Electric power energy storage system and electric power energy storage method

Abstract

The invention discloses an electric power energy storage system and an electric power energy storage method, and particularly relates to the field of electric power system dispatching. Comprising an acquisition processing module, a demand prediction module, a power generation prediction module, an energy storage simulation module, a scheduling optimization module, a target adjustment module, an efficiency evaluation module, a dynamic control module, a decision support module, a risk evaluation module and a monitoring analysis module. Stable operation of a power system in a high-load period is ensured, the problem of power supply interruption or overload caused by improper energy storage strategies is reduced, and the long-term operation efficiency and stability of the power energy storage system are improved; according to the method, decisions in all aspects can be analyzed and optimized in a more detailed manner, deviation or errors caused by a single model are avoided, it is ensured that the energy storage system can make more accurate decisions under different conditions, the possible operation risk of the energy storage system is reduced, and the transparency and interpretability of the decisions are improved.

Description

Technical Field

[0001] The present invention relates to the field of power system dispatching, and in particular to an electric energy storage system and an electric energy storage method. Background Art

[0002] With the transformation of the global energy structure and the large-scale grid connection of renewable energy, the power system is facing increasingly severe challenges. In particular, renewable energy such as wind power and solar energy are volatile and intermittent, which affects the stability and reliability of power supply. In order to solve this problem, the power energy storage system, as an efficient means of regulation, can store excess electricity when power demand is low and release it when demand is peak, so as to balance supply and demand and stabilize the power grid. Through effective energy storage technology, not only can the reliability of the power grid be improved, but also the abandonment of wind and light of renewable energy can be reduced, and the utilization of green energy can be promoted. However, the efficient operation of the energy storage system depends on accurate scheduling strategies. In the traditional energy storage system scheduling, many methods rely on static strategies or simple optimization models, which are difficult to cope with complex dynamic grid load fluctuations and the uncertainty of renewable energy generation.

[0003] After searching, Chinese patent number CN111692070A discloses an electric energy storage system and an electric energy storage method. Although the invention can realize both pumped energy storage and compressed air energy storage, the energy storage is more flexible, and there is no need to rely on geographical conditions, and the choice of construction site is more flexible, but it cannot integrate multi-objective needs, reduce the system operation efficiency, and cannot ensure the stable operation of the power system during high-load periods. The risk of power supply interruption or overload problems caused by improper energy storage strategies is high; in addition, the existing electric energy storage system and electric energy storage method cannot analyze and optimize decisions in various aspects in detail, and are prone to deviations or errors. It is impossible to make more accurate decisions in different situations, which increases the operational risks that the energy storage system may face, and the decision transparency and explainability are low. For this reason, we propose an electric energy storage system and an electric energy storage method. Summary of the invention

[0004] The object of the present invention is to provide an electric energy storage system and an electric energy storage method to solve the above-mentioned problems.

[0005] The purpose of the present invention can be achieved through the following technical solutions: In a first aspect of the present invention, a power storage system is first proposed, the system comprising: Acquisition and processing module, demand forecasting module, power generation forecasting module, energy storage simulation module, scheduling optimization module, target adjustment module, efficiency evaluation module, dynamic control module, decision support module, risk assessment module and monitoring and analysis module; The acquisition and processing module is used to collect and pre-process various data such as grid load, battery status, renewable energy generation and weather forecast in real time; The demand forecasting module is used to forecast subsequent power demand based on historical load data and weather conditions; The power generation prediction module is used to predict power generation fluctuations of renewable energy based on meteorological data and historical power generation data; The energy storage simulation module is used to establish an energy storage device model and simulate the performance of the energy storage system under different scheduling conditions; The scheduling optimization module is used to optimize the charging and discharging strategy of the power storage system according to the power demand and renewable energy generation prediction results; The target adjustment module is used to dynamically adjust the dispatch decision of the power energy storage system; The efficiency evaluation module is used to analyze the real-time data of the power energy storage system, evaluate the charging and discharging efficiency, loss, life of the energy storage equipment and the overall operating cost of the system; The dynamic control module is used to monitor the operating status of the energy storage system in real time and adjust the power storage scheduling according to the actual situation; The decision support module is used to generate corresponding decision plans for reference by staff according to various uncertain factors of the power energy storage system; The risk assessment module is used to comprehensively analyze potential safety hazards of the power energy storage system, simulate, identify and warn of various emergencies; The monitoring and analysis module is used to continuously monitor the operating status of the energy storage system.

[0006] As a further solution of the present invention, the specific steps of the demand forecasting module for forecasting subsequent power demand are as follows: S1.1: The demand forecasting module receives the real-time power data, historical load data and weather data pre-processed by the acquisition and processing module, scales the historical load data and weather data to the interval [0, 1] through normalization, and then divides the processed data into training set, validation set and test set, and constructs a power forecasting model, which includes an input layer, a bidirectional GRU layer, a fully connected layer and an output layer; S1.2: According to the preset requirements, the training set is divided into multiple batches and input into the power forecasting model in sequence. The power forecasting model performs forward propagation on the training set data of each batch, and calculates the forward hidden layer state sequence and the reverse hidden layer state sequence respectively through the bidirectional GRU layer, and then combines the two sets of state sequences using weighted balance. After nonlinear processing through the fully connected layer, the combined state sequence is converted into a power demand forecast value, and then output through the output layer; S1.3: Calculate the gap between the actual power demand and the predicted power demand through the MSE function, perform back propagation on the calculated gap from the output layer of the power prediction model, calculate the gradient of the gap for each network layer of the power prediction model, and adjust the parameters of each network layer through the Adam optimizer. After each round of training, input the validation set into the power prediction model, calculate the MSE function and MAE function, and calculate the prediction ability of the power prediction model on unknown data; S1.4: If the evaluation result does not reach the preset threshold, the power prediction model is retrained and verified until the prediction ability of the model reaches the preset threshold. After that, each group of real-time power data processed by the acquisition and processing module is input into the trained power prediction model. The power prediction model performs forward propagation processing on each group of real-time power data and outputs the power demand forecast value within the preset time interval.

[0007] As a further solution of the present invention, the specific steps of the scheduling optimization module to optimize the charging and discharging strategy of the power energy storage system are as follows: S2.1: The dispatch optimization module receives the power demand forecast value, renewable energy generation forecast value, energy storage equipment status and time information, and constructs a strategy space based on the charging, discharging or non-operation operations that the power storage system can take at each time. Then, the reward function is defined based on the power supply balance of the power grid, the charging and discharging efficiency of the power storage system and the economy to evaluate the effect of each charging and discharging strategy; S2.2: Take the current charging and discharging strategy and state of the power storage system as the root node, and randomly select the selectable operations in the current strategy space, generate a new charging and discharging strategy and state of the power storage system, and use it as a child node, and connect it with the root node to construct a corresponding search tree. Starting from the root node, use the UCT selection strategy to select the node with the highest UCT value in each child node layer by layer; S2.3: When an incompletely expanded child node is selected, based on the optional operations of the child node, all child nodes of the node are expanded to represent different operations at each time step, and each expanded child node is randomly simulated until the preset simulation time of the search tree is reached. During the simulation, the reward value of each operation is calculated according to the reward function. After the simulation is completed, the reward value is fed back to all parent nodes on the path, and the reward value and the number of visits of each node are updated until the root node is reached; S2.4: Repeat multiple selections, expansions, simulations, and backtracking until the reward value changes converge to a preset range, stop iterating, traverse the final search tree, and select the charging and discharging strategy corresponding to the path with the maximum reward value as the charging and discharging behavior of the power energy storage system in the future period. Update the status of the power energy storage system according to the selected charging and discharging strategy.

[0008] As a further solution of the present invention, the specific calculation formula of the reward function in S2.1 is as follows: In the formula, Representative moments The return value at time Representative moments The power demand at that time; Representative moments Renewable energy generation at the time Representative moments Charging power at Representative moments Charging efficiency when Representative moments The discharge power at Representative moments Discharge efficiency when Representative moments The remaining battery capacity of the power storage system at that time; as well as Represents the weighting coefficient, which is used to balance the impact of different objectives.

[0009] As a further solution of the present invention, the specific steps of the target adjustment module dynamically adjusting the dispatching decision of the power energy storage system are as follows: S3.1: According to the current operation of the power energy storage system, minimize the economic cost, maximize the utilization of renewable energy, the stability and reliability of power supply, and minimize the wear of energy storage equipment as the optimization objectives. At the same time, set the energy storage capacity constraint, charge and discharge power constraint, and power demand balance constraint conditions based on the set optimization objectives, and then define the corresponding comprehensive objective function based on the optimization objectives; S3.2: Collect the charging and discharging strategies of the power storage system in each time period, and initialize a group of populations based on each group of charging and discharging strategies. The population size is the total number of charging and discharging strategies. Initialize the position and speed of each individual in the population, where the individual position represents its corresponding charging and discharging strategy, and initially use it as the individual optimal solution of each individual. , calculate the fitness value of the charging and discharging strategy corresponding to each individual position according to the comprehensive objective function value; S3.3: Compare the fitness values ​​of each group of individuals, and take the individual with the lowest fitness value as the global optimal solution , according to the global optimal solution The corresponding individual position is updated, the positions and speeds of the remaining individuals are updated, the fitness values ​​of each group of individuals are calculated, and the individual optimal solution is updated based on the updated fitness values ​​of each individual And the global optimal solution ; S3.4: Compare the updated individual fitness value with its current individual optimal solution , if the current individual fitness value is less than the individual optimal solution , then the individual optimal solution Replace it with the position of the current individual, otherwise it remains unchanged. After the individual optimal solution is updated, compare the fitness values ​​of each individual and reselect the individual position with the lowest fitness value as the global optimal solution , and then repeatedly update the individual position, speed, and individual optimal solution And the global optimal solution , until the global optimal solution of the population The change is less than the preset convergence threshold; S3.5: The population after the update is taken as the initial population, and the corresponding selection probability is calculated based on the fitness value of each individual in the initial population, and the individuals whose selection probability is higher than the preset threshold are extracted, and the selected groups of individuals are taken as parent individuals and replicated. At the same time, the charging and discharging strategies corresponding to each parent individual are regarded as a complete set of control sequences, and then the intersection points in the control sequences of each replicated parent individual are randomly selected for pairwise exchange to generate new individuals, and then a random disturbance value is randomly generated, and any control sequence of each replicated parent individual is adjusted based on the random disturbance value to generate new individuals; S3.6: Generate a new population through selection, crossover and mutation operations, and use the population for fitness evaluation of the next generation. Repeat the evolution and iteration of the population until the fitness value of the best individual in the population is less than the preset threshold. Then stop the iteration and output the charging and discharging strategy corresponding to the best individual as the optimal control strategy. At the same time, use the current population as the initial population for subsequent optimization.

[0010] As a further solution of the present invention, the specific calculation formula for updating the positions and speeds of the remaining individuals in S3.3 is as follows: In the formula, Representative individual In the In the dimension The speed of the moment; represents the inertia weight, which is used to control the influence of the individual's previous velocity; Representing an individual In the In the dimensions The speed of the moment; as well as represents the learning factor, which is used to control the degree of dependence on the individual optimal solution and the global optimal solution; as well as Represents a random number between the interval [0, 1]; Representing an individual In the The individual optimal solution in each dimension; Representing an individual In the The global optimal solution in dimensions; Representing an individual In the In the dimensions The location at the moment; Representing an individual In the In the dimensions The position at the moment.

[0011] In a second aspect of the present invention, a method for storing electric energy is provided, the method comprising the following steps: S101: Collect the load information of the power grid and the power generation of renewable energy in real time, and predict the fluctuation of power demand and renewable energy supply; S102: Establish a power storage system model to simulate its operating characteristics and evaluate the performance and optimization potential of the power storage system under different scenarios; S103: According to the simulation results, simulate different dispatching strategies and evaluate the results of each dispatching strategy, and select the required energy storage dispatching plan for execution; S104: simulating the operation of the entire electric energy storage system based on the selected energy storage scheduling scheme, and optimizing the configuration and scheduling strategy of the energy storage equipment; S105: Based on the operation data of the electric energy storage system, probabilistic modeling is performed on various uncertain factors of the energy storage system to generate an energy storage decision plan; S106: Collect real-time data of the power grid, energy storage equipment and renewable energy to analyze the operating status of the power energy storage system in real time, and dynamically adjust the scheduling plan using feedback control strategies.

[0012] As a further solution of the present invention, the specific steps of performing probability modeling on various uncertain factors of the energy storage system in S105 to generate an energy storage decision plan are as follows: S4.1: Identify and quantify grid loads from electric energy storage systems , renewable energy power generation and energy storage device performance Each group of uncertain factors, based on the collected information of each group of uncertain factors, calculates the prior distribution of grid load, renewable energy power generation and charging and discharging efficiency of energy storage equipment; S4.2: Collect real-time data of each power storage system and determine the control parameters of each power storage system according to the current control strategy , and by setting the likelihood function Calculate the given control parameters The next observed time Observational dataset The probability of Represents an observation dataset, containing the time The grid load, power generation and actual charging and discharging power at that time, Represents the total time, Represents the given control parameters Observed charge and discharge power The probability of S4.3: Based on the calculated prior probability and likelihood function results, calculate the probability of After any observation of data, control parameters The posterior distribution of is normalized by calculating the marginal likelihood, and then the control parameters are adjusted , recalculate the posterior distribution of each power storage system, compare the size of the posterior distribution after each adjustment, and select the control parameter with the largest posterior distribution value through MAP estimation ; S4.4: According to the selected control parameters As well as the parameter values ​​in the current control strategy, the corresponding adjustment strategy is generated, and the operating status of each power energy storage system after the parameter adjustment is simulated through the power energy storage simulation model. At the same time, the actual operating data after the strategy adjustment is collected, and the strategy is re-inferred and adjusted based on the new data.

[0013] As a further solution of the present invention, the specific calculation formula of the prior probability in S4.1 is as follows: In the formula, Represents the grid load The prior probability of Represents the standard deviation of grid load; The average value of the grid load; represents pi; represents the seasonal mean of renewable energy generation; represents amplitude; Represents cycle; represents the phase shift; represents the average power generation; represents the seasonal standard deviation of renewable energy generation; as well as Represent the base value and seasonal variation of the standard deviation respectively; Represents renewable energy generation The prior probability of represents the Beta function, which is used to normalize the probability density function; represents the charge and discharge efficiency, and ; Represents the performance of energy storage equipment The prior probability of as well as Represents the shape parameter of the distribution, which is used to control the shape of the distribution.

[0014] Beneficial effects of the present invention: 1. After the dispatch optimization module completes the selection of the optimal dispatch strategy, the present invention sets weights and optimization targets based on multiple target requirements such as economy, operation efficiency and system stability. Then, the power storage system performs a global search according to the optimization target, explores different operation strategy solutions, and iterates the optimization solution multiple times to achieve a balance between the targets. At the same time, according to the uncertainty factors of the charging and discharging efficiency, capacity limitation, grid load fluctuation and renewable energy power generation of the energy storage equipment, the charging and discharging decision is gradually adjusted. After multiple rounds of decision adjustment and evaluation, the optimal operation strategy is selected, which can integrate multiple target requirements, improve the system operation efficiency, ensure the stable operation of the power system during high load periods, reduce power supply interruptions or overload problems caused by improper energy storage strategies, and improve the long-term operation efficiency and stability of the power storage system.

[0015] 2. The present invention collects real-time data, estimates the possible range of changes and distribution of various uncertain factors of the energy storage system based on historical data and current status, and then comprehensively considers the impact of various scenarios on the operation strategy of the energy storage system. By converting these uncertain factors into probability distribution, a model that includes all potential change scenarios is constructed. After the model is built, the optimal decision-making plan under different uncertainty conditions is derived, which can analyze and optimize various aspects of decision-making in a more detailed manner, avoid deviations or errors caused by a single model, ensure that the energy storage system can make more accurate decisions in different situations, reduce the operational risks that the energy storage system may face, and improve the transparency and explainability of decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described below in conjunction with the accompanying drawings.

[0017] Figure 1 A system block diagram of an electric energy storage system provided by an embodiment of the present invention; Figure 2 A flowchart of an electric energy storage method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0020] The embodiment of the present invention provides an electric energy storage system. Figure 1 , Figure 1 A system block diagram of an electric energy storage system provided in an embodiment of the present invention. The system includes the following modules: Acquisition and processing module, demand forecasting module, power generation forecasting module, energy storage simulation module, scheduling optimization module, target adjustment module, efficiency evaluation module, dynamic control module, decision support module, risk assessment module and monitoring and analysis module.

[0021] The acquisition and processing module is used to collect and pre-process various data such as grid load, battery status, renewable energy generation and weather forecast in real time; the demand forecasting module is used to predict subsequent power demand based on historical load data and weather conditions.

[0022] Specifically, the demand forecasting module receives each set of real-time power data, historical load data and weather data pre-processed by the acquisition and processing module, and scales the historical load data and weather data to the interval [0, 1] through normalization processing. Then, each set of processed data is divided into a training set, a validation set and a test set, and a set of power forecasting models is constructed. The model includes an input layer, a bidirectional GRU layer, a fully connected layer and an output layer. According to preset requirements, the training set is divided into multiple batches and sequentially input into the power forecasting model. The power forecasting model forward propagates the training set data of each batch, and calculates the forward hidden layer state sequence and the reverse hidden layer state sequence respectively through the bidirectional GRU layer, and then combines the two sets of state sequences by weighted balance, and converts the combined state sequence into the power demand forecast value after nonlinear processing of the fully connected layer, and then outputs it through the output layer, calculates the gap between the actual power demand and the power demand forecast value through the MSE function, and performs back propagation from the output layer of the power forecasting model to calculate the gradient of the gap for each network layer of the power forecasting model, and adjusts the parameters of each network layer through the Adam optimizer. After each round of training, the verification set is input into the power forecasting model, and the MSE function and MAE function are calculated to calculate the prediction ability of the power forecasting model on unknown data. If the evaluation result does not reach the preset threshold, the power forecasting model is retrained and verified until the prediction ability of the model reaches the preset threshold. After that, each set of real-time power data processed by the acquisition and processing module is input into the trained power forecasting model, and the power forecasting model performs forward propagation processing on each set of real-time power data and outputs the power demand forecast value within the preset time interval.

[0023] The power generation prediction module is used to predict the power generation fluctuations of renewable energy based on meteorological data and historical power generation data; the energy storage simulation module is used to establish an energy storage device model and simulate the performance of the energy storage system under different scheduling conditions; the scheduling optimization module is used to optimize the charging and discharging strategy of the power energy storage system based on the power demand and renewable energy power generation forecast results.

[0024] The scheduling optimization module receives the power demand forecast value, renewable energy generation forecast value, energy storage equipment status and time information, and constructs a strategy space based on the charging, discharging or non-operation operations that the power storage system can take at each time. Then, based on the power supply balance of the power grid, the charging and discharging efficiency and economy of the power storage system, a reward function is defined to evaluate the effect of each charging and discharging strategy. The current charging and discharging strategy and status of the power storage system are used as the root node, and the selectable operations in the strategy space at the current moment are randomly selected to generate a new charging and discharging strategy and status of the power storage system. It is used as a child node and connected to the root node to construct a corresponding search tree. Starting from the root node, the UCT selection strategy is used to select the node with the highest UCT value in each child node layer by layer. When an incomplete operation is selected, After expanding the child node, based on the optional operation of the child node, all child nodes of the node are expanded to represent different operations at each time step, and each expanded child node is randomly simulated until the preset simulation time of the search tree is reached. During the simulation, the reward value of each operation is calculated according to the reward function. After the simulation is completed, the obtained reward value is fed back to all parent nodes on the path, and the reward value and the number of visits of each node are updated until the root node is reached. Repeat the selection, expansion, simulation and backtracking many times until the reward value change converges to the preset range, stop the iteration, traverse the final search tree, and select the charging and discharging strategy corresponding to the path with the maximum reward value as the charging and discharging behavior of the power energy storage system in the future period. According to the selected charging and discharging strategy, the state of the power energy storage system is updated.

[0025] It should be further explained that the specific calculation formula of the reward function is as follows: In the formula, Representative moments The return value at time Representative moments The power demand at that time; Representative moments Renewable energy generation at the time of Representative moments Charging power at Representative moments Charging efficiency when Representative moments The discharge power at Representative moments Discharge efficiency when Representative moments The remaining battery capacity of the power storage system at that time; as well as Represents the weighting coefficient, which is used to balance the impact of different objectives.

[0026] The target adjustment module is used to dynamically adjust the dispatching decisions of the power storage system.

[0027] Specifically, according to the current operation of the electric energy storage system, minimizing economic costs, maximizing the utilization of renewable energy, the stability and reliability of power supply, and minimizing the wear of energy storage equipment are selected as optimization goals. At the same time, energy storage capacity constraints, charging and discharging power constraints, and power demand balance constraints are set based on the set optimization goals. Then, the corresponding comprehensive objective function is defined based on the optimization goals, and the charging and discharging strategies of the electric energy storage system in each time period are collected. A group of populations is initialized based on each group of charging and discharging strategies, and the population size is the total number of charging and discharging strategies. The position and speed of each individual in the population are initialized, where the individual position represents its corresponding charging and discharging strategy, and it is initially used as the individual optimal solution of each individual. , calculate the fitness value of the charging and discharging strategy corresponding to each individual position according to the comprehensive objective function value, compare the fitness values ​​of each group of individuals, and take the individual with the lowest fitness value as the global optimal solution , according to the global optimal solution The corresponding individual position is updated, the positions and speeds of the remaining individuals are updated, the fitness values ​​of each group of individuals are calculated, and the individual optimal solution is updated based on the updated fitness values ​​of each individual And the global optimal solution , compare the updated individual fitness value with its current individual optimal solution , if the current individual fitness value is less than the individual optimal solution , then the individual optimal solution Replace it with the position of the current individual, otherwise it remains unchanged. After the individual optimal solution is updated, compare the fitness values ​​of each individual and reselect the individual position with the lowest fitness value as the global optimal solution , and then repeatedly update the individual position, speed, and individual optimal solution And the global optimal solution , until the global optimal solution of the population If the change is less than the preset convergence threshold, the population after the update is taken as the initial population, and the corresponding selection probability is calculated based on the fitness value of each individual in the initial population, and the individuals with a selection probability higher than the preset threshold are extracted, and the selected groups of individuals are taken as parent individuals and copied. At the same time, the charging and discharging strategies corresponding to each parent individual are regarded as a complete set of control sequences, and then the intersection points in the control sequences of the copied parent individuals are randomly selected for pairwise exchange to generate new individuals, and then a random disturbance value is randomly generated. Based on the random disturbance value, any control sequence of each copied parent individual is adjusted to generate new individuals. A new population is generated through selection, crossover and mutation operations, and the population is used for the fitness evaluation of the next generation. The population evolution and iteration are repeated until the fitness value of the optimal individual of the population is less than the preset threshold, and the iteration is stopped, and the charging and discharging strategy corresponding to the optimal individual is output as the optimal control strategy, and the current population is used as the initial population for subsequent optimization.

[0028] In addition, it should be noted that the specific calculation formula for updating the positions and speeds of the remaining individuals is as follows: In the formula, Representing an individual In the In the dimensions The speed of the moment; represents the inertia weight, which is used to control the influence of the individual's previous velocity; Representing an individual In the In the dimensions The speed of the moment; as well as represents the learning factor, which is used to control the degree of dependence on the individual optimal solution and the global optimal solution; as well as Represents a random number between the interval [0, 1]; Representing an individual In the The individual optimal solution in each dimension; Representing an individual In the The global optimal solution in dimensions; Representing an individual In the In the dimensions The location at the moment; Representing an individual In the In the dimensions The position at the moment.

[0029] The efficiency evaluation module is used to analyze the real-time data of the power energy storage system, evaluate the charging and discharging efficiency, loss, life of the energy storage equipment and the overall operating cost of the system; the dynamic control module is used to monitor the operating status of the energy storage system in real time and adjust the power storage scheduling according to actual conditions; the decision support module is used to generate corresponding decision plans for reference by staff based on the various uncertain factors of the power energy storage system; the risk assessment module is used to comprehensively analyze the potential safety hazards of the power energy storage system, simulate, identify and warn of various emergencies; the monitoring and analysis module is used to continuously monitor the operating status of the energy storage system.

[0030] Based on an electric energy storage system provided by an embodiment of the present invention, firstly, grid load data, renewable energy generation data, meteorological data, etc. are collected. Through data cleaning and preprocessing, noise is eliminated and missing values ​​are filled to ensure the accuracy and completeness of the data. Then, based on historical load data and weather information, future electricity demand is predicted. Combined with meteorological conditions, seasonal changes and other factors, the renewable energy generation in a certain period of time in the future is predicted to provide input for energy storage scheduling. Based on the prediction results of electricity demand and renewable energy generation, the charging and discharging strategy of the energy storage system is optimized. Then, on the basis of the optimization of the energy storage scheduling strategy, multi-objective optimization is performed to further improve the operating efficiency and economy of the energy storage system, and the performance of the energy storage system in different operating scenarios is simulated to evaluate the stability, flexibility and optimization potential of the system in actual operation. Then, the optimal operation decision is derived in combination with various uncertain factors of the system. Then, the updated operation strategy is executed to ensure that the charging and discharging of the energy storage equipment is carried out according to the optimal solution.

[0031] The embodiment of the present invention also provides a method for storing electric energy, such as Figure 2 As shown, the method comprises the following steps: Collect power grid load information and renewable energy generation in real time, and predict fluctuations in power demand and renewable energy supply.

[0032] Establish an electric energy storage system model to simulate its operating characteristics and evaluate the performance and optimization potential of the electric energy storage system under different scenarios.

[0033] Based on the simulation results, different dispatch strategies are simulated and the results of each dispatch strategy are evaluated, and the required energy storage dispatch scheme is selected for execution.

[0034] Based on the selected energy storage dispatching scheme, the operation of the entire power energy storage system is simulated to optimize the configuration and dispatching strategy of the energy storage equipment.

[0035] Based on the operating data of the power energy storage system, probability modeling is performed on various uncertain factors of the energy storage system to generate an energy storage decision-making plan.

[0036] Specifically, identify and quantify the grid load in the power storage system , renewable energy power generation and energy storage device performance Each group of uncertain factors, based on the collected information of each group of uncertain factors, calculates the prior distribution of grid load, renewable energy power generation and charging and discharging efficiency of energy storage equipment, collects real-time data of each power energy storage system, and determines the control parameters of each power energy storage system according to the current control strategy , and by setting the likelihood function Calculate the given control parameters The next observed time Observational dataset The probability of Represents an observation dataset, containing the time The grid load, power generation and actual charging and discharging power at that time, Represents the total time, Represents the given control parameters Observed charge and discharge power The probability of , based on the calculated prior probability and the likelihood function result, is calculated in a given observation data set After any observation of data, control parameters The posterior distribution of is normalized by calculating the marginal likelihood, and then the control parameters are adjusted , recalculate the posterior distribution of each power storage system, compare the size of the posterior distribution after each adjustment, and select the control parameter with the largest posterior distribution value through MAP estimation , depending on the selected control parameters As well as the parameter values ​​in the current control strategy, the corresponding adjustment strategy is generated, and the operating status of each power energy storage system after the parameter adjustment is simulated through the power energy storage simulation model. At the same time, the actual operating data after the strategy adjustment is collected, and the strategy is re-inferred and adjusted based on the new data.

[0037] It should be noted that the specific calculation formula of the prior probability is as follows: In the formula, Represents the grid load The prior probability of Represents the standard deviation of grid load; The average value of the grid load; represents pi; represents the seasonal mean of renewable energy generation; represents amplitude; Represents cycle; represents the phase shift; represents the average power generation; represents the seasonal standard deviation of renewable energy generation; as well as Represent the base value and seasonal variation of the standard deviation respectively; Represents renewable energy generation The prior probability of represents the Beta function, which is used to normalize the probability density function; represents the charge and discharge efficiency, and ; Represents the performance of energy storage equipment The prior probability of as well as Represents the shape parameter of the distribution, which is used to control the shape of the distribution.

[0038] Collect real-time data from the power grid, energy storage equipment and renewable energy to analyze the operating status of the power storage system in real time, and use feedback control strategies to dynamically adjust the scheduling plan.

[0039] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. An electric energy storage system, characterized in that: include: Acquisition and processing module, demand forecasting module, power generation forecasting module, energy storage simulation module, scheduling optimization module, target adjustment module, efficiency evaluation module, dynamic control module, decision support module, risk assessment module and monitoring and analysis module; The acquisition and processing module is used to collect and pre-process various data such as grid load, battery status, renewable energy generation and weather forecast in real time; The demand forecasting module is used to forecast subsequent power demand based on historical load data and weather conditions; The power generation prediction module is used to predict power generation fluctuations of renewable energy based on meteorological data and historical power generation data; The energy storage simulation module is used to establish an energy storage device model and simulate the performance of the energy storage system under different scheduling conditions; The scheduling optimization module is used to optimize the charging and discharging strategy of the power storage system according to the power demand and renewable energy generation prediction results; The target adjustment module is used to dynamically adjust the dispatch decision of the power energy storage system; The efficiency evaluation module is used to analyze the real-time data of the power energy storage system, evaluate the charging and discharging efficiency, loss, life of the energy storage equipment and the overall operating cost of the system; The dynamic control module is used to monitor the operating status of the energy storage system in real time and adjust the power storage scheduling according to the actual situation; The decision support module is used to generate corresponding decision plans for reference by staff according to various uncertain factors of the power energy storage system; The risk assessment module is used to comprehensively analyze potential safety hazards of the power energy storage system, simulate, identify and warn of various emergencies; The monitoring and analysis module is used to continuously monitor the operating status of the energy storage system.

2. An electric energy storage system according to claim 1, characterized in that: The specific steps of the demand forecasting module for forecasting subsequent power demand are as follows: S1.1: The demand forecasting module receives the real-time power data, historical load data and weather data pre-processed by the acquisition and processing module, scales the historical load data and weather data to the interval [0, 1] through normalization, and then divides the processed data into training set, validation set and test set, and constructs a power forecasting model, which includes an input layer, a bidirectional GRU layer, a fully connected layer and an output layer; S1.2: According to the preset requirements, the training set is divided into multiple batches and input into the power forecasting model in sequence. The power forecasting model performs forward propagation on the training set data of each batch, and calculates the forward hidden layer state sequence and the reverse hidden layer state sequence respectively through the bidirectional GRU layer, and then combines the two sets of state sequences using weighted balance. After nonlinear processing through the fully connected layer, the combined state sequence is converted into a power demand forecast value, and then output through the output layer; S1.3: Calculate the gap between the actual power demand and the predicted power demand through the MSE function, perform back propagation on the calculated gap from the output layer of the power prediction model, calculate the gradient of the gap for each network layer of the power prediction model, and adjust the parameters of each network layer through the Adam optimizer. After each round of training, input the validation set into the power prediction model, calculate the MSE function and MAE function, and calculate the prediction ability of the power prediction model on unknown data; S1.4: If the evaluation result does not reach the preset threshold, the power prediction model is retrained and verified until the prediction ability of the model reaches the preset threshold. After that, each group of real-time power data processed by the acquisition and processing module is input into the trained power prediction model. The power prediction model performs forward propagation processing on each group of real-time power data and outputs the power demand forecast value within the preset time interval.

3. An electric energy storage system according to claim 2, characterized in that: The specific steps of optimizing the charging and discharging strategy of the power storage system by the scheduling optimization module are as follows: S2.1: The dispatch optimization module receives the power demand forecast value, renewable energy generation forecast value, energy storage equipment status and time information, and constructs a strategy space based on the charging, discharging or non-operation operations that the power storage system can take at each time. Then, the reward function is defined based on the power supply balance of the power grid, the charging and discharging efficiency of the power storage system, and the economy to evaluate the effect of each charging and discharging strategy. The specific calculation formula of the reward function is as follows: In the formula, Representative moments The return value at time Representative moments The power demand at that time; Representative moments Renewable energy generation at the time of Representative moments Charging power at Representative moments Charging efficiency when Representative moments The discharge power at Representative moments Discharge efficiency when Representative moments The remaining battery capacity of the power storage system at that time; as well as Represents the weighting coefficient, which is used to balance the impact of different objectives; S2.2: Take the current charging and discharging strategy and state of the power storage system as the root node, and randomly select the selectable operations in the current strategy space, generate a new charging and discharging strategy and state of the power storage system, and use it as a child node, and connect it with the root node to construct a corresponding search tree. Starting from the root node, use the UCT selection strategy to select the node with the highest UCT value in each child node layer by layer; S2.3: When an incompletely expanded child node is selected, based on the optional operations of the child node, all child nodes of the node are expanded to represent different operations at each time step, and each expanded child node is randomly simulated until the preset simulation time of the search tree is reached. During the simulation, the reward value of each operation is calculated according to the reward function. After the simulation is completed, the reward value is fed back to all parent nodes on the path, and the reward value and the number of visits of each node are updated until the root node is reached; S2.4: Repeat multiple selections, expansions, simulations, and backtracking until the reward value changes converge to a preset range, stop iterating, traverse the final search tree, and select the charging and discharging strategy corresponding to the path with the maximum reward value as the charging and discharging behavior of the power energy storage system in the future period. Update the status of the power energy storage system according to the selected charging and discharging strategy.

4. The power storage system according to claim 3, characterized in that: The specific steps of dynamically adjusting the dispatching decision of the power energy storage system by the target adjustment module are as follows: S3.1: According to the current operation of the power energy storage system, minimize the economic cost, maximize the utilization of renewable energy, the stability and reliability of power supply, and minimize the wear of energy storage equipment as the optimization objectives. At the same time, set the energy storage capacity constraint, charge and discharge power constraint, and power demand balance constraint conditions based on the set optimization objectives, and then define the corresponding comprehensive objective function based on the optimization objectives; S3.2: Collect the charging and discharging strategies of the power storage system in each time period, and initialize a group of populations based on each group of charging and discharging strategies. The population size is the total number of charging and discharging strategies. Initialize the position and speed of each individual in the population, where the individual position represents its corresponding charging and discharging strategy, and initially use it as the individual optimal solution of each individual. , calculate the fitness value of the charging and discharging strategy corresponding to each individual position according to the comprehensive objective function value; S3.3: Compare the fitness values ​​of each group of individuals, and take the individual with the lowest fitness value as the global optimal solution , according to the global optimal solution The corresponding individual position is updated, the positions and speeds of the remaining individuals are updated, the fitness values ​​of each group of individuals are calculated, and the individual optimal solution is updated based on the updated fitness values ​​of each individual And the global optimal solution , where the specific calculation formula for updating the positions and speeds of the remaining individuals is as follows: In the formula, Representing an individual In the In the dimensions The speed of the moment; represents the inertia weight, which is used to control the influence of the individual's previous velocity; Representing an individual In the In the dimensions The speed of the moment; as well as represents the learning factor, which is used to control the degree of dependence on the individual optimal solution and the global optimal solution; as well as Represents a random number between the interval [0, 1]; Representing an individual In the The individual optimal solution in each dimension; Representing an individual In the The global optimal solution in dimensions; Representing an individual In the In the dimensions The location at the moment; Representing an individual In the In the dimensions The location at the moment; S3.4: Compare the updated individual fitness value with its current individual optimal solution , if the current individual fitness value is less than the individual optimal solution , then the individual optimal solution Replace it with the position of the current individual, otherwise it remains unchanged. After the individual optimal solution is updated, compare the fitness values ​​of each individual and reselect the individual position with the lowest fitness value as the global optimal solution , and then repeatedly update the individual position, speed, and individual optimal solution And the global optimal solution , until the global optimal solution of the population The change is less than the preset convergence threshold; S3.5: The population after the update is taken as the initial population, and the corresponding selection probability is calculated based on the fitness value of each individual in the initial population, and the individuals whose selection probability is higher than the preset threshold are extracted, and the selected groups of individuals are taken as parent individuals and replicated. At the same time, the charging and discharging strategies corresponding to each parent individual are regarded as a complete set of control sequences, and then the intersection points in the control sequences of each replicated parent individual are randomly selected for pairwise exchange to generate new individuals, and then a random disturbance value is randomly generated, and any control sequence of each replicated parent individual is adjusted based on the random disturbance value to generate new individuals; S3.6: Generate a new population through selection, crossover and mutation operations, and use the population for fitness evaluation of the next generation. Repeat the evolution and iteration of the population until the fitness value of the best individual in the population is less than the preset threshold. Then stop the iteration and output the charging and discharging strategy corresponding to the best individual as the optimal control strategy. At the same time, use the current population as the initial population for subsequent optimization.

5. An electric energy storage method, used to implement the function of an electric energy storage system as claimed in any one of claims 1 to 4, characterized in that: The following steps are involved: S101: Collect the load information of the power grid and the power generation of renewable energy in real time, and predict the fluctuation of power demand and renewable energy supply; S102: Establish a power storage system model to simulate its operating characteristics and evaluate the performance and optimization potential of the power storage system under different scenarios; S103: According to the simulation results, simulate different dispatching strategies and evaluate the results of each dispatching strategy, and select the required energy storage dispatching plan for execution; S104: simulating the operation of the entire electric energy storage system based on the selected energy storage scheduling scheme, and optimizing the configuration and scheduling strategy of the energy storage equipment; S105: Based on the operation data of the electric energy storage system, probabilistic modeling is performed on various uncertain factors of the energy storage system to generate an energy storage decision plan; S106: Collect real-time data of the power grid, energy storage equipment and renewable energy to analyze the operating status of the power energy storage system in real time, and dynamically adjust the scheduling plan using feedback control strategies.

6. A method for storing electric energy according to claim 5, characterized in that: The specific steps of performing probability modeling on various uncertain factors of the energy storage system to generate an energy storage decision plan as described in S105 are as follows: S4.1: Identify and quantify grid loads from electric energy storage systems , renewable energy power generation and energy storage device performance Each group of uncertain factors, based on the collected information of each group of uncertain factors, calculates the prior distribution of grid load, renewable energy power generation and charging and discharging efficiency of energy storage equipment; S4.2: Collect real-time data of each power storage system and determine the control parameters of each power storage system according to the current control strategy , and by setting the likelihood function Calculate the given control parameters The next observed time Observational dataset The probability of Represents an observation dataset, containing the time The grid load, power generation and actual charging and discharging power at that time, Represents the total time, Represents the given control parameters Observed charge and discharge power The probability of S4.3: Based on the calculated prior probability and likelihood function results, calculate the probability of After any observation of data, control parameters The posterior distribution of is normalized by calculating the marginal likelihood, and then the control parameters are adjusted , recalculate the posterior distribution of each power storage system, compare the size of the posterior distribution after each adjustment, and select the control parameter with the largest posterior distribution value through MAP estimation ; S4.4: According to the selected control parameters As well as the parameter values ​​in the current control strategy, the corresponding adjustment strategy is generated, and the operating status of each power energy storage system after the parameter adjustment is simulated through the power energy storage simulation model. At the same time, the actual operating data after the strategy adjustment is collected, and the strategy is re-inferred and adjusted based on the new data.

7. The method for storing electric energy according to claim 6, characterized in that: The specific calculation formula for the prior probability described in S4.1 is as follows: In the formula, Represents the grid load The prior probability of Represents the standard deviation of grid load; The average value of the grid load; represents pi; represents the seasonal mean of renewable energy generation; represents amplitude; Represents cycle; represents the phase shift; represents the average power generation; represents the seasonal standard deviation of renewable energy generation; as well as Represent the base value and seasonal variation of the standard deviation respectively; Represents renewable energy generation The prior probability of represents the Beta function, which is used to normalize the probability density function; represents the charge and discharge efficiency, and ; Represents the performance of energy storage equipment The prior probability of as well as Represents the shape parameter of the distribution, which is used to control the shape of the distribution.

Images