented by the constraint conditions of new energy output-load-storage multi-resources, and comprises the following constraint conditions: a storage related constraint, a flexible load related constraint, a new energy output related constraint, a capacitor related constraint and an SVC related constraint.

4. The method of claim 1, wherein the method further comprises: The constraint conditions considering the active reconstruction behavior of the power grid and the collaborative scheduling of wind, light, load and storage multi-resources in the microgrid group comprise: a microgrid group internal power balance constraint, a power grid active reconstruction related constraint, a power flow constraint, a power grid state quantity constraint and a power grid total power balance constraint.

5. The method of claim 1, wherein the method further comprises: The power prediction model based on reinforcement learning defines the real-time output power of the new energy as an environment state, defines the predicted output power of the new energy in a prediction period as an action, and outputs the predicted output power of the new energy in the prediction period based on the real-time output power of the new energy according to the actor-critic method. The power grid-microgrid group maximum accessible capacity optimization model in the day-ahead-intra-day scale comprises: a risk stochastic optimization day-ahead scheduling model and an intra-day scheduling model. The objective function of the risk stochastic optimization day-ahead scheduling model is constructed in the following manner: based on the chance constraint, the new energy output power is converted into a random variable subject to a specific empirical probability distribution, and the expected value function of the day-ahead power grid-microgrid group scheduling cost under the influence of the random variable is calculated according to the objective function of the power grid-microgrid group maximum accessible capacity simplified evaluation model; based on the risk value function, the day-ahead operation risk cost measurement function under the influence of the random variable is calculated; 6. The method of claim 5, wherein the method further comprises: the day-ahead operation risk cost measurement function and the expected value function of the power grid-microgrid group scheduling cost are combined to obtain the objective function of the risk stochastic optimization day-ahead scheduling model. The coupling solving process of the power prediction model based on reinforcement learning and the power grid-microgrid group maximum accessible capacity optimization model in the day-ahead-intra-day scale comprises: the day-ahead resource model is substituted into the risk stochastic optimization day-ahead scheduling model to obtain a day-ahead optimal scheduling scheme; the power prediction model based on reinforcement learning is used to output the predicted output power of the new energy in a prediction period based on the real-time monitoring quantity of the new energy output power and the day-ahead optimal scheduling scheme; the intra-day scheduling model is solved based on the predicted output power of the new energy in the prediction period to obtain a real-time intra-day scheduling scheme.

7. The method of claim 6, wherein the method further comprises: The process of solving the intraday scheduling model based on the predicted output power of the new energy in the prediction period includes: taking the prediction period as the optimization period, taking the predicted output power of the new energy in the prediction period as the input value, and taking the maximum capacity of the microgrid group accessible to the power distribution network and the minimum scheduling cost of the power distribution network-microgrid group as the target to solve and obtain the real-time intraday scheduling scheme.

8. A system for solving dynamic optimization of maximum access micro-grid capacity of a power distribution network considering uncertainty of new energy output, characterized in that, The method comprises the following steps: a schedulable resource modeling module is configured to model the schedulable resources in the power distribution network to obtain a day-ahead resource model; a simplified evaluation model construction module is configured to construct a simplified evaluation model of the maximum accessible capacity of the power distribution network-microgrid group, including: considering the constraint conditions of the active reconstruction behavior of the power distribution network and the coordinated scheduling of the wind-solar-storage resources in the microgrid group, and taking the maximum capacity of the microgrid group accessible to the power distribution network and the minimum scheduling cost of the power distribution network-microgrid group as the objective function; a multi-scale optimization model construction module is configured to optimize the simplified evaluation model of the maximum accessible capacity of the power distribution network-microgrid group based on the chance constraint to obtain the maximum access capacity optimization model of the power distribution network-microgrid group in the day-ahead-intraday scale; a real-time intraday scheduling scheme acquisition module is configured to construct a power prediction model based on reinforcement learning, and to couple the power prediction model based on reinforcement learning with the maximum access capacity optimization model of the power distribution network-microgrid group in the day-ahead-intraday scale based on the real-time output power of the new energy and the day-ahead resource model to solve and obtain the real-time intraday scheduling scheme.

9. An electronic device, comprising: The non-transitory computer readable storage medium stores computer instructions, and the computer instructions make the computer execute the method of any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, comprising: The non-transitory computer readable storage medium stores computer instructions, and the computer instructions make the computer execute the method of any one of claims 1 to 7.

Images