Scheduling optimization method for gravity energy storage system driven by load balance

Abstract

The invention provides a load balance driven gravity energy storage system scheduling optimization method, and relates to the technical field of data processing, and the method comprises the steps: carrying out the collection of meteorological variables, and obtaining meteorological driving data; generating capacity prediction is carried out to obtain predicted generating capacity; carrying out load demand analysis to obtain a predicted load demand; carrying out load balance deviation calculation to obtain power grid supply and demand deviation; extracting energy storage configuration parameters, performing energy storage scheduling analysis, and outputting a gravity energy storage scheduling strategy; in the charging and discharging scheduling process, the operation states of the multiple gravity energy storage devices are monitored in real time, and strategy dynamic optimization is carried out according to the monitoring result until the termination time of the meteorological time window is reached. According to the method, the technical problem that the running state of the energy storage system cannot be effectively adjusted and the reliability of power supply is further influenced due to the fact that an intelligent power grid in the prior art often depends on a static scheduling rule and lacks flexibility and dynamic adjustment capability when facing power grid requirements and meteorological conditions which change in real time is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a scheduling optimization method for a gravity energy storage system driven by load balancing. Background Art

[0002] The emergence of the smart grid provides solutions for the intelligence and automation of the power system. It can optimize the management, scheduling, and monitoring of the power grid by combining high-efficiency information and communication technologies with the traditional power grid, improving the reliability and efficiency of the power grid. However, with the increasing proportion of renewable energy, how to ensure the balance between power supply and demand in the power grid, especially in the case of large fluctuations in renewable energy generation, remains the main technical problem faced by the smart grid.

[0003] To alleviate the impact of renewable energy volatility on the power grid, energy storage technologies have been widely used. Energy storage devices can store excess power during power supply surpluses and release power during power supply shortages, thus smoothing the supply-demand fluctuations and enhancing the stability of the power grid. However, how to efficiently schedule energy storage devices, especially the coordinated scheduling of multiple devices, to maximize energy utilization efficiency and reduce losses, is the core problem under the current technology. Summary of the Invention

[0004] This application aims to solve the technical problem that the existing smart grid often relies on static scheduling rules and lacks flexibility and dynamic adjustment capabilities when facing real-time changing grid demands and meteorological conditions, resulting in the inability to effectively adjust the operating state of the energy storage system and further affecting the reliability of power supply.

[0005] The scheduling optimization method for a gravity energy storage system driven by load balancing disclosed in this application includes: interacting with a weather API to collect meteorological variables and obtain meteorological driving data, where the meteorological driving data has a meteorological time window identifier; predicting the power generation capacity based on the meteorological driving data to obtain the predicted power generation; taking the meteorological time window as a constraint, analyzing the load demand based on the meteorological driving data to obtain the predicted load demand; calculating the load balance deviation based on the predicted load demand and the predicted power generation to obtain the power grid supply-demand deviation; extracting the energy storage configuration parameters from the gravity energy storage system and performing energy storage scheduling analysis based on the energy storage configuration parameters and the power grid supply-demand deviation to output the gravity energy storage scheduling strategy; during the charge and discharge scheduling of multiple gravity energy storage devices in the gravity energy storage system according to the gravity energy storage scheduling strategy, real-time monitoring the operating states of the multiple gravity energy storage devices and dynamically optimizing the gravity energy storage scheduling strategy according to the monitoring results until the end time of the meteorological time window.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: Collect meteorological variables through an interactive meteorological API to obtain meteorological driving data, providing accurate meteorological environment information for subsequent steps. These information are indispensable bases for predicting power demand and power generation, ensuring that the prediction model reflects current and expected meteorological conditions, reducing possible errors in traditional prediction methods, and improving the response ability and reliability of the smart grid; Use meteorological driving data to predict power generation capacity, especially for energy forms such as wind power and solar energy that are highly dependent on meteorological conditions, which can significantly improve the accuracy of power generation prediction, which is of great significance for balancing grid supply and demand and optimizing power generation resource allocation; By analyzing the relationship between meteorological driving data and electricity demand, accurately predict the electricity demand within a specific future time window, thereby optimizing the scheduling and management of power resources; By calculating the load balance deviation based on the predicted load demand and predicted power generation, obtain the grid supply and demand deviation, and identify possible power surpluses or shortages. This precise prediction and calculation can ensure that the smart grid can make scheduling decisions in a timely manner when dealing with changing electricity demands, ensuring the supply and demand balance of the grid, reducing power surpluses or shortages, and thus improving the stability and operation efficiency of the grid; By combining the specific configuration of the gravity energy storage system and the actual grid supply and demand deviation, optimize the charge and discharge strategy of the energy storage system, providing a basis for the efficient operation of the smart grid, enabling the grid to adjust the energy storage system in a timely manner according to load fluctuations and renewable energy generation fluctuations, thereby reducing energy waste and improving the grid's response speed and load balancing ability; During the charge and discharge scheduling process, by real-time monitoring the state of the energy storage device and dynamically optimizing the charge and discharge strategy according to the monitoring results, support the energy storage system to always operate in the best state, which not only improves the response speed and flexibility of the gravity energy storage system, but also enhances the load balancing ability of the smart grid when facing power fluctuations caused by renewable energy.

[0007] The above description is only an overview of the technical solutions of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the following specific embodiments of this application are specifically given. Brief Description of the Drawings

[0008] Figure 1 This is a schematic flow chart of the scheduling optimization method for a load balance-driven gravity energy storage system provided by an embodiment of this application.

[0009] Figure 2 This is a schematic flow chart of the energy storage scheduling analysis in the scheduling optimization method for a load balance-driven gravity energy storage system provided by an embodiment of this application. Detailed Embodiments

[0010] By providing a scheduling optimization method for a gravity energy storage system driven by load balancing in the embodiments of the present application, the technical problem in the prior art that intelligent power grids often rely on static scheduling rules and lack flexibility and dynamic adjustment capabilities in the face of real-time changing grid demands and meteorological conditions, resulting in the inability to effectively adjust the operating state of the energy storage system and thus affecting the reliability of power supply is solved.

[0011] After introducing the basic principle of the present application, various non-limiting implementation manners of the present application will be specifically introduced below in conjunction with the drawings of the specification.

[0012] As Figure 1 shown, the embodiments of the present application provide a scheduling optimization method for a gravity energy storage system driven by load balancing, and the method includes: Interact with the meteorological API to collect meteorological variables and obtain meteorological driving data, wherein the meteorological driving data has a meteorological time window identifier.

[0013] Interact with the meteorological API to obtain meteorological driving data by requesting meteorological variables. The meteorological driving data includes temperature, humidity, wind speed, wind direction, air pressure, cloud cover, etc., which are all key factors affecting the power generation capacity of wind power and solar power. The obtained meteorological driving data has a meteorological time window identifier. The meteorological time window is a future time period with a fixed duration. For example, if the load and power generation capacity for the next 24 hours are predicted, the time window is 24 hours.

[0014] Predict the power generation capacity based on the meteorological driving data to obtain the predicted power generation.

[0015] Predict the power generation capacity of wind power and solar power based on the meteorological driving data. Different meteorological variables affect different types of power generation methods. For example, meteorological factors such as wind speed and wind direction directly affect the power generation capacity of wind turbines. Within a certain range of wind speed, the power generation efficiency of wind turbines is the highest; light intensity and temperature have a greater impact on the power generation efficiency of solar panels. Specifically, based on the time series data of meteorological variables, train an LSTM (Long Short-Term Memory) model to establish the relationship between the wind power and solar power generation and meteorological variables. The LSTM model is especially suitable for processing time series data because it can remember the impact of past meteorological data on future power generation. Input the real-time meteorological data into the trained model to obtain the predicted power generation for the future time window, and these predicted values are used for subsequent load demand analysis and energy storage scheduling decisions.

[0016] Taking the meteorological time window as a constraint, perform load demand analysis based on the meteorological driving data to obtain the predicted load demand.

[0017] The meteorological time window is the future time period for power generation prediction. Constrained by the meteorological time window, that is, keeping the time period of load demand prediction the same as that of power generation prediction to ensure their temporal correspondence. Based on meteorological driving data, power consumption is predicted using meteorological variables such as temperature, humidity, and wind speed. For example, higher or lower temperatures will increase the power consumption of air conditioning or heating systems. Similar to power generation capacity prediction, based on the time series data of meteorological variables, and at the same time, considering the impact of time characteristics such as weekends and weekdays, day and night on power consumption, an LSTM (Long Short-Term Memory Network) model is trained to establish the relationship between power consumption and meteorological variables. The real-time meteorological data is input into the trained model to obtain the predicted power consumption for the future time window, which is used as the predicted load demand.

[0018] Calculate the load balance deviation based on the predicted load demand and predicted power generation to obtain the power grid supply-demand deviation.

[0019] Compare the predicted power generation with the predicted load demand and calculate the deviation between the two. Specifically, the difference between the two can be directly calculated. For the obtained power grid supply-demand deviation, a positive deviation indicates that the power supply exceeds the demand, and a negative deviation indicates that the demand exceeds the power supply. The acquisition of the power grid supply-demand deviation provides basic data for further energy management and scheduling decisions.

[0020] Extract the energy storage configuration parameters from the gravity energy storage system, and perform energy storage scheduling analysis based on the energy storage configuration parameters and the power grid supply-demand deviation, and output the gravity energy storage scheduling strategy.

[0021] Extract the energy storage configuration parameters from the gravity energy storage system. The energy storage configuration parameters include multiple sets of device configuration parameters of multiple gravity energy storage devices in the gravity energy storage system. Each set of device configuration parameters includes energy storage capacity limit, charge-discharge rate, charge-discharge efficiency, remaining energy storage capacity, and device operating status. These parameters determine the performance and operation limits of the energy storage devices.

[0022] Determine the charging or discharging of the energy storage device according to the power grid supply-demand deviation. For example, if the power generation exceeds the demand, the excess power can be used to charge the energy storage device; if the demand exceeds the power generation, it is necessary to discharge the energy storage device to meet the demand. Combining the configuration parameters of the energy storage device, an optimization algorithm is used to formulate the charging and discharging strategies, including determining which energy storage devices need to be charged, which need to be discharged, and the timing and intensity of related operations. The optimization goal is to minimize the power grid supply-demand deviation, while making the charge-discharge operation of the energy storage system more efficient, reducing losses, and extending the device life. According to the energy storage strategy of each energy storage device, it is integrated into the overall gravity energy storage scheduling strategy of the gravity energy storage system.

[0023] During the charge and discharge scheduling of multiple gravity energy storage devices in the gravity energy storage system according to the gravity energy storage scheduling strategy, the operating states of the multiple gravity energy storage devices are monitored in real time, and the gravity energy storage scheduling strategy is dynamically optimized according to the monitoring results until the termination time of the meteorological time window.

[0024] Apply the energy storage strategy of each device in the gravity energy storage scheduling strategy to the corresponding gravity energy storage device to perform the charge and discharge scheduling of multiple gravity energy storage devices. During the scheduling process, the operating states of each energy storage device, such as the current charge and discharge state, battery health, energy efficiency, temperature, etc., are monitored in real time through sensors and the monitoring system.

[0025] According to the real-time monitoring data, evaluate the effect of the current scheduling strategy. If there is a deviation between the actual operating state and the expectation, adjust the scheduling strategy to adapt to the change. For example, if the charging efficiency of a certain device is lower than expected, it may be necessary to reduce the charging amount of the device or adjust the charging time to avoid overloading or damage. The monitoring and optimization process continues until the end of the current meteorological time window, which ensures that the entire process operates in an optimal state to cope with the changing environment and system state.

[0026] By monitoring the device state in real time during actual operation and dynamically adjusting the strategy according to real-time feedback, the system efficiency is maximized and the device safety is ensured, especially under changing meteorological conditions. This not only improves the energy utilization efficiency but also extends the service life of the device and guarantees the reliability and stability of the system.

[0027] Furthermore, as Figure 2 shown, extract the energy storage configuration parameters from the gravity energy storage system, and perform energy storage scheduling analysis according to the energy storage configuration parameters and the power grid supply-demand deviation, and output the gravity energy storage scheduling strategy. The method includes: Analyze the energy storage configuration parameters to obtain multiple sets of device configuration parameters of the multiple gravity energy storage devices; when the deviation vector of the power grid supply-demand deviation is positive, use the deviation vector of the power grid supply-demand deviation to traverse the multiple sets of device configuration parameters to screen N gravity energy storage devices from the multiple gravity energy storage devices; with the power grid supply-demand deviation as a constraint, perform the charging amount configuration of the N gravity energy storage devices to obtain N initial charging amounts; perform a scheduling fitness evaluation on the N initial charging amounts, and optimize the charging amounts according to the evaluation results to generate N device energy storage scheduling strategies; the N device energy storage scheduling strategies constitute the gravity energy storage scheduling strategy.

[0028] Extract multiple sets of device configuration parameters of multiple gravity energy storage devices from the management software of the gravity energy storage system. Each set of device configuration parameters includes energy storage capacity limit, charge and discharge rate, charge and discharge efficiency, remaining energy storage capacity, and device operating status. Among them, the energy storage capacity limit refers to the maximum amount of electricity that the device can store; the charge and discharge rate refers to the amount of electricity that the device can charge or discharge per hour; the charge and discharge efficiency represents the energy loss rate of the device during the charge and discharge process; the remaining energy storage capacity represents the amount of electricity that can still be stored in the current device in addition to the stored electricity; the device operating status involves whether the device is available or in a maintenance state, etc.

[0029] When the deviation vector of the power grid supply-demand deviation is positive, it means that the power supply of the power grid is greater than the demand. At this time, energy storage devices need to be selected for power storage to avoid energy waste. In this case, according to the deviation vector of the power grid supply-demand deviation, that is, the excess power supply, evaluate the number and capacity of the required energy storage devices. Specifically, traverse multiple sets of device configuration parameters in the gravity energy storage system, select devices that can meet the current energy storage demand, and the selection criteria include but are not limited to the remaining energy storage capacity, charging efficiency, and operating status of the device. Calculate the theoretical storage capacity of each device, compare it with the amount of electricity that needs to be stored currently, and select the most suitable device combination. Priority is given to devices with high efficiency and sufficient capacity. After screening, N gravity energy storage devices are obtained, that is, gravity energy storage devices that can perform charge and discharge operations.

[0030] For the selected N gravity energy storage devices, determine the initial charge of each device according to the power grid supply-demand deviation. The initial charge configuration can be simply set according to the maximum amount of electricity that each device can safely charge. For example, if the power grid supply-demand deviation is 1000 kWh and 5 devices are selected, then this 1000 kWh needs to be allocated to these 5 devices. The initial configuration can be simply allocated according to the maximum remaining capacity of each device. After allocation, N initial charges are obtained as the starting point for subsequent optimization.

[0031] Conduct a scheduling fitness evaluation on the initial charge of each device, including evaluating the load balance, life impact, and energy efficiency of the device, etc., so as to quantify the charging strategy effect of each device. According to the evaluation results, optimize the charge allocation. The optimization goal is to minimize the power grid supply-demand deviation, and at the same time make the charge and discharge operations of the energy storage system more efficient, reduce losses, and extend the device life. For example, if the efficiency of a certain device drops at high load, its charge will be reduced, and the charge of other devices with higher efficiency or lower load will be increased. After optimization, N device energy storage scheduling strategies are generated, including the charging start time, duration, and intensity of each device, to ensure the efficient operation of the energy storage system.

[0032] Integrate the obtained N device energy storage scheduling strategies to form a gravity energy storage scheduling strategy.

[0033] In another case, when the deviation vector of the power grid supply-demand deviation is negative, it indicates that the power supply of the power grid is less than the demand. At this time, it is necessary to select energy storage devices to discharge to avoid insufficient energy supply. Similar steps can be used to generate relevant scheduling strategies for the discharge management of gravity energy storage devices. The implementation process of this case is similar to the previous case. For the sake of simplicity of the specification, it will not be elaborated here.

[0034] Furthermore, among the multiple sets of device configuration parameters, each set of device configuration parameters includes energy storage capacity limit, charge and discharge rate, charge and discharge efficiency, remaining energy storage capacity, and device operation status.

[0035] Each set of device configuration parameters includes energy storage capacity limit, charge and discharge rate, charge and discharge efficiency, remaining energy storage capacity, and device operation status. Among them, the energy storage capacity limit refers to the maximum amount of electricity that the device can store, and the capacity limit determines how much electrical energy the device can store in a charging cycle; the charge and discharge rate refers to how fast the device can charge and discharge, which determines how quickly the device responds to load changes; the charge and discharge efficiency describes the energy loss during the charging and discharging processes, expressed as a percentage. For example, if the charging efficiency is 90%, it means that for every 100 kWh of electrical energy charged, only 90 kWh is actually stored in the device, and the remaining 10 kWh is lost during the conversion process. The charge and discharge efficiency affects the energy utilization rate and operating costs; the remaining energy storage capacity refers to how much electricity is still available in the device at any given time point; the device operation status includes but is not limited to normal operation, maintenance, failure, etc. states. Only when the device is in the normal operation state can it work properly at the predetermined efficiency and capacity.

[0036] Furthermore, perform a scheduling fitness evaluation on the N initial charging amounts, and optimize the charging amounts according to the evaluation results to generate N device energy storage scheduling strategies. The method includes: Calculate the predicted power generation rate based on the predicted power generation and the meteorological time window; call the N charge-discharge rates, N charge-discharge efficiencies, and N remaining energy storage capacities of the N gravity energy storage devices from the multiple sets of device configuration parameters; evaluate the energy consumption of the N initial charge amounts according to the N charge-discharge efficiencies to obtain a first energy consumption coefficient; update the data of the N initial charge amounts with the N remaining energy storage capacities as constraints according to the deviation between the predetermined energy consumption constraint and the first energy consumption coefficient to obtain N updated charge amounts; evaluate the energy consumption of the N updated charge amounts according to the N charge-discharge efficiencies to obtain a second energy consumption coefficient; and so on. According to the deviation between the energy consumption coefficient and the energy consumption constraint, adjust the charge amounts of the N gravity energy storage devices within the capacity range of the N remaining energy storage capacities until N target charge amounts that meet the energy consumption constraint are output; sort the scheduling priorities of the N gravity energy storage devices according to the predicted power generation rate and the N charge-discharge rates, and calculate the scheduling time according to the priority sorting result and the N target charge amounts to obtain N scheduling time intervals; the N scheduling time intervals and the N target charge amounts constitute the N device energy storage scheduling strategies.

[0037] Calculate the predicted power generation rate, which refers to the speed at which the power generation device generates electrical energy per unit time within the meteorological time window, and this rate is calculated based on the ratio of the predicted power generation and the meteorological time window.

[0038] Query and retrieve the corresponding parameters in the multiple sets of device configuration parameters with the query conditions of the N gravity energy storage devices to obtain the N charge-discharge rates, N charge-discharge efficiencies, and N remaining energy storage capacities of the N gravity energy storage devices, where the charge-discharge efficiency is the percentage of energy loss in the charging and discharging processes of the energy storage device.

[0039] Use the N charge-discharge efficiencies to evaluate the energy loss of each device at the initial charge amount. For example, if a device has a charge-discharge efficiency of 95% and plans to charge 100 kWh, then the actual amount of electricity stored in the battery is 95 kWh, with a loss of 5 kWh. The calculated first energy consumption coefficient is the average energy consumption coefficient of all selected devices, which is a quantitative indicator representing the expected percentage of energy loss during the charging process.

[0040] The predetermined energy consumption constraint is set as the energy efficiency target for the overall system. For example, it is set that the energy loss of the overall system does not exceed a certain percentage. Calculate the deviation between the first energy consumption coefficient and this predetermined energy consumption constraint, and determine whether it is necessary to adjust the initial charge amount. If the actual energy consumption is higher than the predetermined constraint, the charge amount needs to be reduced to lower the overall energy loss. Specifically, according to the limitations of the remaining energy storage capacities of N devices, adjust the N initial charge amounts to ensure that any adjustment will not cause the energy storage capacity of the device to be exceeded. Reallocate the charge amounts through an optimization algorithm to achieve a better energy efficiency target while meeting the capacity limitations of the devices, and obtain N updated charge amounts after adjustment.

[0041] Similarly, use the charge-discharge efficiencies of N devices to evaluate the energy loss of each device under the updated charge amounts, so as to evaluate the energy efficiency performance after adjusting the charge amounts. After evaluation, obtain the second energy consumption coefficient. The second energy consumption coefficient is calculated based on the updated charge amounts and may be different from the first energy consumption coefficient, reflecting the efficiency change of the adjusted charging strategy.

[0042] Based on the second energy consumption coefficient, check whether the predetermined energy consumption constraint is met. If not, adjust the charge amounts again, also with the remaining energy storage capacities of each device as the limitation, to ensure that the adjusted charge amounts will not exceed the maximum available capacity of the devices. Keep adjusting until a charge amount configuration that meets the predetermined energy consumption constraint is found, and generate N target charge amounts for preparation for actual operation.

[0043] According to the predicted power generation rate and the charge-discharge rates of each device, determine the scheduling priorities of the devices. The scheduling priorities of the devices can be set based on the high or low of their charge-discharge rates and the matching degree with the predicted power generation rate. For example, the devices with a higher matching degree can obtain higher priorities. Use the determined priorities and the target charge amounts to calculate the scheduling time intervals of each device, that is, determine the charging start time and duration of each device. The scheduling time intervals need to be coordinated among different devices to avoid power congestion during peak periods and maximize the energy utilization during valley periods.

[0044] Integrate the N scheduling time intervals and the N target charge amounts to form the energy storage scheduling strategies for N devices.

[0045] Furthermore, based on the meteorological driving data, perform power generation capacity prediction to obtain the predicted power generation amount. The method includes: Pre - defined power generation - related variables, where the power generation - related variables are the union of wind - power - related variables and solar - power - related variables; taking the power generation - related variables as constraints, interact with the meteorological API to collect meteorological variables and obtain the meteorological driving data, where the meteorological driving data has a meteorological time - window identifier; split the meteorological driving data into first meteorological driving data and second meteorological driving data according to the wind - power - related variables and solar - power - related variables; pre - construct a power generation prediction model, where the power generation prediction model includes first and second power generation prediction branches connected in parallel; through mapping and synchronizing the first meteorological driving data and the second meteorological driving data to the first and second power generation prediction branches of the power generation prediction model respectively for power generation prediction analysis, obtain a first power generation prediction value and a second power generation prediction value; introduce a power generation prediction deviation, and perform a fusion calculation on the first power generation prediction value and the second power generation prediction value based on the power generation prediction deviation, and output the predicted power generation.

[0046] Identify the key meteorological variables that directly affect wind - power and solar - power generation. For wind power, these meteorological variables include wind speed, wind direction, etc., which affect the power generation efficiency and output of wind turbines. For solar power, these meteorological variables include solar radiation intensity and cloud cover, which affect the amount of sunlight received by solar panels and power generation efficiency. Combine these variables into a set as the power generation - related variables for subsequent data collection and power generation prediction.

[0047] Set the parameters for data collection according to the pre - defined power generation - related variables, guide the API call, interact with the meteorological API, request the necessary meteorological data, and the obtained meteorological driving data has a meteorological time - window identifier, such as the next 24 hours, for determining the time range for which power generation prediction is required.

[0048] Extract the first meteorological driving data from the meteorological driving data according to the wind - power - related variables. The first meteorological driving data focuses on wind power. For example, wind speed and wind direction data are used for wind - power prediction; extract the second meteorological driving data from the meteorological driving data according to the solar - power - related variables. The second meteorological driving data focuses on solar power. For example, solar radiation intensity and cloud cover data are used for solar - power prediction. Further, set a time step K, such as 1 hour, and organize the first meteorological driving data and the second meteorological driving data into multiple groups respectively according to this time step K. Each group contains all relevant meteorological data within this time step K for time - series analysis.

[0049] Based on LSTM, a power generation prediction model with two parallel branches is pre-constructed. Each branch focuses on one type of power generation. Among them, the first power generation prediction branch is for wind power, and the second power generation prediction branch is for solar power. Each branch can independently process the input meteorological data, predict the power generation, and summarize the prediction results in the final stage. For each branch, historical data is used for training and validation, so that the model can accurately predict the power generation under different future meteorological conditions.

[0050] Input the first meteorological driving data into the first power generation prediction branch, and analyze the wind power generation according to variables such as wind speed and wind direction; input the second meteorological driving data into the second power generation prediction branch, and analyze the solar power generation according to variables such as solar radiation intensity and cloud cover. Each branch uses the algorithm obtained by training with historical data to predict the power generation in the upcoming time period and output its predicted power generation, that is, the first power generation prediction value and the second power generation prediction value.

[0051] Analyze the deviation between the output of each prediction branch and the actual power generation, such as calculating the error percentage, to obtain the power generation prediction deviation. Use a fusion algorithm, such as the weighted average method, to perform a fusion calculation on the prediction results of the two branches. The weights are set based on the power generation prediction deviation. For example, if the power generation prediction deviation of one branch is usually smaller than that of the other, a higher weight is given to this branch. The predicted power generation is output through the fusion calculation, and this value represents the comprehensive power generation capacity prediction considering all relevant meteorological factors.

[0052] Through this method, not only the power generation of a single energy type is predicted, but also the accuracy and reliability of the prediction are enhanced by integrating different prediction results. Such a method can better cope with changing meteorological conditions and different power generation technology characteristics, and provide more accurate and practical power generation predictions.

[0053] Furthermore, a power generation prediction model is pre-constructed, wherein the power generation prediction model includes a first power generation prediction branch and a second power generation prediction branch in parallel, and the method includes: Under the constraint of the wind power related variables, call and obtain the wind power meteorological data sequence and the power generation data sequence from the historical data; pre-construct a sliding segmentation window, where the sliding segmentation window has the same time span as the meteorological time window, and the sliding segmentation window contains K sample time steps; use the sliding segmentation window to segment the wind power meteorological data sequence and the power generation data sequence into multiple sample meteorological data sequences and multiple sample historical power generations, where each sample meteorological data sequence includes K meteorological data combinations, and the data indicators of each meteorological data combination conform to the wind power related variables; construct the first power generation prediction branch based on the LSTM model; use the multiple sample meteorological data sequences and multiple sample historical power generations as training data to train the first power generation prediction branch until the mean square error of the first power generation prediction branch meets the preset value; and so on, construct the second power generation prediction branch, and complete the construction of the power generation prediction model by paralleling the first power generation prediction branch and the second power generation prediction branch.

[0054] Under the constraint of the wind power related variables, that is, setting the data acquisition parameters with the wind power related variables, extract the wind power meteorological data directly related to wind power production from the storage system, such as wind speed, wind direction and temperature, and extract the power generation data corresponding to the wind power meteorological data. These data usually come from the historical operation records of past wind farms. Construct a time series, match the wind power meteorological data at each time point with the power generation data at the corresponding time, and obtain the wind power meteorological data sequence and the power generation data sequence.

[0055] Define a sliding segmentation window, which has the same time span as the meteorological time window. Set K sample time steps within the sliding segmentation window. K is set according to specific requirements and is used for data segmentation. For example, if it is necessary to predict the wind power generation in the next 1 hour, the sliding segmentation window can be set to 24 hours, which means using the wind speed, wind direction, temperature and other data in the past 24 hours to predict the wind power generation in the next 1 hour.

[0056] Use the sliding segmentation window to segment the wind power meteorological data sequence and the power generation data sequence according to time into multiple samples. Each sample contains continuous data points within a specific time window. After segmentation, multiple sample meteorological data sequences and multiple sample historical power generations are obtained. Among them, each sample meteorological data sequence includes K meteorological data combinations, including wind power related variables such as wind speed, wind direction, temperature and atmospheric pressure. Such a structure enables the sample set to reflect the influence of historical wind power meteorological data on future wind power generation in time series analysis.

[0057] Select the Long Short-Term Memory Network (LSTM), which is a deep learning model particularly suitable for time series data and can learn long-term dependencies in the data. Define the architecture of the LSTM model, including an input layer, one or more LSTM layers, and an output layer. The input layer receives meteorological data for K time steps, and the output layer predicts the power generation for the next time step. Construct the first power generation prediction branch according to this LSTM model.

[0058] Use multiple sample meteorological data sequences and multiple sample historical power generations as training data to train the first power generation prediction branch. During the training process, adjust the network parameters and use a regression loss function to evaluate the model performance. Multiple iterations of training and parameter adjustment are required to achieve the best performance. Monitor the loss value during the training process and continuously optimize until the mean square error drops below a preset threshold.

[0059] Construct a second power generation prediction branch based on the LSTM model. Similar to the first power generation prediction branch, perform data preparation, model training, and performance evaluation on the second power generation prediction branch until a second power generation prediction branch that meets the requirements is obtained. Connect the first and second power generation prediction branches in parallel and use data fusion techniques, such as weighted average, to integrate the power generation predictions of the two prediction branches. In this way, a power generation prediction model that combines wind power generation prediction and solar power generation prediction can be obtained. This method not only enhances the model's prediction ability for different energy types but also improves the overall prediction accuracy and reliability, providing strong data support for energy management and scheduling.

[0060] Furthermore, with the meteorological time window as a constraint, perform load demand analysis based on the meteorological driving data to obtain the predicted load demand. The method includes: Pre-define electricity consumption correlation variables and extract electricity consumption-related meteorological data from the meteorological driving data according to the electricity consumption correlation variables; perform time feature extraction on the meteorological time window to obtain electricity consumption-related time features, where the electricity consumption-related time features include week encoding, time encoding, and season encoding; perform multivariate load demand analysis based on the electricity consumption-related meteorological data and electricity consumption-related time features to obtain multivariate load demand information for multiple electricity consumption types; perform historical load fluctuation analysis on the multiple electricity consumption types and output multivariate load fluctuation coefficients; use the multivariate load fluctuation coefficients to perform data fusion on the multivariate load demand information and output the predicted load demand.

[0061] Identify the key meteorological variables that affect electricity demand as electricity - related variables, such as temperature, humidity, sunlight intensity, sunshine duration, etc. These variables directly or indirectly affect electricity consumption. For example, high temperatures may lead to an increase in air - conditioner usage, thus increasing electricity demand. Extract relevant information from the meteorological - driving data according to the defined electricity - related variables. For example, select temperature and humidity data from the meteorological - driving data, which are usually recorded by time stamps, and take the extracted data as electricity - related meteorological data.

[0062] Extract information about time from the meteorological time window, such as day of the week, time of day (hour), and season. These information have obvious impacts on electricity demand. For example, electricity consumption may be higher during the day on weekdays or in summer. Convert the extracted time information into a format available for machine - learning models. For example, use one - hot encoding to convert the day of the week, time, and season into numerical features, obtaining day - of - week encoding, time encoding, and season encoding, which together form electricity - related time features.

[0063] Utilize the electricity - related meteorological data and electricity - related time features to analyze the load demands of different electricity - consumption types. Electricity - consumption types include industrial electricity, residential electricity, office electricity, commercial electricity, etc. Specifically, use machine - learning models to establish relationship models between each electricity - consumption type and meteorological conditions and time features. Use historical electricity - consumption data and meteorological data for model training to predict load demands under different conditions. Use the trained models to predict the load demands of multiple electricity - consumption types and output multi - variable load - demand information.

[0064] Analyze the volatility of electricity loads for each electricity - consumption type in historical data, and identify electricity - consumption fluctuation patterns under different meteorological and time conditions. Exemplarily, use statistical methods such as analysis of variance and standard - deviation calculation to determine the volatility of each electricity - consumption type under different conditions, and calculate the load - fluctuation coefficients for each electricity - consumption type. These coefficients represent the variability or fluctuation amplitude of electricity consumption under specific meteorological or time conditions. The fluctuation coefficients can help predict changes in electricity demand under extreme meteorological conditions or specific time periods (such as heatwaves, cold snaps, holidays).

[0065] Combine the multi - variable load - fluctuation coefficients with the multi - variable load - demand information, and adjust the load prediction through weighted methods or other statistical techniques to reflect the volatility of different electricity - consumption types under different conditions. Take the fusion result as the predicted load demand. This fusion helps improve the accuracy and reliability of load prediction, especially when facing extreme meteorological conditions or different seasonal changes.

[0066] Furthermore, conduct a multi - variable load - demand analysis based on the electricity - related meteorological data and electricity - related time features to obtain multi - variable load - demand information for multiple electricity - consumption types. The method includes: Perform meteorological change correlation analysis on the multiple electricity consumption types to obtain multiple meteorological correlation degrees; divide the multiple electricity consumption types into a strong correlation type set and a weak correlation type set according to the multiple meteorological correlation degrees; construct H load demand prediction branches for the H electricity consumption types in the strong correlation type set, and complete the construction of the strong correlation load prediction model by paralleling the H load demand prediction branches; construct M load demand prediction branches for the M electricity consumption types in the weak correlation type set, and complete the construction of the weak correlation load prediction model by paralleling the M load demand prediction branches; load the electricity consumption related meteorological data and electricity consumption related time characteristics into the strong correlation load prediction model to obtain H strong correlation load demand information; load the electricity consumption related time characteristics into the weak correlation load prediction model to obtain M weak correlation load demand information; the H strong correlation load demand information and the M weak correlation load demand information constitute the multi - element load demand information.

[0067] Use statistical methods, such as correlation analysis or regression analysis, to evaluate the correlation between different electricity consumption types and meteorological factors. The purpose of the evaluation is to quantify the meteorological correlation degree between each electricity consumption type and specific meteorological variables, that is, to measure how various electricity consumption types respond to these changes when meteorological conditions change. The higher the meteorological correlation degree, the greater the degree to which the electricity consumption type is affected by meteorological factors. For example, temperature is highly correlated with domestic electricity consumption (especially air - conditioner usage).

[0068] According to the obtained multiple meteorological correlation degrees, divide the multiple electricity consumption types into two categories, namely, a strong correlation type set and a weak correlation type set. For example, set a correlation degree threshold. If the meteorological correlation degree is greater than 0.5, it is a strong correlation; if it is less than or equal to 0.5, it is a weak correlation. The strong correlation type set includes those electricity consumption types whose electricity consumption is significantly affected by meteorological changes. For example, the increase in domestic electricity consumption caused by the significant increase in air - conditioner usage in hot weather; the weak correlation type set includes those electricity consumption types with a relatively low correlation with meteorological factors, such as some industrial and office electricity consumption, whose electricity demand may be more determined by production plans and working hours rather than meteorological conditions.

[0069] Use the LSTM model to establish a relationship model between meteorological data and load demand for the H electricity consumption types in the strong correlation type set. This modeling is at the electricity consumption type dimension and is consistent with the logic of LSTM in the aforementioned power generation prediction. Similarly, based on historical data, call the electricity consumption related meteorological data sequence and the load demand data sequence under different electricity consumption types as training data to train the LSTM model to predict the load demand under different conditions. After training, obtain H load demand prediction branches corresponding to the H electricity consumption types. Each branch corresponds to the characteristics of different electricity consumption types. Parallel the multiple branches to obtain the strong correlation load prediction model.

[0070] For M types of electricity consumption in the weakly associated type set, M load demand prediction branches are constructed. Since the demands of these electricity consumption types have a low correlation with meteorological factors, the prediction branches mainly rely on time features, including week, time point, season, etc. Prediction methods suitable for dealing with non-weather-sensitive loads are used to construct and train the prediction branches, such as time series analysis, such as ARIMA, seasonal decomposition, etc. The historical electricity consumption data and time feature data are also used to train each branch, and the model is adjusted to maximize the prediction accuracy under different time conditions. The trained M load demand prediction branches are connected in parallel to form a weakly associated load prediction model. This model supports simultaneously processing M types of electricity consumption in the weakly associated type set, but mainly focuses on the impact of time factors on electricity demand.

[0071] Take the electricity consumption associated meteorological data and electricity consumption associated time features as input data and input them into the strongly associated load prediction model. The model predicts the electricity demands of different electricity consumption types based on these data and outputs the strongly associated load demand information of each strongly associated electricity consumption type. These information reflect the changes in electricity demands under different meteorological and time conditions.

[0072] Take the electricity consumption associated time features as input data and input them into the weakly associated load prediction model. Since the correlation between these electricity consumption types and meteorological factors is low, the model mainly relies on time features to predict electricity demand. According to the time features, the weakly associated load demand information of each weakly associated electricity consumption type is output, including the changes in electricity demands during different time periods such as weekdays and non-weekdays, day and night, different seasons, etc.

[0073] Integrate the obtained H strongly associated load demand information and M weakly associated load demand information to form multi-source load demand information. This method makes accurate predictions for different electricity consumption characteristics and improves the responsiveness of energy use.

[0074] In summary, the load balance-driven gravity energy storage system scheduling optimization method provided by the embodiments of this application has the following technical effects: Collecting meteorological variables through an interactive meteorological API to obtain meteorological driving data, which provides accurate meteorological environment information for subsequent steps. These information are an indispensable basis for predicting power demand and power generation, ensuring that the prediction model reflects the current and expected meteorological conditions, reducing the possible errors in traditional prediction methods, and improving the response ability and reliability of the smart grid; Using meteorological driving data to predict power generation capacity, especially for energy forms such as wind power and solar energy that highly depend on meteorological conditions, can significantly improve the accuracy of power generation prediction, which is of great significance for balancing the power grid supply and demand and optimizing the allocation of power generation resources; By analyzing the relationship between meteorological driving data and electricity demand, accurately predicting the electricity demand within a specific future time window, thereby optimizing the scheduling and management of power resources; By calculating the load balance deviation according to the predicted load demand and predicted power generation, obtaining the power grid supply and demand deviation, and identifying possible power surpluses or shortages. This precise prediction and calculation can ensure that the smart grid can make scheduling decisions in a timely manner when dealing with changing electricity demands, ensuring the power grid supply and demand balance, reducing the situation of power surpluses or shortages, and thus improving the stability and operation efficiency of the power grid; By combining the specific configuration of the gravity energy storage system and the actual power grid supply and demand deviation, optimizing the charge and discharge strategy of the energy storage system, providing a basis for the efficient operation of the smart grid, enabling the power grid to adjust the energy storage system in a timely manner according to the load fluctuation and the power generation fluctuation of renewable energy, thereby reducing energy waste, improving the response speed of the power grid and the load balancing ability; During the charge and discharge scheduling process, by real-time monitoring the state of the energy storage device and dynamically optimizing the charge and discharge strategy according to the monitoring results, supporting the energy storage system to always operate in the best state. This not only improves the response speed and flexibility of the gravity energy storage system, but also enhances the load balancing ability of the smart grid when facing the power fluctuations caused by renewable energy.

[0075] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for optimizing the scheduling of a gravity energy storage system driven by load balancing, characterized in that, The method includes: Interacting with a meteorological API to collect meteorological variables and obtain meteorological driving data, where the meteorological driving data has a meteorological time window identifier; Predicting the power generation capacity based on the meteorological driving data to obtain the predicted power generation; Conducting a load demand analysis based on the meteorological driving data with the meteorological time window as a constraint to obtain the predicted load demand; Calculating the load balance deviation based on the predicted load demand and the predicted power generation to obtain the power grid supply-demand deviation; Extracting energy storage configuration parameters from the gravity energy storage system, and conducting energy storage scheduling analysis based on the energy storage configuration parameters and the power grid supply-demand deviation, and outputting a gravity energy storage scheduling strategy; During the charge and discharge scheduling of multiple gravity energy storage devices in the gravity energy storage system according to the gravity energy storage scheduling strategy, the operation status of the multiple gravity energy storage devices is monitored in real time, and the gravity energy storage scheduling strategy is dynamically optimized according to the monitoring results until the end time of the meteorological time window.

2. The scheduling optimization method for a load-balanced drive gravity energy storage system according to claim 1, characterized in that Extracting energy storage configuration parameters from the gravity energy storage system, and conducting energy storage scheduling analysis based on the energy storage configuration parameters and the power grid supply-demand deviation, and outputting a gravity energy storage scheduling strategy, the method includes: Parsing the energy storage configuration parameters to obtain multiple sets of device configuration parameters of the multiple gravity energy storage devices; When the deviation vector of the power grid supply-demand deviation is positive, using the deviation vector of the power grid supply-demand deviation to traverse the multiple sets of device configuration parameters to screen N gravity energy storage devices from the multiple gravity energy storage devices; Conducting the charging amount configuration of the N gravity energy storage devices with the power grid supply-demand deviation as a constraint to obtain N initial charging amounts; Evaluating the scheduling fitness of the N initial charging amounts, and optimizing the charging amounts according to the evaluation results to generate N device energy storage scheduling strategies; The N device energy storage scheduling strategies constitute the gravity energy storage scheduling strategy.

3. The scheduling optimization method for a load-balanced drive gravity energy storage system according to claim 2, wherein, Among the multiple sets of device configuration parameters, each set of device configuration parameters includes energy storage capacity limit, charge and discharge rate, charge and discharge efficiency, remaining energy storage capacity, and device operation status.

4. The scheduling optimization method for the load-balanced drive gravity energy storage system according to claim 3, wherein, Evaluating the scheduling fitness of the N initial charging amounts, and optimizing the charging amounts according to the evaluation results to generate N device energy storage scheduling strategies, the method includes: Calculating the predicted power generation rate according to the predicted power generation and the meteorological time window; Calling the N charge and discharge rates, N charge and discharge efficiencies, and N remaining energy storage capacities of the N gravity energy storage devices from the multiple sets of device configuration parameters; Evaluating the energy consumption of the N initial charging amounts according to the N charge and discharge efficiencies to obtain a first energy consumption coefficient; Updating the data of the N initial charging amounts with the N remaining energy storage capacities as a constraint according to the deviation between the predetermined energy consumption constraint and the first energy consumption coefficient to obtain N updated charging amounts; Evaluating the energy consumption of the N updated charging amounts according to the N charge and discharge efficiencies to obtain a second energy consumption coefficient; And so on, according to the deviation between the energy consumption coefficient and the energy consumption constraint, within the capacity intervals of the N remaining energy storage capacities, adjust the charging amounts of the N gravity energy storage devices until N target charging amounts that meet the energy consumption constraint are output; Sort the scheduling priorities of the N gravity energy storage devices according to the predicted power generation rate and the N charging and discharging rates, and calculate the scheduling times according to the priority sorting result and the N target charging amounts to obtain N scheduling time intervals; The N scheduling time intervals and the N target charging amounts constitute the energy storage scheduling strategy for the N devices.

5. The scheduling optimization method of the load-balanced drive gravity energy storage system according to claim 1, characterized in that, Based on the meteorological driving data, predict the power generation capacity to obtain the predicted power generation amount. The method includes: Pre-define power generation correlation variables, where the power generation correlation variables are the union of wind power correlation variables and solar power correlation variables; Using the power generation correlation variables as constraints, interact with the meteorological API to collect meteorological variables to obtain the meteorological driving data, where the meteorological driving data has a meteorological time window identifier; According to the wind power correlation variables and solar power correlation variables, split the meteorological driving data into first meteorological driving data and second meteorological driving data; Pre-construct a power generation amount prediction model, where the power generation amount prediction model includes a first power generation amount prediction branch and a second power generation amount prediction branch connected in parallel; By synchronously mapping the first meteorological driving data and the second meteorological driving data to the first power generation amount prediction branch and the second power generation amount prediction branch of the power generation amount prediction model for power generation amount prediction analysis, obtain a first power generation amount prediction value and a second power generation amount prediction value; Introduce a power generation amount prediction deviation, and perform a fusion calculation on the first power generation amount prediction value and the second power generation amount prediction value based on the power generation amount prediction deviation, and output the predicted power generation amount.

6. The scheduling optimization method for a load-balanced drive gravity energy storage system according to claim 5, characterized in that Pre-construct a power generation amount prediction model, where the power generation amount prediction model includes a first power generation amount prediction branch and a second power generation amount prediction branch connected in parallel. The method includes: Under the constraint of the wind power correlation variables, call the wind power meteorological data sequence and the power generation amount data sequence from the historical data; Pre-construct a sliding segmentation window, where the sliding segmentation window has the same time span as the meteorological time window, and the sliding segmentation window contains K sample time steps; Use the sliding segmentation window to split the wind power meteorological data sequence and the power generation amount data sequence into multiple sample meteorological data sequences and multiple sample historical power generation amounts. Each sample meteorological data sequence includes K meteorological data combinations, and the data indicators of each meteorological data combination conform to the wind power correlation variables; Construct the first power generation amount prediction branch based on the LSTM model; Use the multiple sample meteorological data sequences and multiple sample historical power generation amounts as training data to train the first power generation amount prediction branch until the mean square error of the first power generation amount prediction branch meets the preset value; And so on, construct the second power generation amount prediction branch, and complete the construction of the power generation amount prediction model by connecting the first power generation amount prediction branch and the second power generation amount prediction branch in parallel.

7. The scheduling optimization method for a load-balanced drive gravity energy storage system according to claim 6, wherein, Constrained by the meteorological time window, load demand analysis is performed based on the meteorological driving data to obtain predicted load demand. The method includes: Pre-define electricity consumption correlation variables, and extract electricity consumption-related meteorological data from the meteorological driving data according to the electricity consumption correlation variables; Extract time features from the meteorological time window to obtain electricity consumption-related time features, where the electricity consumption-related time features include day-of-week encoding, time encoding, and season encoding; Perform multivariate load demand analysis based on the electricity consumption-related meteorological data and electricity consumption-related time features to obtain multivariate load demand information for multiple electricity consumption types; Perform historical load fluctuation analysis on the multiple electricity consumption types and output multivariate load fluctuation coefficients; Use the multivariate load fluctuation coefficients to perform data fusion on the multivariate load demand information and output the predicted load demand.

8. The scheduling optimization method for a load-balanced drive gravity energy storage system according to claim 7, characterized in that Perform multivariate load demand analysis based on the electricity consumption-related meteorological data and electricity consumption-related time features to obtain multivariate load demand information for multiple electricity consumption types. The method includes: Perform meteorological change correlation analysis on the multiple electricity consumption types to obtain multiple meteorological correlation degrees; According to the multiple meteorological correlation degrees, divide the multiple electricity consumption types into a strongly correlated type set and a weakly correlated type set; Construct H load demand prediction branches for H electricity consumption types in the strongly correlated type set, and complete the construction of the strongly correlated load prediction model by connecting the H load demand prediction branches in parallel; Construct M load demand prediction branches for M electricity consumption types in the weakly correlated type set, and complete the construction of the weakly correlated load prediction model by connecting the M load demand prediction branches in parallel; Load the electricity consumption-related meteorological data and electricity consumption-related time features into the strongly correlated load prediction model to obtain H strongly correlated load demand information; Load the electricity consumption-related time features into the weakly correlated load prediction model to obtain M weakly correlated load demand information; The H strongly correlated load demand information and the M weakly correlated load demand information constitute the multivariate load demand information.

Images