Data management method, device and equipment based on cloud edge-end architecture, and storage medium

Abstract

This invention relates to the technical field of data management, and discloses a data management method, device, equipment, and storage medium based on a cloud-edge-device architecture. It involves real-time monitoring of battery operation based on an edge-side deployment unit, parsing real-time operating parameters using an algorithm model based on the edge-side deployment unit to generate battery operation control strategies, and performing multi-dimensional calculations on the real-time operating parameters to obtain multi-dimensional computational features. The invention also stores the real-time operating parameters and multi-dimensional computational features of batteries from multiple power stations on the cloud side, obtaining datasets for each power station. A comprehensive analysis of the real-time operating parameters and multi-dimensional computational features of each power station's dataset is then performed. Based on the analysis results, the algorithm model deployed on the edge side is trained and optimized, and the optimized algorithm model is then distributed to the edge side. This invention saves cloud computing power by calculating multi-dimensional computational features locally, and simultaneously optimizes the edge-side algorithm model through cloud-based dataset analysis to continuously optimize the operation control of the power stations.

Description

Technical Field

[0001] This invention relates to the technical field of data management, and in particular to a data management method, apparatus, device, and storage medium based on a cloud-edge-device architecture. Background Technology

[0002] With the large-scale development of renewable energy, energy storage systems, as a key means to solve the problems of intermittency and volatility of renewable energy, are constantly expanding in scale. A large number of decentralized energy storage power stations and battery devices are connected to the grid, making the management of energy storage systems increasingly complex. A large energy storage project includes multiple distributed power stations, each of which has a large number of battery packs. Traditional data management involves collecting the operating data of the battery packs and transmitting it to the cloud platform for unified analysis. This analysis method places a heavy burden on the computing power of the cloud platform, making it difficult for the cloud platform to cope with such a large amount of decentralized data. Summary of the Invention

[0003] The purpose of this invention is to provide a data management method, apparatus, device, and storage medium based on a cloud-edge-device architecture, aiming to solve the problem of managing complex battery packs in the prior art.

[0004] The present invention is implemented as follows: Firstly, the present invention provides a data management method based on a cloud-edge-device architecture, comprising: The battery's operating parameters are collected in real time by the edge deployment unit to obtain real-time operating parameters, and the real-time operating parameters are preliminarily processed and cached locally. The algorithm model based on the edge deployment unit performs multi-dimensional calculations on the real-time operating parameters to obtain multi-dimensional calculation features; The energy storage big data platform built on the cloud side stores the real-time operating parameters and multi-dimensional computing characteristics of batteries in multiple power stations, and obtains the datasets of each power station. A comprehensive analysis of the real-time operating parameters and multi-dimensional computational features of the datasets of each power station is conducted. Based on the analysis results, the algorithm model deployed on the edge is trained and optimized, and the optimized algorithm model is then distributed to the edge.

[0005] Secondly, the present invention provides a data management device based on a cloud-edge-device architecture, used to implement the data management method based on a cloud-edge-device architecture as described in any one of the first aspects, comprising: The end-side processing module is used to collect the battery's operating parameters in real time based on the end-side deployment unit, obtain the real-time operating parameters, and perform preliminary processing and local caching of the real-time operating parameters. The edge processing module is used to perform multi-dimensional calculations on the real-time running parameters based on the algorithm model of the edge deployment unit to obtain multi-dimensional calculation features; The cloud-side processing module is used to store the real-time operating parameters and multi-dimensional computational characteristics of batteries from multiple power stations on the cloud-based energy storage big data platform, thereby obtaining the datasets of each power station. The algorithm optimization module is used to perform a comprehensive analysis of the real-time operating parameters and multi-dimensional computational features of the datasets of each power station. Based on the analysis results, it trains and optimizes the algorithm model deployed on the side and distributes the optimized algorithm model to the side so that the side deployment unit can parse the real-time operating parameters and generate an operation control strategy.

[0006] Thirdly, the present invention provides a data management device based on a cloud-edge-device architecture, including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements a data management method based on a cloud-edge-device architecture as described in any one of the first aspects.

[0007] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is run by a processor, the processor causes the processor to perform a data management method based on a cloud-edge-device architecture as described in any one of the first aspects.

[0008] This invention provides a data management method based on a cloud-edge-device architecture, which has the following beneficial effects: This invention enables real-time data acquisition and preliminary processing at the edge, rapid parsing and calculation at the edge, and centralized storage at the cloud, thereby improving data processing efficiency and quality. The edge generates operation control strategies based on the parsing results, which can timely and accurately regulate battery operation and ensure battery safety and stability. After acquisition at the edge, real-time operation data is calculated to generate multi-dimensional computational features, saving computing power for the cloud. The algorithm model is then optimized at the cloud based on the real-time operation data and multi-dimensional computational features, enabling the cloud to cope with the data management pressure of multiple power stations and multiple battery packs. Attached Figure Description

[0009] Figure 1 This is a schematic diagram illustrating the steps of a data management method based on a cloud-edge-device architecture provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a data management device based on a cloud-edge-device architecture provided in an embodiment of the present invention. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0011] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0012] Reference Figure 1 , Figure 2 The diagram shows a preferred embodiment of the present invention.

[0013] In a first aspect, the present invention provides a data management method based on a cloud-edge-device architecture, comprising: S1: Real-time acquisition of battery operating parameters based on the end-side deployment unit, obtaining real-time operating parameters, and performing preliminary processing and local caching of the real-time operating parameters; S2: The algorithm model based on the edge deployment unit performs multi-dimensional calculations on the real-time operating parameters to obtain multi-dimensional calculation features; S3: A cloud-based energy storage big data platform that stores real-time operating parameters and multi-dimensional computational features of batteries from multiple power stations to obtain datasets for each power station. S4: Perform a comprehensive analysis of the real-time operating parameters and multi-dimensional computational features of the datasets of each power station, train and optimize the algorithm model deployed on the side based on the analysis results, and then distribute the optimized algorithm model to the side.

[0014] Specifically, in step S1 of the embodiment provided by the present invention, the battery's operating parameters, such as voltage, current, temperature, SOC (state of charge), and SOH (state of health), are collected by a data acquisition management unit or smart gateway pre-deployed on the end side to obtain real-time operating parameters. The data acquisition management unit and smart gateway have data acquisition function interfaces and can be connected to various sensors of the battery to read the relevant parameters of the battery in real time at a certain sampling frequency (e.g., once per second).

[0015] More specifically, operating parameters such as voltage, current, temperature, SOC, and SOH are key indicators reflecting the battery's operating status and performance. Real-time acquisition of these parameters allows for timely understanding of the battery's working condition, providing basic data for subsequent analysis, control, and decision-making. For example, battery voltage can reflect the battery's charge level and health status; excessively high temperatures can affect battery life and even cause safety issues, requiring real-time monitoring.

[0016] More specifically, the collected real-time operating parameters are filtered, verified, and converted in format to complete the preliminary processing of the real-time operating parameters. Filtering usually uses digital filtering algorithms (such as mean filtering and median filtering) to remove noise interference from the signal; verification checks the accuracy of the data through preset rules (such as the reasonable range of parameter values), and if the data exceeds the reasonable range, it is marked or removed; format conversion converts the collected data into a unified format that is easy for subsequent processing and transmission (such as JSON format).

[0017] More specifically, during the data acquisition process, the acquired signals contain noise due to environmental interference, sensor noise, and other factors. Filtering can improve data quality, making subsequent analysis and decision-making more accurate and reliable, ensuring data accuracy and reliability, and avoiding erroneous decisions caused by incorrect or abnormal data. For example, if the acquired voltage value exceeds the normal operating voltage range of the battery, it may be due to sensor failure or a problem in the acquisition process. Verification can promptly detect and handle these anomalies. Different sensors and devices can output data in different formats. A unified data format can facilitate data processing and analysis on both the edge and cloud sides, improving system compatibility and scalability.

[0018] More specifically, the pre-processed real-time operating parameters are temporarily cached locally, while simultaneously being transmitted to the edge deployment unit. The local cache can use local storage devices (such as hard drives or flash memory) to store data within a certain time period in chronological order; data transmission is performed by sending data to the edge via a network interface (such as Ethernet or wireless communication).

[0019] More specifically, when the network is unstable or the edge device malfunctions, the locally cached data can avoid data loss. At the same time, the local cache can also perform some preprocessing and aggregation of data, reducing the amount of data transmitted to the edge and reducing the pressure on network bandwidth. The edge deployment unit has more powerful computing and analysis capabilities, and the data transmitted to the edge can be analyzed and calculated more deeply to generate battery operation control strategies and multi-dimensional computing characteristics, thereby realizing real-time monitoring and control of the battery.

[0020] Specifically, in step S2 of the embodiment provided by the present invention, the real-time operating parameters transmitted from the end side are analyzed based on the operation control algorithm model of the energy management device and local controller of the EMS (Energy Management System) pre-deployed on the side. The operation control algorithm model generally uses data analysis and pattern recognition technology, including classification and clustering algorithms in machine learning, traditional threshold judgment methods, etc. For example, by comparing real-time voltage, current, temperature and other parameters with preset standard parameter ranges or historical data patterns, the operating status characteristics of the battery at the current moment are identified, such as whether the battery is charging, discharging or idle, and whether there is abnormal heating.

[0021] More specifically, the real-time operating parameters collected at the edge are only raw data. They need to be analyzed to be transformed into meaningful operational features. These features are important bases for subsequent operational control strategies. For example, only by accurately identifying the charging and discharging state of the battery can the charging or discharging power, time and other parameters be adjusted reasonably to ensure the safe and efficient operation of the battery. At the same time, timely detection of abnormal operating features can provide early warnings and take measures to avoid battery failures or safety accidents.

[0022] More specifically, based on the analyzed operational characteristics, corresponding operational control strategies are scheduled from a pre-defined control strategy library. The control strategy library stores a series of predefined strategies for different operational characteristics. For example, when the battery temperature is too high, the corresponding strategy is to reduce the charging or discharging power or activate the heat dissipation device; when the battery SOC reaches the set upper or lower limit, the strategy is to stop charging or discharging. After the side device matches the corresponding strategy according to the current operational characteristics, it sends it to the battery management system for execution.

[0023] More specifically, different operating states require different control measures to ensure battery performance and safety. By establishing a control strategy library and scheduling it according to real-time operating characteristics, intelligent control of the battery can be achieved. This approach can quickly respond to changes in battery operating states, adjust control strategies in a timely manner, avoid the lag and inaccuracy of manual intervention, and improve battery management efficiency and reliability.

[0024] More specifically, based on a multi-dimensional computing algorithm model pre-deployed on the edge, multiple computing tasks are performed on the real-time operating parameters. These computing tasks include statistical analysis, early warning and alarm, feature extraction, fault diagnosis, and trend analysis.

[0025] More specifically, the statistical analysis task calculates statistical quantities such as the average, maximum, minimum, and standard deviation of parameters to assess the stability and performance indicators of battery operation; the early warning and alarm task monitors real-time operating parameters according to preset thresholds and rules, and issues early warning or alarm signals when parameters exceed the normal range; the feature extraction task extracts more representative and discriminative features from the original operating parameters, such as the characteristics of the battery's charge and discharge curves and temperature change rate; the fault diagnosis task uses fault diagnosis algorithms to determine whether the battery has a fault based on abnormal patterns of operating parameters and to identify the type of fault; the trend analysis task predicts the future operating status and performance change trends of the battery through the analysis of historical and real-time operating parameters. The results obtained from each calculation task are collectively used as multi-dimensional computational features.

[0026] More specifically, a single real-time operating parameter can only reflect one aspect of battery operation. Multidimensional computing allows for a comprehensive evaluation and analysis of the battery from multiple perspectives. Statistical analysis helps to understand the overall battery operation; early warning and alarm tasks can promptly identify potential safety hazards; feature extraction can provide more effective data for subsequent fault diagnosis and trend analysis; fault diagnosis tasks can promptly locate and handle battery faults, reducing maintenance costs and downtime; trend analysis can plan battery maintenance and replacement in advance, improving the reliability and economy of the energy storage system. At the same time, transmitting these multidimensional computing features to the cloud for further storage and analysis can provide richer information for the optimization and management of the entire energy storage system.

[0027] Specifically, in step S3 of the embodiment provided by the present invention, the energy storage big data platform built on the cloud side is usually equipped with a dedicated data receiving interface. This interface supports multiple communication protocols (such as HTTP, MQTT, etc.) and can establish a stable network connection with the edge deployment unit. The edge side transmits the processed real-time battery operating parameters and multi-dimensional calculation characteristics of multiple power stations to the cloud side through the network according to the agreed protocol and format. The cloud side's data receiving module will continuously listen to the network port. Once the data is received, it will perform preliminary checks and verifications to ensure the integrity and accuracy of the data.

[0028] More specifically, multiple power stations are distributed in different geographical locations, generating a huge amount of scattered battery data. As the centralized storage and management center for the data, the cloud needs to collect the data from each power station in a unified manner for subsequent analysis and processing. By establishing a stable data receiving mechanism, it can be ensured that the data can be aggregated to the cloud platform in a timely and accurate manner, providing a foundation for subsequent data processing and analysis.

[0029] More specifically, after receiving real-time operating parameters and multi-dimensional calculation features, the cloud platform will tag the received data. The tagging process usually generates corresponding information tags for each piece of data based on information such as the source of the data (e.g., the specific power station name, battery number), timestamp, and data type (e.g., real-time operating parameters such as voltage, current, and temperature, as well as multi-dimensional calculation features such as statistical analysis results and fault diagnosis results). These tags can be attached to the data record in the form of metadata to facilitate subsequent data management and querying.

[0030] More specifically, data from multiple power plants are mixed together, and without effective labeling, it is difficult to distinguish and manage the data. Information tags can provide clear identification for the data, making it more traceable and understandable. For example, when conducting data analysis, data from specific power plants or specific time periods can be filtered out based on tags, improving the targeting and efficiency of the analysis. At the same time, tags also help with data classification and archiving, facilitating subsequent data mining and utilization.

[0031] More specifically, based on the information tags, the real-time operating parameters and multi-dimensional computing features are arranged in an orderly manner. The cloud platform will group the data according to the power station's identifier, and group the data belonging to the same power station together. Then, within each power station's data group, the data is sorted in chronological order to form an ordered dataset. This dataset can be stored in a cloud-based database (such as a relational database like MySQL, or a non-relational database like MongoDB) for subsequent storage and querying.

[0032] More specifically, grouping and arranging data according to power stations can clearly show the battery operation status of each power station. An ordered dataset facilitates statistical analysis, trend prediction, and other operations. For example, by analyzing data from the same power station at different time periods, we can understand the changing trend of battery performance at that power station. By comparing datasets from different power stations, we can evaluate the operating efficiency and management level of each power station. At the same time, an ordered dataset is also conducive to data storage and management, improving data access speed and query efficiency.

[0033] Specifically, in step S4 of the embodiment provided by the present invention, the datasets of each power station are calculated in the energy storage big data platform to calculate the potential feedback information of the real-time operating parameters and multi-dimensional computing characteristics of each power station. The health status of each power station is obtained through calculation, thereby evaluating the value of the operation control strategy executed by the power station at each time.

[0034] More specifically, the operation control strategy is obtained by analyzing real-time operating parameters using an algorithm model deployed on the side. By evaluating the value of the operation control strategy, the defects of the algorithm model can be identified, and the algorithm model can be retrained and deployed to optimize the control operation of the power plant.

[0035] Preferably, the steps of collecting real-time battery operating parameters based on the end-side deployment unit, obtaining real-time operating parameters, and performing preliminary processing and local caching of the real-time operating parameters include: S11: By pre-deploying a data acquisition and management unit or smart gateway on the end side, the operating parameters of the battery, such as voltage, current, temperature, SOC, and SOH, are collected to obtain real-time operating parameters; S12: Filter, verify, and convert the format of the real-time operating parameters to complete the preliminary processing of the real-time operating parameters; S13: Temporarily cache the pre-processed real-time operating parameters locally, and simultaneously transmit the pre-processed real-time operating parameters to the edge deployment unit.

[0036] Specifically, a data acquisition and management unit or smart gateway is pre-installed on the end side. These devices have interfaces for connecting to various battery sensors, such as RS-485, CAN bus, and other communication interfaces, to the battery's voltage, current, and temperature sensors. The data acquisition and management unit or smart gateway periodically reads the battery voltage, current, and temperature values ​​measured by the sensors according to a set sampling frequency (such as once per second or once per minute, the specific frequency can be adjusted according to actual needs). At the same time, it obtains the battery's SOC (State of Charge) and SOH (State of Health) information through specific algorithms or communication with the battery management system (BMS), thereby obtaining real-time operating parameters.

[0037] More specifically, battery voltage, current, temperature, SOC, and SOH are key indicators reflecting its operating status and performance. Real-time acquisition of these parameters allows for timely understanding of the battery's working condition, providing fundamental data for subsequent analysis and control. For example, abnormal voltage indicates a battery malfunction or abnormal charging / discharging; excessively high temperatures can affect battery life and safety, requiring real-time monitoring. These real-time operating parameters form the basis of the entire data management system, and subsequent preliminary processing, analysis, and control strategy formulation all depend on accurate real-time data.

[0038] More specifically, digital filtering algorithms are used to process the acquired real-time operating parameters. Common filtering algorithms include mean filtering and median filtering. For example, mean filtering averages the data within a certain time window to smooth the data curve and reduce noise interference. Real-time operating parameters are verified according to preset rules. For example, the reasonable range for voltage is set to 2V - 4.2V. If the acquired voltage value exceeds this range, the data is considered to be abnormal and needs further inspection or marking. The acquired real-time operating parameters are converted into a unified format that is easy for subsequent processing and transmission. For example, the binary data output by the sensor is converted into decimal values ​​and encapsulated according to standard formats such as JSON or XML.

[0039] More specifically, filtering can remove noise and interference introduced during the acquisition process, making the data smoother and more accurate. Verification can detect abnormal data and prevent erroneous data from affecting subsequent analysis and decision-making. A unified data format facilitates data interaction and sharing between different devices and systems, improving system compatibility and scalability. For example, edge deployment units can more easily receive and process data that has undergone format conversion.

[0040] More specifically, the local storage media (such as hard drives, flash memory, etc.) of the edge devices are used to temporarily store the pre-processed real-time operating parameters. The data can be stored in files or databases in chronological order, with a certain cache capacity and storage time set. When the cache reaches its limit, the data is replaced according to the first-in-first-out principle. The pre-processed real-time operating parameters are transmitted to the edge deployment unit through network communication interfaces (such as Ethernet, Wi-Fi, 4G / 5G, etc.). During the transmission process, encryption and compression technologies can be used to ensure data security and transmission efficiency.

[0041] More specifically, local caching can temporarily store data during network failures or edge device malfunctions to prevent data loss. Once the network returns to normal, the cached data can be transmitted back to the edge to ensure data integrity. Edge deployment units typically have stronger computing and analysis capabilities. Transmitting pre-processed data to the edge for further processing can reduce the computing burden on edge devices and make full use of edge resources. In addition, separating local caching and transmission of data can optimize network bandwidth usage and improve data transmission efficiency.

[0042] Preferably, the steps of analyzing the real-time operating parameters based on the algorithm model of the edge-deployment unit to generate the battery operation control strategy, and performing multi-dimensional calculations on the real-time operating parameters to obtain multi-dimensional computational features include: S21: Based on the operation control algorithm model of the EMS energy management device and local controller pre-deployed on the side, the real-time operation parameters are analyzed to identify the actual operation characteristics of the battery at the current moment. S22: Based on the actual operating characteristics, schedule the corresponding operating control strategy from the preset control strategy library to execute the operating control strategy to control the operation of the battery; S23: Based on the multidimensional computing algorithm model pre-deployed on the edge, perform multiple computing tasks on the real-time running parameters to obtain the computing features of each computing task, which together serve as multidimensional computing features.

[0043] The computational tasks include statistical analysis, early warning and alarm, feature extraction, fault diagnosis, and trend analysis.

[0044] Specifically, the side-mounted EMS energy management device and the local controller receive real-time operating parameters from the receiving end. These parameters include battery voltage, current, temperature, SOC, SOH, etc. Using a pre-deployed operation control algorithm model, machine learning algorithms (such as decision trees, neural networks, etc.) or traditional rule matching methods are used to analyze the real-time operating parameters. For example, the decision tree algorithm determines whether the battery is currently charging, discharging, or idle based on the values ​​of voltage, current, and SOC. By comparing the real-time temperature with a preset temperature threshold, it identifies whether the battery is overheating. After comprehensive analysis, the actual operating characteristics of the battery at the current moment are identified, such as the battery's charging and discharging power, health status, and whether there are any abnormalities.

[0045] More specifically, real-time understanding of the battery's operating status is fundamental to effective battery management and control. By analyzing real-time operating parameters, abnormal battery conditions, such as overcharging, over-discharging, and overheating, can be detected in a timely manner, allowing for appropriate measures to be taken to ensure battery safety and performance. Accurate real-time operating characteristics are a prerequisite for generating suitable operating control strategies. Only by clearly understanding the current state of the battery can the most appropriate strategy be selected from the preset control strategy library to control the battery.

[0046] More specifically, based on the identified operational characteristics, the system searches for the corresponding operational control strategy in a pre-defined control strategy library. This library stores a series of predefined strategies for different operational characteristics. For example, when the battery SOC is too high, the corresponding strategy is to stop charging; when the battery temperature is too high, the strategy is to reduce the charging and discharging power and activate the heat dissipation device. The matched operational control strategy is then sent to the battery management system (BMS) or related control devices, which then execute the corresponding operations to control the battery's operation.

[0047] More specifically, different operating states require different control measures to ensure battery safety and performance. By scheduling appropriate strategies from the control strategy library, the battery's operating parameters can be adjusted in a timely manner to avoid problems such as overcharging, over-discharging, and overheating, thereby extending the battery's lifespan, achieving automated control of the battery, reducing manual intervention, improving the operating efficiency and reliability of the energy storage system, and enabling rapid response and corresponding measures when the battery's operating state changes to ensure the stable operation of the system.

[0048] More specifically, statistical analysis is performed on real-time operating parameters, calculating statistical quantities such as average, maximum, minimum, and standard deviation. For example, the average and standard deviation of battery voltage over a period of time are calculated to assess battery voltage stability. Real-time operating parameters are monitored according to preset thresholds and rules. When parameters exceed normal ranges, early warning or alarm signals are triggered. For example, a high-temperature warning is issued when the battery temperature exceeds a set safety threshold. More representative and discriminative features are extracted from the original operating parameters, such as battery charge / discharge curve characteristics and temperature change rate. Signal processing techniques (such as wavelet transform) can be used to process the data and extract key features.

[0049] More specifically, fault diagnosis algorithms are used to determine whether a battery is faulty based on abnormal patterns in operating parameters, and to identify the type of fault. For example, by analyzing the change patterns of battery voltage and current, it can be determined whether a short circuit fault exists in the battery. By analyzing historical and real-time operating parameters, time series analysis methods (such as the ARIMA model) are used to predict the future operating status and performance change trends of the battery. For example, the SOC change trend of the battery can be predicted so as to make charging and discharging plans in advance.

[0050] More specifically, a single real-time operating parameter can only reflect one aspect of battery operation. Multidimensional computing allows for a comprehensive evaluation and analysis of the battery from multiple perspectives. Statistical analysis helps to understand the overall battery operation; early warning and alarm tasks can promptly identify potential safety hazards; feature extraction can provide more effective data for subsequent fault diagnosis and trend analysis; fault diagnosis tasks can promptly locate and handle battery faults, reducing maintenance costs and downtime; trend analysis can plan battery maintenance and replacement in advance, improving the reliability and economy of the energy storage system. Multidimensional computing features can be transmitted to the cloud for further storage and analysis, providing richer information for the optimization and management of the entire energy storage system. The cloud can then conduct deeper data mining and decision support based on these multidimensional computing features.

[0051] Preferably, the steps for storing real-time operating parameters and multi-dimensional computational characteristics of batteries from multiple power stations and obtaining datasets for each power station based on a cloud-based energy storage big data platform include: S31: A cloud-based energy storage big data platform that receives real-time operating parameters and multi-dimensional computational characteristics of batteries from multiple power stations. S32: Mark the received real-time operating parameters and multi-dimensional computation features to generate corresponding information tags for the real-time operating parameters and multi-dimensional computation features; S33: Arrange the real-time operating parameters and the multi-dimensional calculation features in an orderly manner according to the information tags to obtain the dataset of the power station to which the battery belongs.

[0052] Specifically, the cloud-based energy storage big data platform is equipped with multiple network communication interfaces, supporting common network protocols such as TCP / IP, HTTP, and MQTT to adapt to the data transmission needs of different edge devices. These interfaces continuously listen to specific ports, waiting for edge devices to send data. When edge devices upload real-time battery operating parameters and multi-dimensional calculation characteristics from multiple power stations, the cloud-based platform's receiving module receives the data and performs preliminary verification, checking the data's integrity (e.g., whether the data length and format meet predetermined requirements) and accuracy (e.g., whether the checksum is correct). To ensure the stability of data reception, the cloud-based platform sets up a data buffer to temporarily store the received data. When the data in the buffer reaches a certain quantity or a certain time interval, the data is then processed uniformly.

[0053] More specifically, multiple power stations are distributed in different geographical locations. The battery data of each power station needs to be centrally managed and analyzed in the cloud. The cloud platform, as the core storage and processing center for the data, receives data from each power station, which is the basis for subsequent data processing and mining. Data verification is performed during the receiving process to promptly detect and handle data errors or losses that may occur during transmission, ensuring the data quality for subsequent analysis and application.

[0054] More specifically, detailed tagging rules are formulated based on information such as the source, type, and time of the data. For example, the data source can be tagged according to the name and number of the power station; the data type can be divided into real-time operating parameters such as voltage, current, and temperature, as well as multi-dimensional calculation features such as statistical analysis values ​​and fault diagnosis results; the time information can be accurate to the specific year, month, day, hour, minute, and second. The cloud platform uses preset scripts or programs to automatically add corresponding information tags to each received real-time operating parameter and multi-dimensional calculation feature. These tags can be stored in the data records in the form of metadata, which facilitates subsequent data retrieval and management. As the data is continuously updated and business needs change, the tagging rules are regularly evaluated and adjusted to ensure the accuracy and effectiveness of the tags. At the same time, the tagged data is regularly checked and the tag information is updated in a timely manner.

[0055] More specifically, information tags provide a clear identifier for each piece of data, making the data traceable. Tags can quickly locate the source, generation time and data type of data, facilitating data auditing and quality control. Tags help classify and index massive amounts of data, improving data management efficiency. When querying and analyzing data, tags can be used to quickly filter out data that meets the criteria, reducing data processing time and costs.

[0056] More specifically, based on the power station identifier in the information tag, the received data is grouped according to different power stations, and the data of each power station forms an independent set, which facilitates subsequent individual analysis and management. Within the data set of each power station, the real-time operating parameters and multi-dimensional calculation characteristics are sorted according to the time tag to form a data sequence arranged in chronological order. This can be achieved by using the database sorting function or by writing a sorting algorithm. The sorted data is stored in a cloud-side database, such as a relational database (e.g., MySQL, PostgreSQL) or a non-relational database (e.g., MongoDB, HBase). At the same time, a corresponding index is created for each dataset to improve the data query speed.

[0057] More specifically, ordered data can clearly show the changes in battery data for each power station over time, ensuring data integrity and consistency. This helps analyze battery operating trends and performance changes, promptly identify potential problems, and facilitate various data analysis and mining tasks, such as statistical analysis and machine learning modeling. Through ordered data, patterns and regularities can be extracted more accurately, providing strong support for the optimization and decision-making of energy storage systems.

[0058] Preferably, the method further includes: performing an overall analysis of the datasets of each power station, training and optimizing the algorithm model deployed on the side based on the analysis results, and distributing the optimized algorithm model to the side.

[0059] The steps for conducting a comprehensive analysis of the datasets from each power station and training and optimizing the algorithm model deployed at the periphery based on the analysis results include: S41: Based on the real-time operating parameters and multi-dimensional computational features in each dataset, feature mining of long-cycle historical data is performed on each power station to obtain the long-cycle battery operation characteristics of each power station. S42: Based on the operating characteristics of the long-cycle battery, divide the time intervals of each power station and evaluate its operating value to obtain the operating value parameters of each power station in each time interval. S43: Correct the pre-prepared training data according to the operational value parameters, and use the corrected training data to train and optimize the operational control algorithm model deployed in each power station.

[0060] Specifically, long-term real-time operating parameters (such as battery voltage, current, temperature, SOC, SOH, etc.) and multi-dimensional computational features (such as statistical analysis results, fault diagnosis results, etc.) are extracted from the datasets of various power plants stored in the cloud. Data from the same power plant over a relatively long period (such as one year, three years, etc.) are integrated to form a dataset that can be mined. The integrated data is then cleaned to remove noisy data, outliers, and missing values. Interpolation methods (such as linear interpolation and spline interpolation) can be used to handle missing values, and statistical methods (such as methods based on standard deviation) can be used to identify and remove outliers.

[0061] More specifically, data mining algorithms, such as association rule mining, cluster analysis, and principal component analysis, are used. For example, association rule mining can be used to find the correlation between battery operating parameters, such as the correlation between temperature and battery charging and discharging efficiency; cluster analysis can be used to classify the battery operating state into different categories in order to discover potential operating patterns. Through these algorithms, the long-term battery operating characteristics of each power station can be obtained, such as the charging and discharging patterns of batteries in different seasons and the changing trend of battery performance with the years of use.

[0062] More specifically, long-term historical data can reflect the battery's operation under different environments and usage stages. Through feature mining, we can gain a deeper understanding of the battery's operating patterns and performance change trends, providing a basis for subsequent evaluation and optimization. Uncovering potential problems and patterns hidden in the data can help predict potential battery failures in advance, formulate reasonable maintenance plans, and improve the reliability and stability of the energy storage system.

[0063] More specifically, based on the long-cycle battery operation characteristics, combined with actual business needs and battery operation features, the operating time of each power station is divided into intervals. For example, it can be divided according to seasons (spring, summer, autumn, winter), peak and off-peak electricity consumption periods, and different battery usage stages (such as new battery period, stable period, and degradation period). Indicators for evaluating the operating value of each power station are determined, such as battery charge and discharge efficiency, energy storage utilization rate, failure rate, and economic benefits. These indicators can comprehensively reflect the operating status and value of the power station. For each time interval, based on the determined evaluation indicators, the operating value parameters of each power station within that interval are calculated. For example, by calculating the battery charge and discharge efficiency and energy storage utilization rate during peak electricity consumption periods, its operating value during that period is evaluated; and the failure rate in different seasons is statistically analyzed to evaluate the battery reliability under different environments.

[0064] More specifically, the operating status and value of batteries vary across different time intervals. By dividing time intervals and assessing operating value, we can manage each power station in a more refined manner, develop corresponding operating strategies based on the characteristics of different time periods, and understand the operating value of each power station in different time intervals. This helps to allocate resources rationally and improve the overall efficiency of the energy storage system. For example, we can increase investment in energy storage during periods of high operating value and perform equipment maintenance during periods of low operating value.

[0065] More specifically, based on the operational value parameters of each power station within each time interval, the pre-prepared training data is modified. For example, if a power station has a high failure rate during a certain time period, the number of failure-related data samples during that time period can be increased, or the weights of the data can be adjusted to make the algorithm model pay more attention to failure situations. Suitable machine learning or deep learning algorithms, such as neural networks, decision trees, and support vector machines, are selected to train the operation control algorithm model deployed on the side. The modified training data is used to iteratively train the model, continuously adjusting the model parameters to improve the model's accuracy and performance. During the training process, the model is evaluated using a validation dataset, and evaluation metrics (such as precision, recall, mean squared error, etc.) are used to measure the model's performance. Based on the evaluation results, the model's structure and parameters are further adjusted until the model achieves satisfactory performance.

[0066] More specifically, different power stations have different operating conditions and environmental conditions. By correcting the training data according to the operating value parameters, the algorithm model can better adapt to the actual situation of each power station, improve the accuracy and adaptability of the model, train and optimize the algorithm model deployed on the side, continuously improve the battery operation control strategy, and improve the efficiency, reliability and economic benefits of the energy storage system. At the same time, the optimized model can be distributed to the side to realize real-time intelligent control and better cope with various situations in battery operation.

[0067] Reference Figure 2 As shown, in a second aspect, the present invention provides a data management device based on a cloud-edge-device architecture, used to implement the data management method based on a cloud-edge-device architecture as described in any one of the first aspects, comprising: The end-side processing module is used to collect the battery's operating parameters in real time based on the end-side deployment unit, obtain the real-time operating parameters, and perform preliminary processing and local caching of the real-time operating parameters. The edge processing module is used to analyze the real-time operating parameters based on the algorithm model of the edge deployment unit, generate the battery operation control strategy, and perform multi-dimensional calculations on the real-time operating parameters to obtain multi-dimensional calculation features. The cloud-side processing module is used to store the real-time operating parameters and multi-dimensional computational characteristics of batteries from multiple power stations on the cloud-based energy storage big data platform, thereby obtaining the datasets of each power station. The algorithm optimization module is used to perform a comprehensive analysis of the real-time operating parameters and multi-dimensional computational features of the datasets of each power station. Based on the analysis results, it trains and optimizes the algorithm model deployed on the side and then distributes the optimized algorithm model to the side.

[0068] In this embodiment, the specific implementation of each module in the above system embodiment is described in the above method embodiment, and will not be repeated here.

[0069] Thirdly, the present invention provides a data management device based on a cloud-edge-device architecture, including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements a data management method based on a cloud-edge-device architecture as described in any one of the first aspects.

[0070] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is run by a processor, the processor causes the processor to perform a data management method based on a cloud-edge-device architecture as described in any one of the first aspects.

[0071] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A data management method based on a cloud-edge-device architecture, characterized in that, include: The battery's operating parameters are collected in real time by the edge deployment unit to obtain real-time operating parameters, and the real-time operating parameters are preliminarily processed and cached locally. Based on the edge deployment unit, the real-time operating parameters are calculated in multiple dimensions to obtain multi-dimensional calculation features; The energy storage big data platform built on the cloud side stores the real-time operating parameters and multi-dimensional computing characteristics of batteries in multiple power stations, and obtains the datasets of each power station. A comprehensive analysis of the real-time operating parameters and multi-dimensional computational features of the datasets of each power station is conducted. Based on the analysis results, the algorithm model deployed on the edge is trained and optimized, and the optimized algorithm model is then distributed to the edge.

2. The data management method based on cloud-edge-device architecture as described in claim 1, characterized in that, The steps of collecting real-time battery operating parameters based on the edge deployment unit, obtaining real-time operating parameters, and performing preliminary processing and local caching of the real-time operating parameters include: By pre-deploying a data acquisition and management unit or smart gateway on the edge, the battery's operating parameters such as voltage, current, temperature, SOC, and SOH are collected to obtain real-time operating parameters; The real-time operating parameters are filtered, verified, and converted in format to complete the initial processing of the real-time operating parameters; The pre-processed real-time operating parameters are temporarily cached locally, and then transmitted to the edge deployment unit.

3. The data management method based on cloud-edge-device architecture as described in claim 1, characterized in that, The steps for performing multi-dimensional calculations on the real-time operating parameters based on the edge deployment unit to obtain multi-dimensional calculation features include: Based on a multidimensional computing algorithm model pre-deployed on the edge, multiple computing tasks are performed on the real-time running parameters to obtain the computing features of each computing task, which together serve as multidimensional computing features.

4. The data management method based on cloud-edge-device architecture as described in claim 3, characterized in that, Also includes: Based on the operation control algorithm model of the pre-deployed EMS energy management device and local controller on the side, the real-time operation parameters are analyzed to identify the actual operation characteristics of the battery at the current moment. Based on the actual operating characteristics, the corresponding operating control strategy is scheduled from the preset control strategy library to execute the operating control strategy to control the operation of the battery.

5. The data management method based on cloud-edge-device architecture as described in claim 3, characterized in that, The computational tasks include statistical analysis, early warning and alarm, feature extraction, fault diagnosis, and trend analysis.

6. The data management method based on cloud-edge-device architecture as described in claim 1, characterized in that, The steps involved in building a cloud-based energy storage big data platform to store real-time operating parameters and multi-dimensional computational characteristics of batteries from multiple power stations and obtain datasets for each power station include: The energy storage big data platform built on the cloud side receives real-time operating parameters and multi-dimensional computational characteristics of batteries from multiple power stations. The received real-time operating parameters and multi-dimensional computational features are labeled to generate corresponding information tags for the real-time operating parameters and multi-dimensional computational features; The real-time operating parameters and the multi-dimensional computational features are arranged in an orderly manner according to the information tags to obtain the dataset of the power station to which the battery belongs.

7. The data management method based on cloud-edge-device architecture as described in claim 1, characterized in that, The steps for conducting a comprehensive analysis of the real-time operating parameters and multi-dimensional computational features of the datasets from various power plants, and then training and optimizing the algorithm model deployed at the edge based on the analysis results, include: Based on the real-time operating parameters and multi-dimensional computational features in each dataset, feature mining of long-cycle historical data is performed on each power station to obtain the long-cycle battery operation characteristics of each power station. Based on the aforementioned long-cycle battery operation characteristics, time intervals are divided and operation value is evaluated for each power station to obtain operation value parameters for each power station in each time interval. The pre-prepared training data is corrected based on the operational value parameters, and the operational control algorithm model deployed in each power station is trained and optimized using the corrected training data.

8. A data management device based on a cloud-edge-device architecture, characterized in that, A data management method based on a cloud-edge-device architecture as described in any one of claims 1-7 includes: The end-side processing module is used to collect the battery's operating parameters in real time based on the end-side deployment unit, obtain the real-time operating parameters, and perform preliminary processing and local caching of the real-time operating parameters. The edge processing module is used to perform multi-dimensional calculations on the real-time running parameters based on the algorithm model of the edge deployment unit to obtain multi-dimensional calculation features; The cloud-side processing module is used to store the real-time operating parameters and multi-dimensional computational characteristics of batteries from multiple power stations on the cloud-based energy storage big data platform, thereby obtaining the datasets of each power station. The algorithm optimization module is used to perform a comprehensive analysis of the real-time operating parameters and multi-dimensional computational features of the datasets of each power station. Based on the analysis results, it trains and optimizes the algorithm model deployed on the side and then distributes the optimized algorithm model to the side.

9. A data management device based on a cloud-edge-device architecture, comprising a memory and a processor, wherein the memory stores a computer program that can run on the processor, characterized in that, When the processor executes the computer program, it implements a data management method based on a cloud-edge architecture as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which, when run by a processor, causes the processor to execute a data management method based on a cloud-edge architecture as described in any one of claims 1-7.

Images