Charging and discharging scheduling method and device of energy storage power station and nonvolatile storage medium

Abstract

The invention discloses a charging and discharging scheduling method and device of an energy storage power station and a nonvolatile storage medium. The method comprises the following steps: acquiring power consumption data acquired by an edge computing node of an energy storage power station and historical data corresponding to the energy storage power station; based on the power consumption data and historical data, determining a first power consumption behavior trend of a corresponding area of the energy storage power station through a cloud deep learning network model; based on the first power consumption behavior trend, predicting the charge and discharge quantity of the energy storage power station to obtain a first prediction result; state detection data of the energy storage power station sent by the side end is obtained, and the state detection data comprises operation state data and an abnormal detection result of the energy storage power station; and carrying out charging and discharging scheduling on the energy storage power station in a preset time period based on the first prediction result and the state detection data. According to the invention, the technical problems of rapid charging and discharging of the battery, damage to the service life of the battery and high labor cost caused by temporary adjustment of the charging and discharging strategy of the energy storage power station based on real-time electricity price at present are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage power stations, and in particular to a charge and discharge scheduling method and device for an energy storage power station, and a non-volatile storage medium. Background Art

[0002] With the transformation of the global energy structure and the rapid development of new energy technologies, energy storage has become a core technology for energy conversion and buffering, peak load regulation and efficiency improvement, transmission and scheduling, management, and utilization. Energy storage systems not only balance power supply and demand and optimize grid operation, but also provide the primary driving force for emerging industries such as electric vehicles, thus possessing significant strategic value. Current energy storage power station data collection systems present the following challenges:

[0003] 1. Currently, energy storage power stations mainly make profits by operating at the price difference of regional electricity prices within the day. However, this method usually achieves power transfer through deep charging during off-peak hours and deep discharging during peak hours, which accelerates battery aging, affects battery life, and leads to frequent battery replacement in power stations.

[0004] 2. Heterogeneous energy storage systems, such as mixed energy storage batteries from different processes and production sites, and mixed energy storage batteries of different categories, make it difficult to coordinate the energy storage system and implement optimization strategies.

[0005] 3. The traditional optimization algorithms currently used in energy storage systems are difficult to implement real-time adjustment strategies. Furthermore, they lack intelligence, making the work of energy storage system maintenance personnel more difficult and reducing the operator's profits.

[0006] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0007] Embodiments of the present invention provide a method, device, and non-volatile storage medium for scheduling charging and discharging of an energy storage power station, to at least address the current technical problem of temporarily adjusting the charging and discharging strategy of an energy storage power station based on real-time electricity prices, resulting in rapid charging and discharging of batteries, shortening battery life, and increasing labor costs.

[0008] According to one aspect of an embodiment of the present invention, a method for charging and discharging scheduling of an energy storage power station is provided, comprising: obtaining electricity usage data collected by an edge computing node of the energy storage power station and historical data corresponding to the energy storage power station; determining, based on the electricity usage data and the historical data, a first electricity usage behavior trend in a region corresponding to the energy storage power station through a cloud-based deep learning network model; predicting the charge and discharge amount of the energy storage power station based on the first electricity usage behavior trend to obtain a first prediction result; obtaining status detection data of the energy storage power station sent by the edge, wherein the status detection data includes operating status data and anomaly detection results of the energy storage power station; and scheduling charging and discharging of the energy storage power station within a preset time period based on the first prediction result and the status detection data.

[0009] Optionally, before determining the first electricity consumption behavior trend of the area corresponding to the energy storage power station through the cloud-based deep learning network model based on electricity consumption data and historical data, the method includes: determining the target function corresponding to the charging amount according to the charging amount of the energy storage power station and a preset loss function; determining the target function corresponding to the discharge amount according to the discharge amount of the energy storage power station and the loss function; determining the target profit function according to the target function corresponding to the charging amount, the target function corresponding to the discharge amount and the real-time electricity price of the area corresponding to the energy storage power station; and determining the cloud-based deep learning network model according to the target charging amount function, the target discharge amount function and the target profit function.

[0010] Optionally, based on the electricity consumption data and historical data, a first electricity consumption behavior trend of the area corresponding to the energy storage power station is determined through a cloud-based deep learning network model, including: standardizing the electricity consumption data and historical data to obtain processed electricity consumption data and processed historical data, wherein the standardization processing includes data cleaning, compression and formatting; and inputting the processed electricity consumption data and processed historical data into the cloud-based deep learning network model to obtain the first electricity consumption behavior trend of the area corresponding to the energy storage power station.

[0011] Optionally, based on the first electricity consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a first prediction result, including: obtaining a prediction feature, wherein the prediction feature includes at least one of the following: average electricity consumption, peak electricity consumption time and electricity price; based on the first electricity consumption behavior trend, determining a numerical value corresponding to the prediction feature; based on the numerical value corresponding to the prediction feature, predicting the charge and discharge amount of the energy storage power station to obtain a first prediction result, wherein the prediction result includes a target charge amount and a target discharge amount.

[0012] Optionally, when an abnormality occurs in the network corresponding to the deep learning network model in the cloud, the control edge determines a second electricity consumption behavior trend in the area corresponding to the energy storage power station based on the pre-deployed local model through electricity consumption data and historical data; based on the second electricity consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a second prediction result; based on the second prediction result and status detection data, the energy storage power station is scheduled to charge and discharge within a preset time period.

[0013] Optionally, the local model is obtained through the following steps: inputting preset sample electricity consumption data into the cloud-based deep learning network model to obtain a first output result; building an algorithm model based on the initial deep learning architecture to obtain an initial model; training the initial model based on the sample electricity consumption data and the first output result until the output result of the initial model matches the first output result to obtain a target model; sending the target model to the edge and deploying it to obtain a local model, wherein the edge monitors the operating status of the energy storage power station based on the local model.

[0014] According to another aspect of an embodiment of the present invention, a charging and discharging scheduling system for an energy storage power station is also provided, including: a cloud end, used to execute any of the above-mentioned charging and discharging scheduling methods for an energy storage power station; an edge end, used to monitor the operating status of the energy storage power station, perform anomaly detection, collect power consumption data and historical data based on edge computing nodes, and upload the operating status data, anomaly detection results, power consumption data and historical data to the cloud end.

[0015] Optionally, in the event of a network anomaly in the cloud, the edge is also used to determine a second electricity consumption behavior trend in the area corresponding to the energy storage power station based on a pre-deployed local model using electricity consumption data and historical data; based on the second electricity consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a second prediction result; based on the second prediction result, the charge and discharge of the energy storage power station is scheduled within a preset time period.

[0016] According to another aspect of an embodiment of the present invention, a charging and discharging scheduling device for an energy storage power station is also provided, including: a first acquisition module, used to obtain electricity consumption data collected by an edge computing node of the energy storage power station and historical data corresponding to the energy storage power station; a determination module, used to determine, based on the electricity consumption data and historical data, a first electricity consumption behavior trend in the area corresponding to the energy storage power station through a cloud-based deep learning network model; a prediction module, used to predict the charging and discharging amount of the energy storage power station based on the first electricity consumption behavior trend to obtain a first prediction result; a second acquisition module, used to obtain status detection data of the energy storage power station sent by the edge, wherein the status detection data includes: operating status data and anomaly detection results of the energy storage power station; and a scheduling module, used to schedule the charging and discharging of the energy storage power station within a preset time period based on the first prediction result and the status detection data.

[0017] According to another aspect of an embodiment of the present invention, a non-volatile storage medium is provided. The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute any of the above-mentioned charge and discharge scheduling methods for energy storage power stations.

[0018] According to another aspect of an embodiment of the present invention, a computer device is provided. The computer device includes a processor, and the processor is used to run a program. When the program is run, any one of the above-mentioned charge and discharge scheduling methods for an energy storage power station is executed.

[0019] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program. When the computer program is executed by a processor, it implements any of the above-mentioned charge and discharge scheduling methods for an energy storage power station.

[0020] In an embodiment of the present invention, a charging and discharging scheduling method for an energy storage power station is adopted, by obtaining electricity consumption data collected by an edge computing node of the energy storage power station and historical data corresponding to the energy storage power station; based on the electricity consumption data and historical data, a first electricity consumption behavior trend of the area corresponding to the energy storage power station is determined through a cloud-based deep learning network model; based on the first electricity consumption behavior trend, the charging and discharging amount of the energy storage power station is predicted to obtain a first prediction result; status detection data of the energy storage power station sent by the edge is obtained, wherein the status detection data includes: operating status data and anomaly detection results of the energy storage power station; based on the first prediction result and the status detection data, the energy storage power station is charged and discharged within a preset time period, thereby achieving the purpose of predicting the charging and discharging amount of the energy storage power station based on the edge-cloud collaborative deep learning algorithm model, thereby realizing the technical effect of intelligently scheduling the charging and discharging amount of the energy storage power station, and further solving the current technical problem of temporarily adjusting the charging and discharging strategy of the energy storage power station based on the real-time electricity price, resulting in fast charging and discharging of the battery, damaged battery life, and high labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0022] Figure 1 A hardware structure block diagram of a computer terminal for implementing a charge and discharge scheduling method for an energy storage power station is shown;

[0023] Figure 2 1 is a flow chart of a method for scheduling charging and discharging of an energy storage power station according to an embodiment of the present invention;

[0024] Figure 3 is a schematic diagram of a charge and discharge scheduling method for an energy storage power station provided according to an optional embodiment of the present invention;

[0025] Figure 4 4 is a structural block diagram of a charge and discharge scheduling device for an energy storage power station provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described 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 making creative efforts should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] According to an embodiment of the present invention, a method embodiment of a method for scheduling charging and discharging of an energy storage power station is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0029] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 The hardware structure block diagram of a computer terminal for implementing a charge and discharge scheduling method for an energy storage power station is shown. Figure 1 As shown, the computer terminal 10 may include one or more (illustrated as 102a, 102b, ..., 102n in the figure) processors (the processor may include but is not limited to a microprocessor MCU or a programmable logic device FPGA and other processing devices), a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0030] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry." The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuitry may be a single, independent processing module, or may be incorporated in whole or in part into any of the other components of the computer terminal 10. As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0031] Memory 104 can be used to store software programs and modules for application software, such as the program instructions / data storage device corresponding to the method for charging and discharging scheduling of an energy storage power station in the embodiments of the present invention. The processor executes the software programs and modules stored in memory 104 to execute various functional applications and data processing, thereby implementing the method for charging and discharging scheduling of an energy storage power station for the aforementioned application. Memory 104 can include high-speed random access memory (RAM) and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 104 can further include memory remotely located from the processor, which can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0032] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0033] Figure 2 FIG. 1 is a flow chart of a method for scheduling charging and discharging of an energy storage power station according to an embodiment of the present invention. Figure 2 As shown, the method includes the following steps:

[0034] Step S202: Obtain electricity consumption data collected by the edge computing node of the energy storage power station and historical data corresponding to the energy storage power station.

[0035] In this step, in order to predict the future charge and discharge capacity of the energy storage power station, it is necessary to collect historical data of the energy storage power station and electricity consumption data in the area. Therefore, edge computing nodes are deployed at the energy storage power station to collect data. Both the historical data of the energy storage power station and the electricity consumption data in the area can be collected at the same time. Based on the current electricity consumption data and the historical data of the energy storage power station, it is convenient to predict the future charge and discharge capacity of the energy storage power station in subsequent steps.

[0036] Specifically, edge computing nodes can be deployed based on data acquisition devices to collect data. Multi-protocol data acquisition (such as Modbus, CAN, and IEC61850) can be supported. Power usage data includes battery load data, which can be understood as the battery's immediate power input and output, reflecting the battery's current operating intensity. Battery status data includes information such as the battery's state of health (SOH), state of charge (SOC), and battery temperature. Obtaining this information can prevent overcharging and discharging, thereby extending battery life. Real-time electricity price data can also be included, including peak and valley prices and spot market prices. Historical data includes historical behavior data of energy storage power stations, i.e., records of past battery voltage and current changes, which can be used to analyze long-term battery performance. Historical battery charge and discharge data, i.e., records of battery charge and discharge at different time points, can help identify battery operating modes, such as peak and valley periods, and trends in charge and discharge efficiency. Economic and environmental historical data, i.e., historical battery temperature values ​​and historical electricity price data, can be used to analyze the impact of temperature on battery performance, organize past electricity price records, and identify patterns and patterns in electricity price fluctuations. It enables real-time acquisition of comprehensive electricity consumption data and historical data, providing a data basis for subsequent accurate predictions.

[0037] Step S204: Based on the electricity consumption data and historical data, a first electricity consumption behavior trend of the area corresponding to the energy storage power station is determined through a cloud-based deep learning network model.

[0038] In this step, the cloud-based deep learning network model is an algorithm model deployed in the cloud server. It is an algorithm model based on a complex deep learning architecture. It can be a trained model obtained by pre-training with massive historical data. Therefore, the collected data can be directly processed to obtain the electricity consumption behavior trend in the area based on the preset logical algorithm.

[0039] Specifically, deep learning architectures suitable for processing time series data can be used, such as long short-term memory networks (LSTMs), gated recurrent units (GRUs), or Transformer models combined with attention mechanisms. These models can capture long-term dependencies between data and are suitable for predicting electricity usage trends. A trained deep learning model can predict electricity usage behavior for a period of time based on current and historical electricity data, including peak and valley values, as well as changing trends. It can also predict future electricity price trends within a region, helping energy storage systems make optimal charging and discharging decisions based on price fluctuations and maximize economic benefits. Combined with historical battery data, it can also predict the battery's health status, enabling early detection of battery performance degradation and providing a basis for maintenance plans. Through in-depth analysis of electricity usage and historical data using cloud-based deep learning network models, patterns and trends in electricity usage can be identified, providing a foundation for subsequent predictions.

[0040] Step S206 : Based on the first electricity usage behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a first prediction result.

[0041] In this step, based on the electricity consumption trends in the area corresponding to the energy storage station, the electricity consumption in that area can be determined. Based on this electricity consumption, the charging and discharging capacity of the energy storage station can be predicted. For example, when electricity consumption is high, the electricity price may also be high. In this case, discharging operations are necessary. Integrating into the grid to supply power to the area can reduce the pressure on the grid. When electricity consumption is low, the electricity price may be relatively low. In this case, charging operations are necessary. The energy storage station can operate independently and no longer need to supply power to the area. Predictions based on electricity consumption trends avoid the problems of rapid charging and discharging of energy storage stations, which lead to short battery life and high labor costs.

[0042] And / or, step S208, obtaining status detection data of the energy storage power station sent by the edge, wherein the status detection data includes: operating status data and abnormality detection results of the energy storage power station.

[0043] In this step, the cloud server receives the status detection data sent by the edge, where the edge can be understood as a locally deployed system that collects and summarizes the power consumption data and historical data collected by the edge computing node. At the same time, it monitors the status of the energy storage power station, identifies abnormal conditions of the energy storage equipment, and uploads the status detection data to the cloud system.

[0044] Specifically, the edge can collect and transmit real-time operational status data of the energy storage power station, including but not limited to key indicators such as battery voltage, current, temperature, state of charge (SOC), state of health (SOH), charge and discharge power, charge and discharge efficiency, remaining capacity, and internal impedance. It can also detect abnormal behavior of the energy storage device in real time, such as battery overcharge, overdischarge, abnormal temperature, voltage instability, and abnormal SOC / SOH changes. Once an anomaly is detected, the edge immediately reports detailed anomaly detection results, including the anomaly type, time of the anomaly, abnormal parameter values, and possible anomaly severity. After receiving the status detection data, the cloud first parses the data packets and structures the data for further processing and storage. The parsed operational status data and anomaly detection results are stored in a high-performance database. Real-time data can be stored using in-memory databases or cache technologies to support low-latency data access. Historical data can be stored in persistent databases such as Clickhouse to facilitate data analysis and model training. When the cloud receives anomaly detection results from the edge, it can trigger an anomaly response mechanism, including sending alerts to operations and maintenance personnel, initiating emergency plans, adjusting charging and discharging strategies to protect equipment safety, or notifying relevant personnel to conduct on-site inspections and maintenance when necessary. Through edge-cloud collaboration, the cloud can comprehensively and promptly obtain energy storage power station status detection data from the edge, including operating status data and anomaly detection results. This allows the cloud to effectively monitor the operation of the energy storage system, respond to anomalies in a timely manner, and dynamically adjust charging and discharging strategies to ensure the safe operation and economic efficiency of energy storage equipment.

[0045] Step S210 : Based on the first prediction result and the status detection data, the energy storage power station is scheduled to charge and discharge within a preset time period.

[0046] In this step, the first prediction result includes a forecast of the energy storage device's charge and discharge demand within a preset time period. In addition to charge and discharge behavior, the prediction result can also include an estimate of the energy storage device's future state, such as the SOC (State of Charge) prediction, which can be used to plan charge and discharge strategies in advance to avoid excessive battery charge and discharge. Status detection data can confirm whether the energy storage device is currently in normal operation, for example, checking whether key parameters such as the current SOC level, battery temperature, and voltage are within a preset healthy range. Based on the first prediction result and status detection data, the charging and discharging of the energy storage power station can be scheduled.

[0047] Specifically, if status monitoring data indicates an equipment anomaly, such as battery overheating, excessive voltage fluctuations, or a sudden drop in SOC, the system can immediately take action, such as suspending charging and discharging operations, reducing charging and discharging power, or activating a backup battery pack, to ensure equipment safety. Combining the prediction results with the equipment's real-time status, the system generates a specific charging and discharging scheduling strategy. This strategy is then converted into specific operational instructions and sent to the energy storage plant's control system via the communication network. These instructions include charging and discharging start and stop times, charging and discharging power levels, and emergency response measures. For example, a scheduling cycle might run from 10 pm to 6 am the following morning, when market electricity prices are typically low. The model predicts the optimal charging and discharging strategy for the energy storage plant during this period, including charging at maximum power between 11 pm and 2 am to take advantage of the lowest electricity prices. However, real-time status monitoring data indicates that the energy storage equipment's SOC is nearing saturation, which, if left unchecked, could lead to battery overcharging. Therefore, the system adjusts charging operations, reducing charging and discharging power to prevent battery overcharging while also ensuring economic efficiency. Charging and discharging scheduling based on deep learning predictions and real-time status detection data can effectively balance the economic benefits and equipment safety of energy storage power stations, and is a key link in realizing intelligent management of energy storage systems.

[0048] Through the above steps, the goal of predicting the charge and discharge capacity of the energy storage power station based on the edge-cloud collaborative deep learning algorithm model can be achieved, thereby realizing the technical effect of intelligently scheduling the charge and discharge capacity of the energy storage power station, and then solving the current technical problem of temporarily adjusting the charge and discharge strategy of the energy storage power station based on real-time electricity prices, resulting in rapid charging and discharging of batteries, damaged battery life, and high labor costs.

[0049] As an optional embodiment, it can be achieved through the following steps: before determining the first electricity consumption behavior trend of the area corresponding to the energy storage power station through a cloud-based deep learning network model based on electricity consumption data and historical data, it includes: determining the target function corresponding to the charging amount according to the charging amount of the energy storage power station and a preset loss function; determining the target function corresponding to the discharge amount according to the discharge amount of the energy storage power station and the loss function; determining the target profit function according to the target function corresponding to the charging amount, the target function corresponding to the discharge amount and the real-time electricity price of the area corresponding to the energy storage power station; and determining the cloud-based deep learning network model according to the target charging amount function, the target discharge amount function and the target profit function.

[0050] Optionally, the cloud-based deep learning network model is a complex deep learning algorithm model. During model construction, the deep learning algorithm architecture parameters must be set, with optimal charge and discharge as targets, and the objective function determined. The optimal revenue can also be automatically calculated based on the optimal charge and discharge and real-time electricity prices. This enables automatic and intelligent prediction of energy storage power station charge and discharge capacity, ensuring increased revenue while reducing costs and maintaining battery life.

[0051] For example, the costs and benefits of charging and discharging operations are directly related to the current electricity price level. Charging during low-price periods and discharging during high-price periods can significantly reduce costs and increase benefits. Therefore, the objective functions for charging and discharging are designed to optimize this cost-benefit ratio. The depth of battery charge and discharge has a direct impact on battery life. Excessive charge and discharge accelerate battery aging and increase maintenance costs. By introducing a loss function, the model can be trained to avoid unfavorable charge and discharge depths, thereby extending battery life. The objective benefit function combines the economic benefits of charging and discharging operations while also considering the potential costs of deep charge and discharge. Using real-time electricity price information, the function can dynamically adjust the charge and discharge strategy to adapt to changing market conditions. By optimizing the objective benefit function, the system can maximize the economic benefits of the energy storage power station while meeting battery health constraints. This is a key goal in commercial operations and can directly impact the return on investment and operational sustainability.

[0052] Optionally, the optimal charging capacity objective function is as follows:

[0053]

[0054] in, Indicates i th The data at the moment, c i Indicates i th Charging data at all times, represents the trainable parameters, Indicates the next day corresponding to i th The optimal charging capacity prediction value at the moment, represents the data feature extractor, represents the cross entropy loss function, Indicates the next day corresponding to i th The estimated value of the charge capacity at each moment.

[0055] The optimal discharge capacity objective function is as follows:

[0056]

[0057] in, Indicates i th Data at the moment, dci Indicates i th Discharge data at the moment, represents the trainable parameters, Indicates the next day corresponding to i th The optimal discharge amount prediction value at the moment, represents the data feature extractor, represents the cross entropy loss function, Indicates the next day corresponding to i th Discharge prediction value at each moment

[0058] Based on the predicted optimal charge and discharge volume and real-time electricity price data for the next day, the profit for the next day is calculated. The optimal profit objective function is as follows:

[0059]

[0060] Among them, E pred Indicates the future day's profit, P t Indicates i th Time-of-day electricity price, DC t and C t Indicates the next day i th The absolute values ​​of the discharge and charge capacities obtained by prediction at each moment.

[0061] Based on the above objective function, a cloud-based deep learning network model is designed. During pre-training, model parameters such as the learning rate, number of network layers, and number of neurons can be adjusted to improve the model's prediction accuracy. Techniques such as early stopping and regularization can also be used to prevent overfitting and ensure good performance on unseen data. This process enables the construction and training of a cloud-based deep learning network model, laying the foundation for subsequent accurate predictions.

[0062] As an optional embodiment, it can be achieved through the following steps: based on electricity consumption data and historical data, a cloud-based deep learning network model is used to determine the first electricity consumption behavior trend of the area corresponding to the energy storage power station, including: standardizing the electricity consumption data and historical data to obtain processed electricity consumption data and processed historical data, wherein the standardization processing includes data cleaning, compression and formatting; inputting the processed electricity consumption data and processed historical data into the cloud-based deep learning network model to obtain the first electricity consumption behavior trend of the area corresponding to the energy storage power station.

[0063] Optionally, data collected by edge computing nodes deployed based on data acquisition devices often contain issues such as missing values, outliers, noise, and duplicate data. The purpose of data cleaning is to identify and address these issues to ensure data quality and integrity. Cleaned data can improve model training efficiency and prediction accuracy, thereby avoiding erroneous decisions. When processing large datasets, data compression can reduce data size, lower storage and transmission costs, and accelerate data processing. During data transmission between edge devices and the cloud, compression can significantly reduce bandwidth requirements and improve real-time performance. Heterogeneous energy storage systems, such as those with mixed energy storage batteries from different processes, production sites, or different types of energy storage batteries, result in collected data in different formats. Deep learning models typically require data in specific formats as input. Formatting ensures that data meets model input requirements, such as converting data to a time series format, unifying timestamps, or adjusting data types, so that the model can correctly interpret and process the data. Standardized data can be read and processed more quickly by deep learning models, reducing model training time. Clean, uniformly formatted, and standardized data makes it easier for models to identify patterns and trends, thereby improving the accuracy of predicting electricity usage trends. Standardizing electricity consumption and historical data ensures that the data input into the cloud-based deep learning network model is of high quality, uniform in format, and easy to process, thereby improving the efficiency of model training and the accuracy of predictions, and ultimately obtaining the first electricity consumption behavior trend in the corresponding area of ​​the energy storage power station.

[0064] Specifically, outliers in the dataset are identified and processed. These may be caused by measurement errors, equipment failures, or other external factors. Outlier processing methods include deletion, replacement, or correction using statistical methods (such as the median and mean). Electricity consumption data and historical data may contain missing values, which require appropriate methods to fill in. Common methods include using neighboring values, time series interpolation, rule-based filling, or prediction using deep learning models. While ensuring predictive effectiveness, feature selection and extraction can be used to reduce data dimensionality, remove irrelevant or redundant features, and reduce data storage and transmission overhead. Data can also be encoded, such as using differential encoding (difference encoding) or Huffman coding, to reduce the number of data bits and improve data transmission efficiency between edge devices and the cloud. Furthermore, data can be converted to a time series format, ensuring that each data point is associated with a specific timestamp so that deep learning models can identify temporal patterns and trends. Converting feature values ​​in the dataset to a uniform numerical range, such as using Z-score normalization, maximum and minimum value scaling, or decimal scaling, can improve model convergence speed and prediction accuracy. Before feeding the data into the model, a final data preprocessing step can be performed, such as partitioning the data into training, validation, and test sets to ensure that the model can be trained, validated, and tested on different datasets. The processed electricity consumption data and historical data are used to construct the input datasets. These datasets may require further feature engineering, such as creating sliding windows and adding lag features, to capture the temporal dependencies of the data.

[0065] As an optional embodiment, it can be achieved through the following steps: based on the first electricity consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a first prediction result, including: obtaining a prediction feature, wherein the prediction feature includes at least one of the following: average electricity consumption, peak electricity consumption time and electricity price; based on the first electricity consumption behavior trend, determining the numerical value corresponding to the prediction feature; based on the numerical value corresponding to the prediction feature, the charge and discharge amount of the energy storage power station is predicted to obtain a first prediction result, wherein the prediction result includes a target charge amount and a target discharge amount.

[0066] Optionally, within the electricity consumption trends output by the model, it is first necessary to analyze the trends and extract useful eigenvalues ​​to facilitate subsequent predictions. Average electricity consumption is a fundamental indicator for measuring daily electricity consumption within a region, reflecting users' average daily or weekly electricity demand. Understanding the peak times of electricity demand during peak hours generally results in higher electricity prices. Therefore, predicting peak times can help determine when to release energy storage to meet high demand and recharge during periods of low electricity prices. Fluctuations in electricity prices directly impact the operating costs and revenues of energy storage power stations. By analyzing historical electricity price data and market trends to predict future electricity prices, energy storage power stations can charge during expected low-price periods and discharge during high-price periods, maximizing economic benefits.

[0067] Specifically, average electricity consumption must first be extracted from historical electricity data. This average electricity consumption reflects the typical electricity demand level of a user or region and serves as the basis for developing charging plans. Average electricity consumption patterns over different time periods (e.g., weekdays, weekends, and holidays) are analyzed to more accurately predict electricity demand on specific days. Time series analysis of historical data identifies peak demand periods, taking into account the impact of market signals and special events (such as holidays and extreme weather conditions) on peak demand times to more accurately reflect potential anomalies or changes in the forecast. Future electricity prices are predicted by analyzing historical price trends in the electricity market and incorporating macroeconomic factors, supply and demand conditions, and renewable energy production. By collecting and analyzing predictive features, determining feature values ​​based on electricity consumption trends, and using deep learning networks to predict the charge and discharge capacity of energy storage power plants, accurate forecasts of electricity demand and market prices can be achieved, facilitating intelligent scheduling of energy storage power plants, optimizing the allocation of power resources, and reducing energy waste.

[0068] As an optional embodiment, it can be achieved through the following steps: when an abnormality occurs in the network corresponding to the deep learning network model in the cloud, the control edge determines the second power consumption behavior trend of the area corresponding to the energy storage power station based on the pre-deployed local model through power consumption data and historical data; based on the second power consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a second prediction result; based on the second prediction result and status detection data, the energy storage power station is scheduled to charge and discharge within a preset time period.

[0069] Optionally, by deploying a local model on the edge, in the event of a network anomaly on the cloud server, the cloud model can be used to predict power usage trends and adjust charging and discharging strategies based on the predicted results. This can be understood as using an edge-cloud collaborative mechanism to obtain power plant operation strategies from the cloud when the network is normal, and to generate power plant operation strategies on the edge when the network is abnormal, thus ensuring system reliability.

[0070] Specifically, the edge system continuously monitors its connection to the cloud. Upon detecting abnormal network latency or a connection interruption, it immediately triggers the activation of a local prediction mechanism. Upon detecting a network anomaly, the edge automatically switches to a pre-deployed local model. This model is a lightweight version of the model obtained through cloud distillation under normal conditions, designed to accommodate edge computing and storage constraints. The edge collects the latest operational status data from energy storage power station sensors and system logs in real time, including battery state of charge (SOC), state of health (SOH), current electricity prices, and external environmental conditions. This collected real-time data undergoes necessary preprocessing, including data cleaning (removing outliers), data normalization (ensuring data is on a consistent scale), and possible feature engineering (extracting key features relevant to prediction). This preprocessed data is fed into the local model to generate new predictions, including charge and discharge forecasts for a preset future time period. Based on these new predictions, the edge system automatically or with the assistance of operations personnel formulates a charge and discharge strategy for the preset time period. After the network anomaly is resolved, the edge continues to monitor the network status to ensure a stable connection to the cloud. Once the network stabilizes, the edge proactively requests the cloud to synchronize the latest model parameters or update the local model. In the event of an anomaly in the network corresponding to the cloud-based deep learning network model, the control edge predicts the charge and discharge capacity of the energy storage power station based on the local model. Through real-time data collection, local model prediction, strategy formulation and monitoring, and model synchronization after network recovery, the edge can ensure that when cloud services are unreachable, the system can still make reasonable and economical charging and discharging decisions, thereby maintaining the efficient operation and economic benefits of the power system.

[0071] As an optional embodiment, it can be implemented through the following steps: the local model can be implemented through the following steps: inputting the preset sample electricity consumption data into the cloud-based deep learning network model to obtain a first output result; building an algorithm model based on the initial deep learning architecture to obtain an initial model; training the initial model based on the sample electricity consumption data and the first output result until the output result of the initial model matches the first output result to obtain a target model; sending the target model to the edge and deploying it to obtain a local model, wherein the edge monitors the operating status of the energy storage power station based on the local model.

[0072] Optionally, a local model, or model deployed at the edge, is a lightweight model. During the creation of the local model, it is built based on the algorithm model of the initial deep learning architecture. Compared to the architecture of the cloud-based deep learning network model, this model is simpler and easier to deploy at the edge. At the same time, it is necessary to ensure that the local model has the same operational logic as the cloud-based model, that is, to ensure that the prediction results match. In this case, the initial model needs to be trained multiple times until the output meets the expected results. The initial model after training is the local model. The local model can also be used to monitor the status of the energy storage power station and provide fault notifications.

[0073] Specifically, distillation can be used to simplify cloud-based deep learning network models, resulting in a cloud-based distilled model. This method, based on the concept of knowledge distillation, involves having a smaller model (the student model) learn and mimic the behavior of a larger pre-trained model (the teacher model). By learning the teacher model's predictions on a large amount of data, the student model can learn similar decision boundaries, retaining much of the teacher model's performance while being smaller and more suitable for real-time and resource-constrained scenarios.

[0074] First, the cloud-based deep learning network model needs to be used as the teacher model to ensure that there is a large enough data set for the teacher model training. These data sets contain key features such as historical electricity consumption data, battery status information, and market electricity prices. The architecture of the student model should be simpler than the teacher model, for example, reducing the number of layers, nodes, or using simpler neural network units. In this embodiment, the algorithm model of the initial deep learning architecture is used for construction, which can be understood as the parameters of the model are all initialization parameters.

[0075] Secondly, in each training iteration, the student model's output should be as close as possible to the teacher model's output, not just matching the final classification or regression results. The distillation objective function typically consists of two parts: the prediction error on the original data and the difference with the teacher model's output. The pretrained dataset is fed into the teacher model to generate its output. These outputs are then used as the training target for the student model, and the student model's parameters are adjusted by minimizing the distillation objective function. This process is repeated until the student model's performance approaches that of the teacher model. After the distillation process, the student model's prediction performance on the validation set is evaluated to ensure that its performance loss is within an acceptable range. Based on the performance evaluation results, the distillation temperature, learning rate, or other hyperparameters may need to be adjusted to further optimize the student model's performance. In addition to distillation, model pruning (removing unimportant neurons or connections) and quantization (reducing the precision of the weight representation) may be used to further reduce model size and improve computational efficiency. The resulting lightweight model is the cloud-based distillation model, which can run efficiently in the cloud and can be updated in real time or periodically to maintain optimal prediction performance.

[0076] Finally, the cloud-based distillation model can be further distributed to edge devices, which use it to make real-time predictions. For example, these predictions can predict the charge and discharge of energy storage power plants, enabling cost-effective energy scheduling. The goal of using distillation to simplify models is to reduce model complexity and resource requirements while maintaining predictive performance, making the model adaptable to diverse computing environments in the cloud and edge.

[0077] Optionally, the edge system can also continuously monitor the operating status of the energy storage power station, including key indicators such as charge and discharge capacity, battery temperature, and SOC / SOH, to ensure the safe operation of the energy storage station during policy execution. The local model at the edge can identify abnormal behavior or equipment failures during charging and discharging operations, such as battery overheating and voltage anomalies. Once an abnormality or violation is detected, the system will immediately trigger an alarm and notify relevant personnel through voice, text messages, and other means to handle it. This helps to promptly identify and resolve problems, ensuring the safe and stable operation of the energy storage power station. The alarm recording and analysis function can also record detailed information for each alarm and compile statistics and analysis on the alarm data, providing strong support for subsequent improvement and optimization.

[0078] As an optional embodiment, Figure 3 is a schematic diagram of a method for determining the remaining charging time according to an optional embodiment of the present invention. Figure 3 As shown, the method includes the following steps:

[0079] S1. First, the energy storage power station data acquisition device deploys an edge computing node at the energy storage power station or local gateway to perform lightweight data cleaning, compression, and feature extraction, and transmit the corresponding data to the cloud in a standard format.

[0080] Based on the characteristics and data volume of the energy storage power station, select edge computing devices with sufficient computing power, storage space, and low power consumption, such as embedded computing units, IoT gateways, or dedicated edge servers. Configure the necessary software environment to ensure that the edge computing nodes can run the relevant programs for data collection, processing, and transmission. Edge computing nodes need to be connected to various sensors and data collection points on the energy storage power station, including battery status (SOC, SOH), voltage, current, temperature, environmental information (such as temperature, humidity), and data interacting with the power grid (such as real-time electricity prices). In order to communicate with different types of devices and systems, edge nodes can support multiple industrial communication protocols, such as Modbus, CAN, IEC61850, OPC-UA, etc., to ensure seamless data collection from the energy storage system. Perform preliminary data cleaning at the edge to remove noise, invalid data, or fill missing values ​​to ensure data quality and integrity. Due to network bandwidth and storage resource limitations, edge computing nodes need to compress cleaned data to reduce transmission data volume and storage requirements. Common compression methods include, but are not limited to, lossless or lossy data compression algorithms, such as Huffman coding, LZ77 (e.g., gzip), or compression techniques specifically designed for time series data. Preliminary feature extraction is performed at the edge to convert raw data into meaningful features, such as time series features of power usage and battery health indicators. This helps reduce data dimensionality and improves cloud-based model training efficiency. Extracted features are normalized to ensure the data is on the same scale, facilitating model training and preventing numerical instability. Normalization methods can include Z-score normalization, min-max normalization, or other normalization techniques. Secure transmission protocols, such as HTTPS and TLS, can also be used to encrypt transmitted data to protect data security and privacy during transmission. A mechanism for handling data transmission anomalies can be designed to buffer and retransmit data when network issues are detected, ensuring data integrity and system reliability. Processed data is converted into a unified, cloud-interpretable standard format, such as JSON, CSV, or a specific API format, to facilitate storage, querying, and analysis on the cloud data platform. By deploying edge computing nodes at the energy storage power station or local gateway to perform lightweight data cleaning, compression, and feature extraction, and securely transmit the processed data to the cloud in a standard format, the energy storage power station intelligent scheduling system can achieve an efficient data flow pipeline.

[0081] S2: The cloud then receives the data uploaded by the edge, standardizes it, and stores it in Clickhouse, a big data storage database. The data uploaded by the edge includes, but is not limited to, historical values ​​of battery voltage and current, battery charge and discharge capacity, battery temperature, and battery SOC / SOH.

[0082] The cloud-based data receiving module first parses the raw data packets uploaded by the edge, extracting historical values ​​such as battery voltage, current, charge / discharge capacity, temperature, SOC (State of Charge), and SOH (State of Health). All data is presented in consistent units, for example, voltage and current are expressed in volts (V) and amperes (A), temperature in degrees Celsius (°C), and SOC and SOH are expressed as percentages. Based on the data's characteristics, numerical scaling can be performed, such as normalizing voltage and current values ​​to a range of 0–1 or appropriately transforming temperature values ​​to fit the input range of the machine learning model. Outliers in the data are detected and addressed to prevent them from negatively impacting model training. Statistical methods, such as the interquartile range (IQR), or machine learning methods can be used for anomaly detection. For missing data, interpolation, forward / backward filling, or predictions based on other data can be used to ensure data continuity and integrity. ClickHouse is a column-based database designed specifically for online analytical processing (OLAP). It's suitable for storing and querying large amounts of time-series data and supports distributed data storage, meaning data can be distributed across multiple servers. This improves data processing speed and enhances the system's fault tolerance. Create a table in ClickHouse and define columns for each type of historical value, such as "timestamp," "battery_voltage," "battery_current," "charge_discharge_energy," "temperature," "SOC," and "SOH." Ensure that the data type of each column matches the standardized data. Insert the standardized data into ClickHouse according to the defined table structure. Based on the frequency and characteristics of the data, adopt appropriate partitioning strategies and compression algorithms to improve data access speed and reduce storage space. Encrypt sensitive data stored in ClickHouse to protect its security. Establish a data backup mechanism and regularly back up important data to prevent data loss. Develop an API or query interface to allow applications and services to access ClickHouse data in real time for immediate analysis and decision-making. Through the above process, the cloud can effectively receive and manage various historical data uploaded by the edge, providing strong data support for the intelligent scheduling of energy storage power plants.

[0083] S3. The cloud system establishes a deep learning network model and uses the distillation method to simplify the model to obtain a cloud distillation model.

[0084] Based on the characteristics of the energy storage scheduling problem, select an appropriate time series analysis model, such as an LSTM (Long Short-Term Memory) network, a GRU (Gated Recurrent Units), or a CNN (Convolutional Neural Network) combined with an RNN (Recurrent Neural Network). The model inputs include historical battery voltage and current values, battery charge and discharge values, battery temperature values, battery SOC / SOH values, and market electricity prices. The output is a predicted charge and discharge value, SOC trend, or abnormal behavior detection results. Initialize the model parameters and define an optimization objective, such as minimizing the root mean square error (RMSE) or cross-entropy loss of the prediction error. Extract a training dataset from the Clickhouse database, including historical electricity usage data, battery status information, and other relevant features. Train the model using backpropagation and an appropriate optimizer (such as Adam or SGD), continuously adjusting parameters to minimize prediction error. Evaluate model performance on a validation set to ensure strong generalization and avoid overfitting.

[0085] Distillation is used to simplify the model. The teacher model is a trained cloud-based deep learning network model with high accuracy and computational intensiveness. The student model architecture is simplified to reduce computational resource requirements, but still requires sufficient expressiveness to capture the learning results of the teacher model. During the distillation process, the output of the remote deep learning network model is made more detailed by adjusting model parameters, allowing the student model to learn more details during training. The distillation objective function is typically defined as consisting of two parts: a loss between the student model's predictions and the actual labels, and a KL divergence loss between the student model's predictions and the teacher model's soft labels. Using the teacher model's soft labels as additional supervision, the student model is trained until its performance approaches that of the teacher model while maintaining low computational complexity. The student model's performance is evaluated on an independent test set to ensure that it meets the expected accuracy requirements. Based on the test results, the student model's structure or parameters may need to be fine-tuned to further optimize its performance. In addition to distillation, model pruning and weight quantization can also be used to further reduce model size and improve its efficiency. The optimized student model is encapsulated as a cloud-deployable service, and a final test is performed on the cloud before model deployment to ensure that the model performs stably in different environments and data sets. After deployment, the performance of the model is continuously monitored to ensure that it can cope with various complex scenarios in actual applications. Based on the model's performance and business needs, the model is updated regularly or on demand. The teacher model can be retrained and then another round of knowledge distillation is performed. The cloud system successfully established a deep learning network model and used the distillation method to effectively simplify the model to generate a cloud-based distillation model. This model can not only accurately predict the charge and discharge capacity and other key states of the energy storage power station, but also run efficiently on resource-constrained edge devices, significantly improving the real-time and economic performance of the energy storage intelligent scheduling system.

[0086] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present invention is not limited by the order of the actions described, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0087] Through the description of the above embodiments, those skilled in the art can clearly understand that the charge and discharge scheduling method for the energy storage power station according to the above embodiment can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a more preferred embodiment. Based on this understanding, the technical solution of the present invention, or the part that contributes to the existing technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0088] According to an embodiment of the present invention, a charge and discharge scheduling system for an energy storage power station is also provided for implementing the above-mentioned charge and discharge scheduling method for an energy storage power station. The charge and discharge system of the energy storage power station includes: a cloud for executing any of the above-mentioned charge and discharge scheduling methods for an energy storage power station; and an edge for monitoring the operating status of the energy storage power station, performing anomaly detection, collecting power consumption data and historical data based on edge computing nodes, and uploading the operating status data, anomaly detection results, power consumption data, and historical data to the cloud.

[0089] Optionally, in the event of a network anomaly in the cloud, the edge is also used to determine a second electricity consumption behavior trend in the area corresponding to the energy storage power station based on a pre-deployed local model using electricity consumption data and historical data; based on the second electricity consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a second prediction result; based on the second prediction result, the charge and discharge of the energy storage power station is scheduled within a preset time period.

[0090] According to an embodiment of the present invention, there is also provided a charge and discharge scheduling device for an energy storage power station for implementing the above-mentioned charge and discharge scheduling method for an energy storage power station. Figure 4 FIG. 1 is a structural block diagram of a charge and discharge scheduling device for an energy storage power station according to an embodiment of the present invention. Figure 4 As shown, the charge and discharge scheduling device of the energy storage power station includes: a first acquisition module 402, a determination module 404, a prediction module 406, a second acquisition module 408 and a scheduling module 410. The charge and discharge scheduling device of the energy storage power station is described below.

[0091] A first acquisition module 402 is configured to acquire power consumption data collected by edge computing nodes of the energy storage power station and historical data corresponding to the energy storage power station;

[0092] A determination module 404, connected to the first acquisition module 402, is configured to determine a first electricity usage behavior trend in a region corresponding to the energy storage power station based on the electricity usage data and historical data using a cloud-based deep learning network model;

[0093] The prediction module 406 is connected to the determination module 404 and is used to predict the charge and discharge amount of the energy storage power station based on the first electricity usage behavior trend to obtain a first prediction result;

[0094] The second acquisition module 408 is connected to the first prediction module 406 and is used to obtain the status detection data of the energy storage power station sent by the edge, wherein the status detection data includes: the operating status data of the energy storage power station and the abnormality detection results;

[0095] The scheduling module 410 is connected to the second acquisition module 408 and is used to schedule charging and discharging of the energy storage power station within a preset time period based on the first prediction result and the status detection data.

[0096] It should be noted that the first acquisition module 402, determination module 404, prediction module 406, second acquisition module 408, and scheduling module 410 described above correspond to steps S202 to S210 in the embodiment. The examples and application scenarios implemented by these modules and corresponding steps are the same, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above modules, as part of the apparatus, can be run in the computer terminal 10 provided in the embodiment.

[0097] An embodiment of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0098] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the method and apparatus for scheduling charging and discharging energy storage power stations in the embodiments of the present invention. The processor executes the software programs and modules stored in the memory to perform various functional applications and data processing, thereby implementing the aforementioned method for scheduling charging and discharging energy storage power stations. The memory can include high-speed random access memory (RAM) and can also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory can further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0099] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain the power consumption data collected by the edge computing node of the energy storage power station and the historical data corresponding to the energy storage power station; based on the power consumption data and the historical data, determine the first power consumption behavior trend of the area corresponding to the energy storage power station through the cloud-based deep learning network model; based on the first power consumption behavior trend, predict the charge and discharge amount of the energy storage power station to obtain a first prediction result; obtain the status detection data of the energy storage power station sent by the edge, wherein the status detection data includes: the operating status data and abnormality detection results of the energy storage power station; based on the first prediction result and the status detection data, schedule the charge and discharge of the energy storage power station within a preset time period.

[0100] Optionally, the processor may also execute program code for the following steps: before determining the first electricity consumption behavior trend of the area corresponding to the energy storage power station through a cloud-based deep learning network model based on electricity consumption data and historical data, the steps include: determining a target function corresponding to the charging amount according to the charging amount of the energy storage power station and a preset loss function; determining a target function corresponding to the discharging amount according to the discharging amount of the energy storage power station and the loss function; determining a target profit function according to the target function corresponding to the charging amount, the target function corresponding to the discharging amount, and the real-time electricity price of the area corresponding to the energy storage power station; and determining a cloud-based deep learning network model according to the target charging amount function, the target discharging amount function, and the target profit function.

[0101] Optionally, the processor may also execute the program code of the following steps: based on electricity consumption data and historical data, determining a first electricity consumption behavior trend in the area corresponding to the energy storage power station through a cloud-based deep learning network model, including: performing standardization processing on the electricity consumption data and historical data to obtain processed electricity consumption data and processed historical data, wherein the standardization processing includes data cleaning, compression, and formatting; and inputting the processed electricity consumption data and processed historical data into the cloud-based deep learning network model to obtain a first electricity consumption behavior trend in the area corresponding to the energy storage power station.

[0102] Optionally, the processor may also execute program code for the following steps: based on the first electricity consumption behavior trend, predicting the charge and discharge amount of the energy storage power station to obtain a first prediction result, including: obtaining prediction features, wherein the prediction features include at least one of the following: average electricity consumption, peak electricity consumption time, and electricity price; based on the first electricity consumption behavior trend, determining a numerical value corresponding to the prediction feature; based on the numerical value corresponding to the prediction feature, predicting the charge and discharge amount of the energy storage power station to obtain a first prediction result, wherein the prediction result includes a target charge amount and a target discharge amount.

[0103] Optionally, the processor may also execute the program code of the following steps: when an abnormality occurs in the network corresponding to the deep learning network model in the cloud, the control edge determines a second electricity consumption behavior trend of the area corresponding to the energy storage power station based on the pre-deployed local model through electricity consumption data and historical data; based on the second electricity consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a second prediction result; based on the second prediction result and status detection data, the charge and discharge scheduling of the energy storage power station is performed within a preset time period.

[0104] Optionally, the above-mentioned processor can also execute the program code of the following steps: the local model is obtained by the following steps: inputting the preset sample electricity consumption data into the cloud-based deep learning network model to obtain a first output result; building an algorithm model based on the initial deep learning architecture to obtain an initial model; training the initial model based on the sample electricity consumption data and the first output result until the output result of the initial model matches the first output result to obtain a target model; sending the target model to the edge and deploying it to obtain a local model, wherein the edge monitors the operating status of the energy storage power station based on the local model.

[0105] By adopting the embodiment of the present invention, a scheme for a charge and discharge scheduling method of an energy storage power station is provided. The power consumption data collected by the edge computing node of the energy storage power station and the historical data corresponding to the energy storage power station are obtained; based on the power consumption data and the historical data, a first power consumption behavior trend of the area corresponding to the energy storage power station is determined through a cloud-based deep learning network model; based on the first power consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a first prediction result; the status detection data of the energy storage power station sent by the edge is obtained, wherein the status detection data includes: the operating status data and abnormality detection results of the energy storage power station; based on the first prediction result and the status detection data, the charge and discharge of the energy storage power station is scheduled within a preset time period, achieving the purpose of predicting the charge and discharge amount of the energy storage power station based on the cloud-edge collaborative deep learning algorithm model, thereby realizing the technical effect of intelligently scheduling the charge and discharge amount of the energy storage power station, and further solving the current technical problem of temporarily adjusting the charge and discharge strategy of the energy storage power station based on the real-time electricity price, resulting in fast charging and discharging of the battery, damaged battery life, and high labor costs.

[0106] A person skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a non-volatile storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0107] The embodiment of the present invention further provides a non-volatile storage medium. Optionally, in this embodiment, the non-volatile storage medium can be used to store program codes executed by the charge and discharge scheduling method for the energy storage power station provided in the embodiment.

[0108] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0109] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: obtaining electricity consumption data collected by the edge computing node of the energy storage power station and historical data corresponding to the energy storage power station; determining a first electricity consumption behavior trend in the area corresponding to the energy storage power station through a cloud-based deep learning network model based on the electricity consumption data and the historical data; predicting the charge and discharge amount of the energy storage power station based on the first electricity consumption behavior trend to obtain a first prediction result; obtaining status detection data of the energy storage power station sent by the edge, wherein the status detection data includes: operating status data and abnormality detection results of the energy storage power station; and scheduling the charge and discharge of the energy storage power station within a preset time period based on the first prediction result and the status detection data.

[0110] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: before determining the first electricity consumption behavior trend of the area corresponding to the energy storage power station through a cloud-based deep learning network model based on electricity consumption data and historical data, including: determining the target function corresponding to the charging amount according to the charging amount of the energy storage power station and a preset loss function; determining the target function corresponding to the discharge amount according to the discharge amount of the energy storage power station and the loss function; determining the target profit function according to the target function corresponding to the charging amount, the target function corresponding to the discharge amount and the real-time electricity price of the area corresponding to the energy storage power station; determining the cloud-based deep learning network model according to the target charging amount function, the target discharge amount function and the target profit function.

[0111] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: based on the electricity consumption data and historical data, determining the first electricity consumption behavior trend of the area corresponding to the energy storage power station through a cloud-based deep learning network model, including: standardizing the electricity consumption data and historical data to obtain processed electricity consumption data and processed historical data, wherein the standardization processing includes data cleaning, compression and formatting; inputting the processed electricity consumption data and processed historical data into the cloud-based deep learning network model to obtain the first electricity consumption behavior trend of the area corresponding to the energy storage power station.

[0112] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: based on the first electricity consumption behavior trend, predicting the charge and discharge amount of the energy storage power station to obtain a first prediction result, including: obtaining a prediction feature, wherein the prediction feature includes at least one of the following: average electricity consumption, peak electricity consumption time and electricity price; based on the first electricity consumption behavior trend, determining a numerical value corresponding to the prediction feature; based on the numerical value corresponding to the prediction feature, predicting the charge and discharge amount of the energy storage power station to obtain a first prediction result, wherein the prediction result includes a target charge amount and a target discharge amount.

[0113] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: when an abnormality occurs in the network corresponding to the deep learning network model in the cloud, the control edge determines a second power consumption behavior trend of the area corresponding to the energy storage power station based on the pre-deployed local model through power consumption data and historical data; based on the second power consumption behavior trend, the charge and discharge amount of the energy storage power station is predicted to obtain a second prediction result; based on the second prediction result and status detection data, the charge and discharge scheduling of the energy storage power station is performed within a preset time period.

[0114] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: the local model is obtained by the following steps: inputting preset sample electricity consumption data into the cloud-based deep learning network model to obtain a first output result; building an algorithm model based on the initial deep learning architecture to obtain an initial model; training the initial model based on the sample electricity consumption data and the first output result until the output result of the initial model matches the first output result to obtain a target model; sending the target model to the edge and deploying it to obtain a local model, wherein the edge monitors the operating status of the energy storage power station based on the local model.

[0115] An embodiment of the present invention also provides a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can achieve: obtaining electricity consumption data collected by the edge computing node of the energy storage power station and historical data corresponding to the energy storage power station; based on the electricity consumption data and historical data, determining a first electricity consumption behavior trend in the area corresponding to the energy storage power station through a cloud-based deep learning network model; based on the first electricity consumption behavior trend, predicting the charge and discharge amount of the energy storage power station to obtain a first prediction result; obtaining status detection data of the energy storage power station sent by the edge, wherein the status detection data includes: operating status data and abnormality detection results of the energy storage power station; based on the first prediction result and the status detection data, scheduling the charge and discharge of the energy storage power station within a preset time period.

[0116] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0117] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0118] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0119] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0120] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0121] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, and other media that can store program code.

[0122] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for charging and discharging scheduling of an energy storage power station, characterized in that: include: Obtaining electricity consumption data collected by the edge computing nodes of the energy storage power station and historical data corresponding to the energy storage power station; Based on the electricity consumption data and the historical data, determining a first electricity consumption behavior trend in the area corresponding to the energy storage power station through a cloud-based deep learning network model; Based on the first electricity usage behavior trend, predicting the charge and discharge amount of the energy storage power station to obtain a first prediction result; Acquire status detection data of the energy storage power station sent by the edge, wherein the status detection data includes: operating status data and abnormality detection results of the energy storage power station; Based on the first prediction result and the status detection data, charging and discharging scheduling is performed on the energy storage power station within a preset time period.

2. The method according to claim 1, characterized in that Before determining the first electricity usage behavior trend of the area corresponding to the energy storage power station based on the electricity usage data and the historical data through a cloud-based deep learning network model, the method includes: Determining an objective function corresponding to the charging amount based on the charging amount of the energy storage power station and a preset loss function; Determining an objective function corresponding to the discharge amount according to the discharge amount of the energy storage power station and the loss function; Determining a target revenue function based on the target function corresponding to the charging amount, the target function corresponding to the discharging amount, and the real-time electricity price in the area corresponding to the energy storage power station; The cloud-based deep learning network model is determined according to the target charge capacity function, the target discharge capacity function, and the target profit function.

3. The method according to claim 1, characterized in that The determining, based on the electricity usage data and the historical data, through a cloud-based deep learning network model, a first electricity usage behavior trend in the area corresponding to the energy storage power station includes: Performing standardization processing on the electricity usage data and the historical data to obtain processed electricity usage data and processed historical data, wherein the standardization processing includes data cleaning, compression, and formatting; The processed electricity usage data and the processed historical data are input into the cloud-based deep learning network model to obtain a first electricity usage behavior trend in the area corresponding to the energy storage power station.

4. The method according to claim 1, wherein The step of predicting the charge and discharge amount of the energy storage power station based on the first electricity usage behavior trend to obtain a first prediction result includes: Obtaining prediction features, wherein the prediction features include at least one of the following: average power consumption, peak power consumption time, and power price; determining a value corresponding to the prediction feature based on the first electricity usage behavior trend; Based on the numerical value corresponding to the prediction feature, the charge and discharge amount of the energy storage power station is predicted to obtain the first prediction result, wherein the first prediction result includes a target charge amount and a target discharge amount.

5. The method according to claim 1, wherein Also includes: When an abnormality occurs in the network corresponding to the cloud-based deep learning network model, controlling the edge to determine a second power consumption behavior trend of the area corresponding to the energy storage power station based on the power consumption data and the historical data based on a pre-deployed local model; Based on the second electricity usage behavior trend, predicting the charge and discharge amount of the energy storage power station to obtain a second prediction result; Based on the second prediction result and the status detection data, charging and discharging scheduling is performed on the energy storage power station within the preset time period.

6. The method according to claim 5, characterized in that The local model is obtained by the following steps: Inputting the preset sample electricity consumption data into the cloud-based deep learning network model to obtain a first output result; Build an algorithm model based on the initial deep learning architecture to obtain an initial model; Based on the sample electricity usage data and the first output result, the initial model is trained until the output result of the initial model matches the first output result, thereby obtaining a target model; The target model is sent to the edge and deployed to obtain the local model.

7. A charging and discharging scheduling system for an energy storage power station, characterized in that: include: A cloud, configured to execute the charge and discharge scheduling method for the energy storage power station according to any one of claims 1 to 6; The edge is used to monitor the operating status of the energy storage power station, perform anomaly detection, collect power consumption data and historical data based on the edge computing node, and upload the operating status data, anomaly detection results, the power consumption data and the historical data to the cloud.

8. The system according to claim 7, characterized in that Also includes: In the event of a network anomaly in the cloud, the edge is further configured to determine, based on a pre-deployed local model and using the power consumption data and the historical data, a second power consumption behavior trend in the area corresponding to the energy storage power station; Based on the second electricity usage behavior trend, predicting the charge and discharge amount of the energy storage power station to obtain a second prediction result; Based on the second prediction result, charging and discharging scheduling is performed on the energy storage power station within the preset time period.

9. A charging and discharging scheduling device for an energy storage power station, characterized in that: include: A first acquisition module is used to acquire power consumption data collected by the edge computing node of the energy storage power station and historical data corresponding to the energy storage power station; a determination module, configured to determine, based on the electricity consumption data and the historical data, a first electricity consumption behavior trend in the area corresponding to the energy storage power station through a cloud-based deep learning network model; a prediction module, configured to predict the charge and discharge amount of the energy storage power station based on the first electricity usage behavior trend to obtain a first prediction result; A second acquisition module is configured to acquire status detection data of the energy storage power station sent by an edge end, wherein the status detection data includes: operating status data and abnormality detection results of the energy storage power station; A scheduling module is used to schedule charging and discharging of the energy storage power station within a preset time period based on the first prediction result and the status detection data.

10. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the charging and discharging scheduling method of the energy storage power station according to any one of claims 1 to 6.

11. A computer device, characterized in that: include: memory and processor, The memory stores a computer program; The processor is configured to execute a computer program stored in the memory, and when the computer program is run, the processor executes the charge and discharge scheduling method for an energy storage power station according to any one of claims 1 to 6.

12. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the charging and discharging scheduling method of the energy storage power station according to any one of claims 1 to 6 is implemented.

Images