Method for optimizing electric quantity of energy storage power station based on graph calculation

Abstract

The invention belongs to the technical field of energy storage power stations, and particularly relates to an energy storage power station electric quantity optimization method based on graph calculation, which comprises the following steps: S1, predicting future power generation capacity and load demand of an energy storage power station based on graph calculation: acquiring historical power generation data and load data of a distributed power station in a grid-connected area of the energy storage power station, and external influence data, constructing a time-space diagram neural network model, and obtaining prediction results of the power generation data and the load data in each hour period of the next 24 hours; s2, calculating the power and load value of each node in the power grid through power flow analysis; and according to the predicted future power generation capacity and the load demand, the electric quantity needing to be provided or consumed by the energy storage power station is calculated. Optimization of energy of the power storage station is achieved, and the cost of the power storage station is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy storage power stations, and specifically relates to a method for optimizing the power of an energy storage power station based on graph calculation. Background Art

[0002] At present, wind power and photovoltaic power generation are widely connected to the power system. However, both wind power and photovoltaic power generation are intermittent and volatile, which aggravates the peak load regulation pressure of the power system and inevitably leads to wind and photovoltaic power abandonment. The existence of energy storage can effectively alleviate the problem of wind and photovoltaic power abandonment, realize the consumption of wind power and photovoltaic power generation, reduce the production cost of the power system and reduce carbon emissions. However, energy storage power stations have the following problems:

[0003] High safety and reliability requirements: The power balance between power generation and load is an important basis for the safe operation of the power system. The access of energy storage systems, especially large-scale battery energy storage systems, must ensure their safety and reliability.

[0004] Low prediction accuracy: The power grid environment is complex. Traditional time series prediction methods are often based on statistical models, such as ARIMA and VAR. However, for complex time series data, the prediction accuracy is often not high.

[0005] There is a waste of electricity in energy storage power stations, which also require large investments and have a long payback period. Summary of the invention

[0006] In order to solve the above technical problems, namely, the energy storage power station has a waste of electricity and high operating costs, the present invention provides the following technical solutions:

[0007] A method for optimizing the power of an energy storage power station based on graph computing comprises the following steps:

[0008] Step S1, predicting the future power generation capacity and load demand of the energy storage power station based on graph calculation: obtaining the historical power generation data and load data of the distributed power stations in the grid-connected area of ​​the energy storage power station, as well as the external impact data, constructing the power grid topology structure, and constructing a spatiotemporal graph neural network model based on the power grid topology structure, and predicting the power generation capacity and load demand of each hour in the next 24 hours through the spatiotemporal graph neural network model;

[0009] Step S2: Obtain the technical parameters and real-time operation data of the distribution network in the power grid topology structure, calculate the power and load values ​​of each node in the power grid through power flow analysis, and output the power and voltage of each node in the power grid as the power flow analysis result; and calculate the amount of electricity that the energy storage power station needs to provide or consume based on the predicted future power generation capacity and load demand.

[0010] Furthermore, step S1 specifically includes the following steps:

[0011] Step S11, data collection and preprocessing: obtaining historical power generation data of distributed power stations including solar and wind power stations in the energy storage power station grid-connected area, grid load data, holiday data, weather data, and equipment operation status data in the energy storage power station grid-connected area; preprocessing the collected data, including cleaning data, processing missing values ​​and outliers, and normalizing data, to obtain relevant time series data of the energy storage power station grid-connected area after preprocessing;

[0012] Step S12, constructing a power grid topology graph structure: converting the time series data processed in step S11 into a graph structure, mapping the timestamp and attribute information in the time series data to the nodes and edges in the graph structure, defining nodes, edges and their attributes, where nodes represent power stations, substations, and load points, and edges represent paths for power flow, including transmission lines or distribution lines, and distribution line edges, to obtain a power grid topology graph structure;

[0013] Step S13, using the power grid topology structure of step S12 as the input of the graph neural network model to construct a spatiotemporal graph neural network model;

[0014] Step S14: Use the historical data of step S11 to train and verify the spatiotemporal graph neural network model, evaluate the performance of the spatiotemporal graph neural network model by comparing the difference between the prediction results of the spatiotemporal graph neural network model and the actual power grid state, and make adjustments and optimizations;

[0015] Step S15, prediction and verification: Use the trained spatiotemporal graph neural network model to predict future power generation capacity and load demand, and evaluate the accuracy and generalization ability of the dynamic relationship graph model through cross-validation or holdout method

[0016] Furthermore, in step S13, a spatiotemporal graph neural network model is constructed, including:

[0017] The changing trend and spatial distribution characteristics of the power grid topology structure in the time series are calculated to obtain the embedded representation of the graph and input it into the encoder. The encoder then performs feature extraction and combines the attention mechanism to output a dynamic graph of the power grid status at a preset time in the future in the decoder, finally obtaining a spatiotemporal graph neural network model.

[0018] Furthermore, step S13 more specifically includes:

[0019] Firstly, the time series analysis technology is used to capture the changing trend of each node and its attributes in the power grid topology graph over time, and the power grid topology structure and time series data are converted into the embedded representation of the graph to obtain the embedded representation of the graph;

[0020] Input the graph embedding representation into the encoder;

[0021] Feature extraction combined with attention mechanism: The attention mechanism is introduced into the encoder to capture the key changes in the power grid. After being processed by the encoder, the extracted features are input into the decoder. The decoder outputs a dynamic graph of the power grid status within the preset hourly period in the future to obtain a spatiotemporal graph neural network model.

[0022] Further, specifically, step S2 includes: obtaining technical parameters and real-time operating data of the distribution network in the power grid topology structure, including the load and voltage of each node, inputting the power grid topology structure and real-time operating data into the power flow calculation tool, using the power flow calculation method to obtain the power flow analysis results of the distribution network, and outputting the power and voltage of each node in the power grid as the power flow analysis results.

[0023] Further, specifically, step S2 includes:

[0024] Step S21, obtain the power generation capacity and load demand for each hour of the next day predicted by the spatiotemporal graph neural network model, and obtain the power value of each intermediate node in the circuit through power flow calculation:

[0025] The spatiotemporal graph neural network model is used to predict the power generation capacity and load demand for each hour of the next day. The predicted power generation capacity and load demand are used as inputs to the power flow calculation tool. The power flow calculation is performed again, and the power and voltage of each node in the power grid are output as the power flow analysis results, thereby calculating the power value of each intermediate node in the circuit.

[0026] Step S22: Obtain the load forecast value of the intermediate node and the load value calculated based on the power flow analysis result, and calculate the amount of electricity that the energy storage power station needs to provide or consume:

[0027] Based on the results of the power flow analysis, the actual load value of the intermediate node is calculated. By comparing the load prediction value of the intermediate node with the load value of the power flow analysis, the amount of electricity that the energy storage power station needs to provide or consume is calculated.

[0028] Furthermore, it also includes:

[0029] Step S3: Based on the results of the power flow analysis, a profit maximization model is established to output the best profit model, including the way in which the power storage station participates in the power market, the charging power and discharging power in each hour period, and the charging time and discharging time;

[0030] Step S3 specifically includes:

[0031] Calculate the life cycle cost of energy storage, i.e. the levelized cost of storage (LCOS);

[0032] Using the life cycle cost LCOS of energy storage, determine the objective function and constraints for maximizing the economic benefits of energy storage power stations;

[0033] By using optimization methods to solve the optimal combination of energy storage participating in battery leasing, electricity market trading, and power auxiliary services, as well as the optimal charging and discharging power, charging and discharging time period, and charging and discharging duration, the economic income of the energy storage power station can be maximized.

[0034] Furthermore, the full life cycle cost of energy storage, namely the levelized cost of energy storage LCOS, is calculated using the following formula:

[0035]

[0036] Among them, d is the discharge time at rated power, η is the cycle efficiency of the energy storage station, T is the system life, n(t) is the number of cycles per year, r is the discount rate, Q&M(t) is the operation and maintenance cost in the tth year, C is the E Installation cost that varies with capacity, C P Installation costs vary with power.

[0037] Furthermore, the objective function of maximizing the economic benefits of the energy storage power station is expressed as:

[0038]

[0039] P is the price, represents the price of electricity sold in the wth electricity market activity mode during the d1th hour period, Represents the price of electricity purchased in the wth electricity market activity mode during the d1th hour period;

[0040] S is power, It represents the discharge power under the w-th electricity market activity mode during the d1-th hour period, and the negative sign represents discharge; It represents the charging power under the w-th electricity market activity mode during the d1-th hour period, and the positive sign represents charging;

[0041] It represents the discharge duration of participating in the wth electricity market activity mode in the d1th hour period, and the negative sign represents discharge; It represents the charging time of participating in the wth electricity market activity mode in the d1th hour period, and the positive sign represents charging;

[0042] Income rent is the rental income for the τ period.

[0043] Furthermore, the constraints of the objective function include energy storage capacity constraints, charging power and discharging power constraints, energy storage energy balance constraints, and initial and final energy level constraints of the energy storage system. The specific formula for energy storage capacity constraints is as follows:

[0044] E min -Erent <E d1 <E max -E rent

[0045]

[0046] Among them, E min Indicates the minimum energy level of energy storage, E max Indicates the maximum energy level of energy storage, E rent Indicates the capacity of the leased energy storage battery, E d1 Represents the electric energy at the energy storage time d1, and d1=1,2,3,...,τ.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. The present invention obtains the historical power generation data and load data of distributed power stations in the grid-connected area of ​​the energy storage power station, as well as external influence data such as holidays and weather data, to construct a power grid topology structure, and obtains the prediction results of power generation data and load data for each hour in the next 24 hours; obtains the power flow analysis results of the distribution network; outputs the power and voltage of each node in the power grid as the power flow analysis results; allocates the power of the energy storage power station on demand, saves the power of the energy storage power station, and reduces the power operation cost of the energy storage power station.

[0049] 2. According to the results of the power flow analysis, a profit maximization model is established to output the best profit model, including the participation method, the power of charging and discharging in each hour period, and the sufficient duration. The present invention realizes the optimization of energy and reduces the cost of the power storage station. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The present invention is a flow chart of the method for optimizing the power generation of an energy storage power station. DETAILED DESCRIPTION

[0051] The technical solution of the present invention will be clearly described below in conjunction with the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention, and all other embodiments obtained by ordinary technicians in the field without making creative work are within the protection scope of the present invention.

[0052] Peak-valley electricity price difference refers to the difference in electricity prices during peak and valley periods. Energy storage power stations can store energy when electricity prices are low and release energy when electricity prices are high, thereby earning profits from the price difference.

[0053] Ancillary services refer to the paid value-added services provided by power companies to users, such as peak load regulation (peak-valley electricity price difference), frequency regulation, reactive power regulation, etc. Energy storage power stations can earn revenue by providing these ancillary services.

[0054] Power flow calculation is a term in electrical engineering, which refers to the calculation of active power, reactive power and voltage distribution in the power grid under given power system network topology, component parameters and power generation and load parameters. Power flow calculation is the calculation of steady-state operating state parameters of each part of the power system based on given power grid structure, parameters and operating conditions of components such as generators and loads. Usually, the given operating conditions include the power of each power source and load point in the system, the voltage at the hub point, the voltage and phase angle at the balance point. The operating state parameters to be calculated include the voltage amplitude and phase angle of each bus node of the power grid, as well as the power distribution of each branch, the power loss of the network, etc.

[0055] like Figure 1 As shown, the present invention provides a method for optimizing the power of an energy storage power station based on graph calculation, comprising the following steps:

[0056] Step S1: Predict the future power generation capacity and load demand of the energy storage power station based on the graph calculation:

[0057] The historical power generation data and load data of distributed power stations in the grid-connected area of ​​the energy storage power station, as well as external impact data, are obtained to build a power grid topology structure. Based on the power grid topology structure, a spatiotemporal graph neural network model is built to predict the power generation demand and load demand for each hour in the next 24 hours.

[0058] Power generation capacity forecast: Consider the impact of weather forecasts (especially for solar and wind power), seasonal changes, and other factors on renewable energy power generation. Load demand forecast: Analyze historical data and consider the impact of factors such as weekdays and holidays, and temperature changes on load demand.

[0059] Step S1 specifically includes the following steps:

[0060] Step S11, data collection and preprocessing:

[0061] Obtain the historical power generation data of distributed power stations in the energy storage power station grid-connected area, including solar and wind power stations, grid load data in the energy storage power station grid-connected area, holiday data, weather data, and equipment operating status data. Preprocess the collected data, including cleaning data, processing missing values ​​and outliers, and normalizing data.

[0062] Step S12: constructing a power grid topology structure:

[0063] The time series data processed in step S11 is converted into a graph structure, and the timestamps and attribute information in the time series data are mapped to the nodes and edges in the graph structure.

[0064] Define nodes: In the graph, nodes represent power stations, substations, and load points. Define edges: Edges represent the path of power flow, which can be transmission lines or distribution lines. Transmission line edges are the power transmission paths connecting power stations and substations, or between substations; distribution line edges are the power distribution paths from substations to load points. Attribute node and edge assignment: Node attributes: such as power generation capacity, load level, holiday data, weather data, weather. Edge attributes: such as line capacity, etc.

[0065] Graph Neural Networks (GNNs), as a new neural network model, can process data with non-Euclidean structures, such as graphs, networks, etc. By converting time series data into graph structures and using graph neural networks for feature extraction and prediction, it is possible to process nonlinear relationships and complex time series data, thereby improving the accuracy of time series prediction.

[0066] Step S13: Use the power grid topology structure of step S12 as the input of the graph neural network model to construct a spatiotemporal graph neural network model:

[0067] The changing trend and spatial distribution characteristics of the power grid topology structure in the time series are calculated to obtain the embedded representation of the graph and input it into the encoder; the encoder then extracts features and combines the attention mechanism to output a dynamic graph of the power grid state at a preset time in the future in the decoder, and finally obtains a spatiotemporal graph neural network model. Specifically:

[0068] First, time series analysis techniques (such as difference, moving average, autoregressive model, etc.) are used to capture the changing trend of each node (such as power station, substation, load point) and its attributes (such as power generation capacity, load level, weather conditions, etc.) in the power grid topology graph structure over time. At the same time, the spatial relationship between nodes is analyzed, such as the similarity of adjacent nodes and the power flow between nodes, to capture the spatial distribution characteristics of the power grid, and the power grid topology and time series data are converted into an embedded representation of the graph to obtain an embedded representation of the graph.

[0069] Input the graph embedding representation to the encoder: The obtained graph embedding representation is used as the input of the encoder. The encoder is usually a graph neural network model used to extract the feature information of the graph. In the encoder, graph neural network models such as graph convolutional network (GCN) and graph attention network (GAT) can be used to process graph structure data and extract its high-level features.

[0070] Feature extraction with attention mechanism: Introduce the attention mechanism in the encoder to dynamically focus on important nodes and edges in the graph during feature extraction. The attention mechanism can assign different weights based on the attribute information of nodes and edges and the relationship between them, thereby capturing key changes in the power grid. The decoder outputs a dynamic graph of the power grid state in the future preset hour period: After being processed by the encoder, the extracted features are input into the decoder. The decoder is usually also a neural network model used to predict the future power grid state based on the extracted features.

[0071] The decoder can generate dynamic graphs for a period of time in the future, which reflect the changes in the properties of nodes and edges in the power grid over time. Finally, a spatiotemporal graph neural network model is obtained.

[0072] Step S14, model training: Divide the historical data preprocessed in step S11 into a training set, a validation set, and a test set. Use the historical data to train and validate the spatiotemporal graph neural network model. By comparing the difference between the prediction results of the spatiotemporal graph neural network model and the actual power grid state, the performance of the spatiotemporal graph neural network model can be evaluated, and necessary adjustments and optimizations can be made.

[0073] Choose loss function and optimizer: Define the loss function (such as mean squared error) and optimizer (such as Adam).

[0074] In the power forecasting task, the commonly used loss function is the mean square error (MSE). MSE calculates the average of the squares of the differences between the predicted value and the true value, which can reflect the accuracy of the model prediction. For each time step or hour period, the MSE between the predicted load or power generation and the actual load or power generation is calculated. During the training process, the goal is to minimize the MSE to improve the prediction accuracy of the model.

[0075] Commonly used optimizers include stochastic gradient descent (SGD), Adam, etc.

[0076] The model parameters are continuously updated through iterative training. In each round of iteration, forward propagation is first performed to calculate the model's predicted value, then the value of the loss function is calculated, and then back propagation is performed to calculate the gradient, and the optimizer is used to update the parameters.

[0077] During training, monitor the value of the loss function and the performance of the model on the validation set.

[0078] When the performance on the validation set no longer improves, training can be stopped to avoid overfitting.

[0079] Step S15, prediction and verification: Use the trained spatiotemporal graph neural network model to predict future power generation capacity and load demand, and evaluate the accuracy and generalization ability of the model through cross-validation or holdout method.

[0080] The trained spatiotemporal graph neural network model is used to predict the power generation capacity and load demand in the future.

[0081] The accuracy and generalization ability of the model are evaluated through cross-validation or holdout method. Common evaluation indicators include accuracy, recall, F1 score, MSE, MAE, etc.

[0082] Step S2: Obtain the technical parameters and real-time operation data of the distribution network in the power grid topology structure, calculate the power and load values ​​of each node in the power grid through power flow analysis, and output the power and voltage of each node in the power grid as the power flow analysis result; and calculate the amount of electricity that the energy storage power station needs to provide or consume based on the future power generation capacity and load demand predicted by the spatiotemporal graph neural network model. Specifically:

[0083] Obtain the technical parameters and real-time operation data of the distribution network in the power grid topology diagram, input the power grid topology diagram structure and real-time operation data into the power flow calculation tool, and use the power flow calculation method to obtain the power flow analysis results of the distribution network; output the power and voltage of each node in the power grid as the power flow analysis results. Real-time operation data includes power injection (or load) and voltage of each node.

[0084] Power flow analysis is an important task in the power system, which aims to calculate the power and voltage distribution of each node in the power grid. By obtaining the power grid topology diagram and the technical parameters and real-time operation data of the distribution network, we can use the power flow calculation tool to obtain the power flow analysis results of the power grid. These results are of great significance for the stable operation, optimal dispatch and fault analysis of the power system.

[0085] More specifically, step S2 includes:

[0086] 1) Obtain the power generation capacity and load demand for each hour of the next day predicted by graph calculation, and obtain the power value of each intermediate node in the circuit through power flow calculation.

[0087] Specifically, graph computing technology (such as spatiotemporal graph neural network model, etc.) is used to predict the power generation capacity and load demand for each hour of the next day. The predicted power generation capacity and load demand are used as inputs to the power flow calculation tool, and the power flow calculation is performed again. The power and voltage of each node in the power grid are output as the power flow analysis results, and the power value of each intermediate node in the circuit is obtained.

[0088] 2) Obtain the load forecast value of the intermediate node and the load value calculated based on the power flow analysis results, and calculate the amount of electricity that the energy storage power station needs to provide or consume.

[0089] Based on the results of the power flow analysis, the actual load value of the intermediate node is calculated. By comparing the load forecast value of the intermediate node with the load value of the power flow analysis, the amount of electricity that the energy storage power station needs to provide or consume is calculated. This can be achieved by calculating the difference between the two. A positive value indicates that the energy storage power station needs to provide electricity, and a negative value indicates that the energy storage power station needs to consume electricity.

[0090] Through the above steps, we can use power flow analysis technology to calculate the power and voltage of each node in the power grid, and calculate the amount of electricity that the energy storage power station needs to provide or consume based on the predicted future power generation capacity and load demand. This will provide strong support for the stable operation and optimized dispatch of the power system.

[0091] Step S3, establish a profit maximization model and output the best profit model: According to the results of the power flow analysis, establish a profit maximization model and output the best profit model, including the way the power storage station participates in the power market activities, the charging power and discharging power in each hour period, and the charging time and discharging time. Step S3 specifically includes:

[0092] 1) Calculate the full life cycle cost of energy storage, i.e. the levelized energy storage cost:

[0093] The full life cycle cost of energy storage is the levelized cost of storage (LCOS). LCOS can be summarized as the full life cycle cost of an energy storage technology divided by its cumulative transmitted electrical energy or power, reflecting the internal average electricity price when the net present value is zero, that is, the profit point of the investment. The levelized cost of storage (LCOS) quantifies the discounted cost of unit discharge under specific energy storage technology and application scenarios, taking into account all technical and economic parameters that affect the discharge life cost. The specific calculation formula and the indicators involved are as follows:

[0094]

[0095] The definition of each parameter is shown in Table 1:

[0096] Table 1

[0097] parameter unit definition d Hour Discharge time at rated power η ％ Energy storage station cycle efficiency T Year System life n(t) Times / year Annual cycles r ％ Discount rate Q&M(t) ％ Operation and maintenance costs in the tth year (installation ratio) <![CDATA[C E ]]> Yuan / kWh Installation costs as a function of capacity <![CDATA[C P ]]> Yuan / kW Installation costs as a function of power

[0098] The levelized cost of storage (LCOS) is a key indicator for evaluating the economic feasibility of energy storage technology. Its calculation formula takes into account the full life cycle cost of the energy storage system, including installation cost, operation and maintenance costs, and system life.

[0099] The total cost includes the installation cost that varies with capacity, the installation cost that varies with power, and the operation and maintenance costs (installation ratio) of each year, and the discount rate must be considered. The cumulative transmitted electric energy is the total electric energy transmitted by the energy storage system during its life. The specific definition and calculation method of each parameter have been given in the problem description.

[0100] 2) Using the full life cycle cost LCOS of energy storage, determine the objective function and constraints for maximizing the economic benefits of the energy storage power station, and maximize the economic benefits of the energy storage power station without affecting the operation of the power grid and the safety of the energy storage power station. The economic benefits of the energy storage power station mainly come from two aspects: one is the rental income obtained by renting the energy storage capacity to other users, and the other is the income obtained from participating in the electricity market (such as wind power, solar energy, frequency modulation, peak regulation, etc.). Therefore, the objective function for maximizing the economic benefits of the energy storage power station is expressed as:

[0101]

[0102] Income in the formula rent Represents rental income,

[0103] Represents the income obtained from participating in the electricity market (such as wind power, solar energy, frequency regulation, peak regulation, etc.).

[0104] Among them, τ is the total duration of the cycle, d1 is the hour segment, for example, the cycle of τ is 24 hours, and d1 is every hour in the cycle; n is the total number of ways in which the energy storage power station participates in the electricity market, and w represents the way in which the energy storage power station participates in the electricity market. For example, w=1 means that the energy storage power station participates in the electricity market in the wind power mode, w=2 means that the energy storage power station participates in the electricity market in the solar power mode, w=3 means that the energy storage power station participates in the electricity market in the frequency modulation mode, and w=4 means that the energy storage power station participates in the electricity market in the peak load regulation mode...;

[0105] P is the price, represents the price of electricity sold in the wth electricity market activity mode during the d1th hour period, Represents the price of electricity purchased in the wth electricity market activity mode during the d1th hour period;

[0106] S is power, It represents the discharge power under the w-th electricity market activity mode during the d1-th hour period, and the negative sign represents discharge; It represents the charging power under the w-th electricity market activity mode during the d1-th hour period, and the positive sign represents charging;

[0107] It represents the discharge duration of participating in the wth electricity market activity mode in the d1th hour period, and the negative sign represents discharge; It represents the charging time under the w-th electricity market activity mode within the d1-th hour period, and the positive sign represents charging.

[0108] The income in the objective function maxProfirent It reflects capacity leasing, which refers to renting the capacity of the energy storage power station to other users for use, thereby obtaining leasing income.

[0109] Income rent is the rental income of τ period, and its formula is:

[0110]

[0111] in represents the rental price of the energy storage battery within the τ period, E rent Indicates the capacity of the leased energy storage battery.

[0112] The constraints of the objective function include energy storage capacity constraints, charging power and discharging power constraints, energy storage energy balance constraints, and initial and final energy level constraints of the energy storage system.

[0113] (1) Energy storage capacity constraint: Since energy storage has a maximum energy storage capacity, it cannot store electricity indefinitely. The capacity rented out cannot exceed the maximum energy storage capacity.

[0114] The specific formula of the constraint is as follows:

[0115] E min -E rent <E d1 <E max -E rent

[0116]

[0117] Where E min Indicates the minimum energy level of energy storage, E max Indicates the maximum energy level of energy storage, E rent Indicates the capacity of the leased energy storage battery, E d1 Represents the electric energy at the energy storage time d1, and d1=1,2,3,...,τ.

[0118] The minimum energy level of energy storage means that when the remaining capacity of the energy storage after discharge drops to 5% of the rated capacity, further discharge will be stopped; the maximum energy level of energy storage means that when the capacity of the energy storage after charging rises to 95% of the rated capacity, further charging will be stopped.

[0119] (2) Charging power and discharging power constraints: Charging and discharging efficiency constraints require that the charging power and discharging power of energy storage must be within the allowed range. The specific formula is as follows:

[0120]

[0121]

[0122] Indicates the maximum discharge power of the energy storage power station, Indicates the maximum charging power of the energy storage power station.

[0123] (3) Energy storage energy balance constraints:

[0124]

[0125] E d1,power It indicates the load value obtained by the flow calculation in the hour d1.

[0126] E d1,predict It represents the load forecast value of the node in hour d1.

[0127] (4) Initial and final energy level constraints of the energy storage system:

[0128]

[0129] It means that after a period of time, the energy state of the energy storage battery pack should remain balanced, that is, the total amount of charge and discharge of the energy storage battery pack should be equal within a period of time.

[0130] 3) Solve the optimal profit model: Input the known parameters of the objective function, such as installed capacity, participation method, electricity price, predicted power load of each node and actual load calculated by flow, etc., and use the optimization method to solve the optimal combination of energy storage participation in battery leasing, electricity market trading, power auxiliary services, etc., as well as the optimal charging and discharging power, charging and discharging time period, and charging and discharging time, to maximize the economic income of the energy storage power station.

[0131] Solve using optimization methods: linear programming, nonlinear programming or heuristic algorithms (such as genetic algorithms, particle swarm algorithms, etc.) are used to solve the above objective functions and optimal solutions under the constraints. The solution results will include the optimal combination of various profit-making methods for energy storage power stations, the optimal charging and discharging power, charging and discharging time periods, and charging and discharging durations.

[0132] Output the best profit model: Based on the solution results, the best profit model of the energy storage power station is output, including the way to participate in electricity market activities, the charging and discharging power and charging and discharging time in each hour period, etc. This will provide operators of energy storage power stations with clear operating strategies and economic benefit maximization plans.

[0133] Through the above steps, we can establish a profit maximization model for energy storage power stations and output the best profit model to maximize the economic benefits of energy storage power stations. The above technical features constitute the best implementation of the present invention, which has strong adaptability and best implementation effect. Non-essential technical features can be added or subtracted according to actual needs to meet the needs of different situations.

[0134] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. A method for optimizing the power consumption of an energy storage power station based on graph computing, characterized in that: The steps include: Step S1, predicting the future power generation capacity and load demand of the energy storage power station based on graph calculation: obtaining the historical power generation data and load data of the distributed power stations in the grid-connected area of ​​the energy storage power station, as well as the external impact data, constructing the power grid topology structure, and constructing a spatiotemporal graph neural network model based on the power grid topology structure, and predicting the power generation capacity and load demand of each hour in the next 24 hours through the spatiotemporal graph neural network model; Step S2: Obtain the technical parameters and real-time operation data of the distribution network in the power grid topology structure, calculate the power and load values ​​of each node in the power grid through power flow analysis, and output the power and voltage of each node in the power grid as the power flow analysis result; and calculate the amount of electricity that the energy storage power station needs to provide or consume based on the predicted future power generation capacity and load demand.

2. The method for optimizing the power generation of an energy storage power station according to claim 1, characterized in that: Step S1 specifically includes the following steps: Step S11, data collection and preprocessing: obtaining historical power generation data of distributed power stations including solar and wind power stations in the energy storage power station grid-connected area, grid load data, holiday data, weather data, and equipment operation status data in the energy storage power station grid-connected area; preprocessing the collected data, including cleaning data, processing missing values ​​and outliers, and normalizing data, to obtain relevant time series data of the energy storage power station grid-connected area after preprocessing; Step S12, constructing a power grid topology graph structure: converting the time series data processed in step S11 into a graph structure, mapping the timestamp and attribute information in the time series data to the nodes and edges in the graph structure, defining nodes, edges and their attributes, where nodes represent power stations, substations, and load points, and edges represent paths for power flow, including transmission lines or distribution lines, and distribution line edges, to obtain a power grid topology graph structure; Step S13, using the power grid topology structure of step S12 as the input of the graph neural network model to construct a spatiotemporal graph neural network model; Step S14: Use the historical data of step S11 to train and verify the spatiotemporal graph neural network model, evaluate the performance of the spatiotemporal graph neural network model by comparing the difference between the prediction results of the spatiotemporal graph neural network model and the actual power grid state, and make adjustments and optimizations; Step S15, prediction and verification: Use the trained spatiotemporal graph neural network model to predict future power generation capacity and load demand, and evaluate the accuracy and generalization ability of the dynamic relationship graph model through cross-validation or holdout method.

3. The method for optimizing the power generation of an energy storage power station according to claim 2, characterized in that: In step S13, a spatiotemporal graph neural network model is constructed, including: The changing trend and spatial distribution characteristics of the power grid topology structure in the time series are calculated to obtain the embedded representation of the graph and input it into the encoder. The encoder then performs feature extraction and combines the attention mechanism to output a dynamic graph of the power grid status at a preset time in the future in the decoder, finally obtaining a spatiotemporal graph neural network model.

4. The method for optimizing the power generation of an energy storage power station according to claim 3, characterized in that: More specifically, step S13 includes: Firstly, the time series analysis technology is used to capture the changing trend of each node and its attributes in the power grid topology graph over time, and the power grid topology structure and time series data are converted into the embedded representation of the graph to obtain the embedded representation of the graph; Input the graph embedding representation into the encoder; Feature extraction combined with attention mechanism: The attention mechanism is introduced into the encoder to capture the key changes in the power grid. After being processed by the encoder, the extracted features are input into the decoder. The decoder outputs a dynamic graph of the power grid status within the preset hourly period in the future to obtain a spatiotemporal graph neural network model.

5. The method for optimizing the power generation of an energy storage power station according to claim 1, characterized in that: Specifically, step S2 includes: obtaining technical parameters and real-time operation data of the distribution network in the power grid topology structure, including the load and voltage of each node, inputting the power grid topology structure and real-time operation data into the power flow calculation tool, using the power flow calculation method to obtain the power flow analysis results of the distribution network, and outputting the power and voltage of each node in the power grid as the power flow analysis results.

6. The method for optimizing the power generation of an energy storage power station according to claim 5, characterized in that: Specifically, step S2 includes: Step S21: Obtain the power generation capacity and load demand for each hour of the next day predicted by the spatiotemporal graph neural network model, and obtain the power value of each intermediate node in the circuit through power flow calculation: The spatiotemporal graph neural network model is used to predict the power generation capacity and load demand for each hour of the next day. The predicted power generation capacity and load demand are used as inputs to the power flow calculation tool. The power flow calculation is performed again, and the power and voltage of each node in the power grid are output as the power flow analysis results, thereby calculating the power value of each intermediate node in the circuit. Step S22: Obtain the load forecast value of the intermediate node and the load value calculated based on the power flow analysis result, and calculate the amount of electricity that the energy storage power station needs to provide or consume: Based on the results of the power flow analysis, the actual load value of the intermediate node is calculated. By comparing the load prediction value of the intermediate node with the load value of the power flow analysis, the amount of electricity that the energy storage power station needs to provide or consume is calculated.

7. The method for optimizing the power generation of an energy storage power station according to claim 1, characterized in that: Also includes: Step S3: Based on the results of the power flow analysis, a profit maximization model is established to output the best profit model, including the way in which the power storage station participates in the power market, the charging power and discharging power in each hour period, and the charging time and discharging time; Step S3 specifically includes: Calculate the life cycle cost of energy storage, i.e. the levelized cost of storage (LCOS); Using the life cycle cost LCOS of energy storage, determine the objective function and constraints for maximizing the economic benefits of energy storage power stations; By using optimization methods to solve the optimal combination of energy storage participating in battery leasing, electricity market trading, and power auxiliary services, as well as the optimal charging and discharging power, charging and discharging time period, and charging and discharging duration, the economic income of the energy storage power station can be maximized.

8. The method for optimizing the power generation of an energy storage power station according to claim 7, characterized in that: The full life cycle cost of energy storage is calculated as the levelized cost of energy storage (LCOS). The specific calculation formula is as follows: Among them, d is the discharge time at rated power, η is the cycle efficiency of the energy storage station, T is the system life, n(t) is the number of cycles per year, r is the discount rate, Q&M(t) is the operation and maintenance cost in the tth year, C is the E Installation cost that varies with capacity, C P Installation costs vary with power.

9. The method for optimizing the power generation of an energy storage power station according to claim 8, characterized in that: The objective function of maximizing the economic benefits of the energy storage power station is expressed as: P is the price, represents the price of electricity sold in the wth electricity market activity mode during the d1th hour period, Represents the price of electricity purchased in the wth electricity market activity mode during the d1th hour period; S is power, It represents the discharge power under the w-th electricity market activity mode during the d1-th hour period, and the negative sign represents discharge; It represents the charging power under the w-th electricity market activity mode during the d1-th hour period, and the positive sign represents charging; It represents the discharge duration of participating in the wth electricity market activity mode in the d1th hour period, and the negative sign represents discharge; It represents the charging time of participating in the wth electricity market activity mode in the d1th hour period, and the positive sign represents charging; Income rent is the rental income for the τ period.

10. The method for optimizing the power consumption of an energy storage power station according to claim 9, characterized in that: The constraints of the objective function include energy storage capacity constraints, charging power and discharging power constraints, energy storage energy balance constraints, and initial and final energy level constraints of the energy storage system. The specific formula for energy storage capacity constraints is as follows: AND min -AND rent <E d1 <E max -AND rent Among them, E min Indicates the minimum energy level of energy storage, E max Indicates the maximum energy level of energy storage, E rent Indicates the capacity of the leased energy storage battery, E d1 Represents the electric energy at the energy storage time d1, and d1=1,2,3,...,τ.

Images