Energy storage battery health management system based on digital twinning and energy storage system

Abstract

The invention discloses an energy storage battery health management system based on digital twinning and an energy storage system, and the system comprises a demand analysis module which is used for determining operation and maintenance demand parameters of equipment in the energy storage system, and determining a demand element frame according to the operation and maintenance demand parameters of the energy storage system; the demand element framework represents a result of carrying out importance classification on the operation and maintenance demand parameters; the digital twinborn model construction module is used for determining an operation scene of the energy storage system according to the demand element framework, matching an equipment simulation model meeting the operation scene and a hyper-parameter combination of the equipment simulation model from a model library, and fusing to obtain a digital twinborn model of the energy storage system; and the data analysis module is used for receiving the real-time operation feedback data of the energy storage system and outputting the virtual simulation data and a health assessment result corresponding to the real-time operation feedback data based on the digital twin model. In the embodiment of the invention, the digital twin model matched with the energy storage system can be provided.

Description

Technical Field

[0001] The present application relates to the field of electrochemical energy storage technology, and in particular to a digital twin-based energy storage battery health management system and energy storage system. Background Art

[0002] Electrochemical energy storage systems have the advantages of fast power response, energy-intensive storage, flexible deployment, and many other features, such as improving power quality, peak-shaving, demand control, frequency regulation, peak-shaving, and grid-friendly. They have become one of the fastest-growing and most widely used energy storage technologies. The increased dynamics and complexity of the entire "source-grid-load-storage" link has put forward higher requirements for the stable operation of electrochemical energy storage systems and battery health management.

[0003] In related technologies, digital twin technology is used to build a digital twin model of batteries, and intelligent battery health management is performed through the digital twin model. The performance of the digital twin model is related to the operation effect of the battery health management system. At present, the digital twin method model of the electrochemical energy storage system is solidified, the model's general ability is weak, and the operation and maintenance demand data related to the model parameters have a large number of subjective factors, lacking objectivity and reliability. Summary of the invention

[0004] The embodiments of the present application provide a digital twin-based energy storage battery health management system and an energy storage system, which can provide a digital twin model matching the energy storage system.

[0005] In a first aspect, an embodiment of the present application provides a digital twin-based energy storage battery health management system, including:

[0006] A demand analysis module, used to determine the operation and maintenance demand parameters of the equipment in the energy storage system, and determine a demand factor framework according to the operation and maintenance demand parameters of the energy storage system; the demand factor framework represents the result of classifying the importance of the operation and maintenance demand parameters;

[0007] A digital twin model construction module is used to determine the operation scenario of the energy storage system according to the demand factor framework, and match the device simulation model that meets the operation scenario and the hyperparameter combination of the device simulation model from the model library to obtain the digital twin model of the energy storage system by fusion;

[0008] A data analysis module is used to receive real-time operation feedback data of the energy storage system, and output virtual simulation data and health assessment results corresponding to the real-time operation feedback data based on the digital twin model.

[0009] In some embodiments, the energy storage battery health management system also includes a model optimization module, which is used to optimize the device simulation model and the hyperparameter combination through a reinforcement learning algorithm based on the difference between the real-time operation feedback data and the virtual simulation data, with the goal of maximizing the accuracy of the model, and update the model library according to the optimized results.

[0010] In some embodiments, the digital twin model construction module is also used to build a new equipment simulation model based on the mechanism model of the energy storage battery and the operation and maintenance demand parameters corresponding to the demand element framework if a device simulation model that meets the operation scenario cannot be matched from the model library, and add the new equipment simulation model to the model library.

[0011] In some embodiments, the digital twin model construction module is further used to add an operation scenario label to the new device simulation model, and the operation scenario label corresponds to the operation scenario of the energy storage system;

[0012] The digital twin model construction module is also used to determine the operation scenario label according to the demand factor framework, automatically match the equipment simulation model that meets the operation scenario from the model library according to the operation scenario label, and automatically set the hyperparameter combination of the equipment simulation model.

[0013] In some embodiments, the architecture of the digital twin model includes a physical layer, a data layer, a mechanism layer, a presentation layer and an interaction layer; the physical layer is a physical bottom layer of the digital twin built according to the data acquisition components installed in the energy storage system, the data layer is a data processing layer that performs data integration and normalization on the real-time operation feedback data, the mechanism layer is a model layer that analyzes the mapping relationship between the mechanism model of the energy storage battery and the data-driven model of the digital twin, the presentation layer is a display interface of the digital twin model for the mirror data of the energy storage system, and the interaction layer is a user interaction interface for the health management of the energy storage battery.

[0014] In some embodiments, the demand analysis module is specifically used to convert the operation and maintenance demand parameters into demand elements of a preset demand analysis model, classify the attributes of the demand elements according to the demand analysis model, and sort the importance of the demand elements through a hierarchical analysis method, and then adjust the importance of the sorted demand elements according to the quality function deployment method to obtain a demand element framework.

[0015] In some embodiments, the digital twin model includes an SOC prediction model, which is used to obtain characteristic parameters that affect the SOC of the energy storage battery in the real-time operation feedback data, input the characteristic parameters into a Kalman filter for state estimation, and obtain a state estimation value, and then use the characteristic parameters and the state estimation value as input parameters for prediction of a timing algorithm to obtain a SOC prediction value of the energy storage battery; wherein the timing algorithm is selected from a preset algorithm library.

[0016] In some embodiments, the SOC prediction model is constructed by offline modeling; the construction method is to obtain historical aging data of the energy storage battery, use the historical aging data to construct a training data set, and after time series processing, the training data set is input into a Kalman filter, and adjusted by a state prediction equation, a covariance prediction equation and a Kalman gain equation, and a state estimation value of the training data set is obtained by a state update equation and a covariance update equation, and then the SOC prediction model is constructed according to the historical aging data and the state estimation value of the training data set; wherein the historical aging data includes the current, voltage and historical SOC value of the energy storage battery, and the state equation of the Kalman filter is obtained by a battery electrical model based on a Thevenin equivalent circuit.

[0017] In some embodiments, the energy storage battery health management system also includes a data interactive sharing model and an interactive interface, wherein the interactive interface is connected to a data acquisition component of the energy storage system to collect data, and the data interactive sharing model is used to adjust the data received by the interactive interface based on a predefined data structure to form standardized real-time operation feedback data, and the data interactive sharing model is also used to forward the standardized real-time operation feedback data to the digital twin model in real time.

[0018] In a second aspect, an embodiment of the present application further provides an energy storage system, including the energy storage battery health management system of the embodiment of the first aspect.

[0019] The energy storage battery health management system and energy storage system based on digital twins of the embodiments of the present application have at least the following beneficial effects: the demand analysis module determines the operation and maintenance demand parameters of the equipment in the energy storage system, and can obtain a demand factor framework classified according to importance, and then matches the equipment simulation model corresponding to the operation scenario and the hyperparameter combination of the equipment simulation model from the model library based on the demand factor framework, thereby integrating to obtain a digital twin model of the energy storage system. The digital twin model can display the virtual simulation data and health assessment results of the energy storage system. Therefore, it can be seen that the energy storage battery health management system solves the problem of many subjective factors in the operation demand parameters through demand classification, and obtains a demand factor framework that can observably reflect the importance of the demand factors. Then, the corresponding digital twin model is built through the model library that meets different operation scenarios to form a high-fidelity mirror image of the energy storage system, effectively improving the response speed and accuracy of data interaction, so that the built digital twin model has better adaptability and can provide battery health management solutions for energy storage systems in different scenarios.

[0020] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained through the structures particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a schematic diagram of an energy storage battery health management system provided by an embodiment of the present application;

[0022] Figure 2 It is a schematic diagram of the architecture of a digital twin provided by an embodiment of the present application;

[0023] Figure 3 It is a schematic diagram of predicting the health status of an energy storage battery provided by an embodiment of the present application;

[0024] Figure 4 is a schematic diagram of the architecture of an energy storage system provided in one embodiment of the present application;

[0025] Figure 5 It is a schematic diagram of the hardware device structure of the energy storage system provided in one embodiment of the present application;

[0026] Figure 6 It is a schematic diagram of the system data flow provided by an example of the present application. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. In addition, the characteristics, operations or features described in the specification can be combined in any appropriate manner to form various implementation methods. At the same time, the steps or actions in the method description can also be replaced or adjusted in order in a manner that is obvious to those skilled in the art. Therefore, the various sequences in the specification and the accompanying drawings are only for the purpose of clearly describing a certain embodiment and are not meant to be a necessary sequence, unless otherwise specified that a certain sequence must be followed.

[0028] In the description of this application, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed", etc. are understood to exclude the number itself, and "above", "below", "within", etc. are understood to include the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0029] The serial numbers of the components in this document, such as "first", "second", etc., are only used to distinguish the objects described and do not have any order or technical meaning. The "connection" and "coupling" mentioned in this application, unless otherwise specified, include direct and indirect connections (couplings).

[0030] Electrochemical energy storage systems have the advantages of fast power response, energy-intensive storage, flexible deployment, and many other features, such as improving power quality, peak-shaving, demand control, frequency regulation, peak-shaving, and grid-friendly. They have become one of the fastest-growing and most widely used energy storage technologies. The increased dynamics and complexity of the entire "source-grid-load-storage" link has put forward higher requirements for the stable operation of electrochemical energy storage systems and battery health management.

[0031] Due to the complex reaction and performance degradation mechanism inside the battery, it is difficult for the external characteristic parameters to directly reflect the safety status and life stage of the battery. It is impossible to track and diagnose the entire process of battery use during its life cycle and provide safety warnings for the battery in advance. At the same time, the data collected by the existing energy storage system has the characteristics of large interference, large time scale, low specification, non-standard and weak promotion. There are a lot of subjective factors in the operation and maintenance demand data, which leads to the lack of objectivity and reliability of its classification method. At the same time, the electrochemical energy storage system faces problems such as diversified operation scenarios, random environmental uncertainty, and insufficient general adaptive capabilities of the model.

[0032] That is to say, the digital twin technology is used in related technologies to build a digital twin model of the battery, and the digital twin model is used to perform intelligent battery health management. The performance of the digital twin model is related to the operation effect of the battery health management system. At present, the digital twin method model of the electrochemical energy storage system is solidified, the model's general ability is weak, and the operation and maintenance demand data related to the model parameters have a large number of subjective factors, lacking objectivity and reliability.

[0033] Based on this, the embodiment of the present application provides a digital twin-based energy storage battery health management system and energy storage system. The demand analysis module determines the operation and maintenance demand parameters of the equipment in the energy storage system, and can obtain a demand factor framework classified according to importance. Then, based on the demand factor framework, the device simulation model corresponding to the operation scenario and the hyperparameter combination of the device simulation model are matched from the model library, thereby integrating the digital twin model of the energy storage system. The digital twin model can display the virtual simulation data and health assessment results of the energy storage system. Therefore, it can be seen that the energy storage battery health management system solves the problem of many subjective factors in the operation demand parameters through demand classification, and obtains a demand factor framework that can observably reflect the importance of the demand factors. Then, the corresponding digital twin model is built through the model library that meets different operation scenarios to form a high-fidelity mirror image of the energy storage system, effectively improving the response speed and accuracy of data interaction, so that the built digital twin model has better adaptability and can provide battery health management solutions for energy storage systems in different scenarios.

[0034] The following is an explanation of the energy storage battery health management system and energy storage system based on digital twins with the attached drawings:

[0035] Reference Figure 1 As shown, Figure 1 It is a schematic diagram of an energy storage battery health management system provided in one embodiment of the present application.

[0036] In some embodiments, the energy storage battery health management system includes a demand analysis module 100, a digital twin model construction module 200 and a data analysis module 300. These three modules are described in detail below.

[0037] The demand analysis module 100 is used to determine the operation and maintenance demand parameters of the equipment in the energy storage system, and determine the demand factor framework based on the operation and maintenance demand parameters of the energy storage system, wherein the demand factor framework represents the result of classifying the importance of the operation and maintenance demand parameters. Among them, the operation and maintenance demand parameters of the embodiment of the present application include the operation state prediction parameters of the equipment, the health state rapid response parameters and the remote monitoring parameters. Specifically, the operation state prediction parameters refer to the parameters of the functional realization requirements of the equipment in the energy storage system, such as the operation state parameters of the battery cell, PCS (Power Conversion System, energy storage converter) and other equipment, the health state rapid response parameters refer to the parameters that are controlled and adjusted according to the operation state, such as the balance control of the battery cell, the safety control during the abnormal warning of the PCS, etc., and the remote monitoring parameters refer to the parameters of the real-time feedback of the equipment parameters.

[0038] In the process of determining the demand factor framework based on the operation and maintenance demand parameters of the energy storage system, the embodiment of the present application first selects the Kano demand analysis model to classify the attributes of the operation and maintenance demand parameters, and divides the operation and maintenance demand parameters into necessary type, expected type, exciting type, indifferent type and reverse type through the Kano demand analysis model, and uses the hierarchical analysis method to sort the importance of the operation and maintenance demand parameters. Secondly, the importance of data management is adjusted based on the GQFD (Grey Quality Function Deployment) theory. Demand analysis is performed by combining demand classification and importance analysis to solve the problem of more subjective factors in demand data, thereby obtaining objective demand classification and the importance of similar demand factors. Finally, based on the results of the demand analysis, the demand factor framework is determined to make it suitable for the needs of the energy storage battery health management system.

[0039] It can be understood that the energy storage system in the embodiments of the present application can be a wind-solar energy storage system, a thermal energy storage system, an industrial and commercial energy storage system, etc. Specifically, the energy storage system can be applied to but not limited to operating scenarios such as wind-solar storage, thermal storage, and industrial and commercial storage, and the type of energy storage battery can be a lithium battery, a lead-acid battery, and a graphene battery, etc., which is not specifically limited in the embodiments of the present application.

[0040] In some embodiments, due to the large differences in operating scenarios such as industrial and commercial storage, wind and solar storage, etc., separate modeling is required for different scenarios. Specifically, the digital twin model construction module 200 in the embodiment of the present application is used to determine the operating scenario of the energy storage system according to the demand factor framework, and match the device simulation model and the hyperparameter combination of the device simulation model that meet the operating scenario from the model library, so as to improve the multi-scenario adaptability through scenario adaptation, and integrate the digital twin model of the energy storage system to achieve self-learning adaptation of the scenario model.

[0041] It is worth noting that the energy storage battery health management system of the embodiment of the present application is suitable for multiple operating scenarios through two-layer optimization, wherein the first layer has a built-in model library and algorithm library, so as to improve the adaptability to multiple scenarios through scenario adaptation; the second layer can evolve the optimization algorithm to realize the self-learning adaptation of the scenario model, that is, to match the device simulation model and the hyperparameter combination of the device simulation model that meets the operating scenario from the model library.

[0042] In some embodiments, the data analysis module 300 is used to receive real-time operation feedback data of the energy storage system, and output virtual simulation data and health assessment results corresponding to the real-time operation feedback data based on the digital twin model, so as to realize real-time interaction of the operation data of the digital twin of the energy storage battery health management system and improve the overall operation and regulation level of the energy storage battery health management system.

[0043] It should be noted that the data analysis module 300 is also used to perform data anomaly cleaning and correction, and adaptive repair of time and space for various basic data, so as to provide more accurate intelligent panoramic situation awareness analysis. In addition, the data analysis module 300 is also used to extract valuable information from the data, mine and extract valuable information, and use distributed storage and real-time extraction to reduce the size and complexity of the data.

[0044] In some embodiments, the energy storage battery health management system also includes a model optimization module, which is used to optimize the device simulation model and hyperparameter combination through a reinforcement learning algorithm based on the difference between real-time operation feedback data and virtual simulation data, with the goal of maximizing the accuracy of the model, and update the model library according to the optimized results to achieve adaptive update of the model library, and achieve self-learning of the core parameters of the scenario model through an evolvable reinforcement learning method, thereby improving the ability to cope with random uncertainties and realizing dynamic correction and self-evolution of key models and parameters.

[0045] Specifically, the agent in the model optimization module aims to maximize the accuracy of the model, using the accuracy as the reward value, and continuously learns through the reinforcement learning algorithm until the optimal model and hyperparameter combination are found. Then, the model is corrected through the optimization of the model's hyperparameter combination, and online prediction is achieved. At the same time, feedback is used to achieve adaptive updates of the model library model and realize a closed-loop model.

[0046] Reference Figure 2 As shown, Figure 2 This is a schematic diagram of the architecture of a digital twin provided in one embodiment of the present application.

[0047] In some embodiments, the digital twin model construction module 200 is also used to build a new equipment simulation model based on the operation and maintenance demand parameters corresponding to the mechanism model of the energy storage battery and the demand factor framework if a device simulation model that meets the operation scenario cannot be matched from the model library, and add the new equipment simulation model to the model library. The model solidification problem of the traditional twin method is solved through the constructed multi-scenario model library and algorithm library, thereby improving the self-selection capability of the twin system and the flexible application capability of the model.

[0048] In some embodiments, the architecture of the digital twin of the embodiment of the present application is a multi-level architecture. Specifically, the architecture of the digital twin model includes a physical layer, a data layer, a mechanism layer, a presentation layer and an interaction layer; the physical layer is the physical bottom layer of the digital twin built according to the data acquisition components installed in the energy storage system, the data layer is the data processing layer that performs data integration and normalization on real-time operation feedback data, the mechanism layer is the model layer that analyzes the mapping relationship between the mechanism model of the energy storage battery and the data-driven model of the digital twin, the presentation layer is the display interface of the digital twin model for the mirror data of the energy storage system, and the interaction layer is the user interaction interface for energy storage battery health management.

[0049] Specifically, the digital twin model first determines the data required for battery health management based on the mechanism model and mathematical model of the energy storage battery, installs the corresponding data collection equipment, and completes the construction of the physical layer of the digital twin. Then, the data of the energy storage battery health management system is integrated and standardized, and the data layer of the digital twin is built to provide a data basis for the operation and optimization of the energy storage battery health management system. At the mechanism layer, from the evolution law model of the "electricity-heat-power-gas" characteristics of the energy storage battery throughout its life cycle, a mathematical or mechanism model for the complex reaction and performance degradation mechanism inside the energy storage battery is established to provide a theoretical basis for the prediction of the equipment's operating life and status; at the presentation layer, the digital twin realizes a high-fidelity mirroring of the energy storage system, so that on-site personnel can quickly locate the battery fault point and implement repairs based on the information provided by the digital twin. Finally, at the interaction layer, the real-time operating status information of the real system is integrated into the forward-looking prediction of the digital twin, and finally a standardized multi-level digital twin architecture for different application scenarios is proposed.

[0050] It should be noted that the physical layer in the embodiment of the present application covers multiple elements such as transmission and storage, and is used to establish the physical architecture of the digital twin; the data layer examines the data storage and management mechanism, and analyzes the digital twin data processing method; the mechanism layer is used to analyze the mapping relationship between the mechanism model of the physical object and the data-driven model of the digital twin; the presentation layer is used to analyze the output expression form of the digital twin model; the interaction layer analyzes the interaction mechanism between the digital twin system and the actual system and users.

[0051] In some embodiments, the digital twin model construction module 200 is further used to add an operation scenario tag to the new device simulation model, and the operation scenario tag corresponds to the operation scenario of the energy storage system;

[0052] The digital twin model construction module 200 is also used to determine the operation scenario label according to the demand factor framework, automatically match the equipment simulation model that meets the operation scenario from the model library according to the operation scenario label, and automatically set the hyperparameter combination of the equipment simulation model.

[0053] It should be noted that the embodiments of the present application will set labels for different scenarios. One scenario will have one label, and this label will be associated with the model library and algorithm library under a scenario, so the agent can directly select it. The hyperparameter combination here actually means that after the model is adopted, the model parameters will be optimized according to the data of the scenario. The hyperparameters will be determined according to the algorithm. Different algorithms have different hyperparameters, such as learning rate, penalty coefficient, number of networks, etc.

[0054] In some embodiments, the demand analysis module 100 is specifically used to convert operation and maintenance demand parameters into demand elements of a preset demand analysis model, classify the attributes of the demand elements according to the demand analysis model, and sort the demand elements by importance through a hierarchical analysis method, and then adjust the importance of the sorted demand elements according to the quality function deployment method to obtain a demand element framework.

[0055] Reference Figure 3 As shown, Figure 3 It is a schematic diagram of predicting the health status of an energy storage battery provided in an embodiment of the present application.

[0056] In some embodiments, the digital twin model includes a SOC (State of Charge) prediction model, which is used to obtain characteristic parameters that affect the SOC of the energy storage battery in real-time operation feedback data, input the characteristic parameters into a Kalman filter for state estimation, obtain state estimation values, and then use the characteristic parameters and state estimation values ​​as input parameters for prediction of a timing algorithm to obtain a SOC prediction value of the energy storage battery; wherein the timing algorithm is selected from a preset algorithm library.

[0057] Specifically, the SOC prediction model includes a data acquisition module, a state estimation module and a prediction module. In the process of obtaining the SOC prediction value of the energy storage battery, the acquisition module in the SOC prediction model first collects characteristic parameters such as current, voltage and ambient temperature that affect the battery SOC, and the state estimation module implements the state estimation of the characteristic parameter input Kalman filter, wherein the current and voltage are adjusted through the state prediction equation, the covariance prediction equation and the Kalman gain equation, and the Kalman filter state estimation value is finally obtained through the state update equation and the covariance update equation, wherein the state equation of the Kalman filter is obtained by the battery kinetic model based on the Thevenin equivalent circuit; the prediction module adopts a time series prediction method, and uses the current, voltage, ambient temperature measurement values ​​and the Kalman filter state estimation value as input characteristic parameters to obtain the battery SOC prediction value.

[0058] It should be noted that the time series algorithm comes from the constructed algorithm library, including but not limited to RNN (Recurrent Neural Network), LSTM (Long Short-Term Memory), GRU (Gated Recurrent Unit) and other algorithms, which introduce memory and gating mechanisms to process and model sequence data.

[0059] It is worth noting that the state estimation value in the embodiment of the present application includes an SOH (State of Health) analysis value and a loss analysis value. Among them, the SOH analysis involves the calculation of the energy storage SOH decay trend and the equivalent number of battery cycles. According to the battery failure mechanism, the loss analysis includes reversible capacity loss and irreversible capacity loss, among which the evaluation of irreversible capacity loss is based on the evaluation analysis model for data supervised learning, and the main indicator parameters include loss data size, cycle stability performance indicators, etc.

[0060] In some embodiments, the SOC prediction model is constructed by offline modeling; the construction method is to obtain historical aging data of the energy storage battery, use the historical aging data to construct a training data set, and after time series processing of the training data set, input it into the Kalman filter, adjust it through the state prediction equation, covariance prediction equation and Kalman gain equation, and obtain the state estimation value of the training data set through the state update equation and covariance update equation, and then construct the SOC prediction model according to the historical aging data and the state estimation value of the training data set; wherein the historical aging data includes the current, voltage and historical SOC value of the energy storage battery, and the state equation of the Kalman filter is obtained by a battery electrical model based on the Thevenin equivalent circuit.

[0061] It should be noted that the embodiment of the present application constructs an SOC prediction model through a Kalman filter and a timing algorithm, the Kalman filter performs state estimation, and the timing algorithm realizes online SOC prediction. Specifically, SOC prediction monitoring is divided into offline model construction and online calculation, wherein offline modeling mainly uses current, voltage, temperature and historical SOC values ​​to construct an SOC prediction model through timing prediction, and uses historical aging data such as current, voltage and ambient temperature of different battery working conditions as a training data set, and inputs the training data into a Kalman filter in a time series to obtain a set of Kalman filter state estimation values; then the battery historical aging data and the Kalman filter state estimation values ​​are normalized and other data preprocessing is performed, and used as a data set to construct a timing prediction model offline.

[0062] In some embodiments, the energy storage battery health management system also includes a data interactive sharing model and an interactive interface 400. The interactive interface 400 is connected to the data acquisition component of the energy storage system to collect data. The data interactive sharing model is used to adjust the data received by the interactive interface 400 based on a predefined data structure to form standardized real-time operation feedback data. The data interactive sharing model is also used to forward the standardized real-time operation feedback data to the digital twin model in real time. The established data interactive sharing model is used to achieve panoramic perception of energy storage battery health management, drive the continuous improvement of the digital twin model and parameters, and the synchronous evolution of the virtual-real system.

[0063] It should be noted that the design of the interactive interface 400 follows the design principles of segmentation, high cohesion, and low coupling, minimizes the degree of integration between systems and modules within the system, reduces operational complexity, and all interface designs must comply with industry interface specifications to ensure the universality of implementation, improve the reusability and scalability of the system, and ensure the consistency of interface data among the various systems involved in the interface.

[0064] It can be understood that the energy storage battery health management system of the embodiment of the present application establishes a data interactive sharing model in the form of a wide area service agent. The message bus is used as the information transmission channel to realize data transmission, the data flow model is defined using a standardized data structure, and real-time data delivery is performed according to the data flow definition, so as to realize the real-time interaction of the operation data of the digital twin of the energy storage battery health management system, and finally improve the overall operation and control level of the energy storage battery health management system.

[0065] Reference Figure 4 As shown, Figure 4 It is a schematic diagram of the architecture of an energy storage system provided in one embodiment of the present application.

[0066] An embodiment of the present application also provides an energy storage system, including the energy storage battery health management system of the above embodiment.

[0067] In some embodiments, the energy storage system includes an application layer, a platform layer, and a basic resource layer. The basic resources provide data collection, storage, transmission, computing and other resources to support the platform layer's data analysis, edge computing and digital twin architecture capabilities, thereby realizing the application layer's real-time monitoring, real-time interaction, status diagnosis, predictive maintenance, battery health assessment and other functions. The energy storage battery health management system is used to achieve battery health management throughout its life cycle, detect the battery cell performance characteristic parameters in real time, and locate degraded units online by analyzing changes in key performance parameters during battery aging. At the same time, it provides services such as battery health status assessment, battery status correction, and battery remaining life prediction. In the process of data processing and management, a clear data flow and communication mechanism need to be established to ensure smooth connections between various links, as well as the accuracy, timeliness and effectiveness of the data.

[0068] In some embodiments, the basic resource layer is the bottom layer of the architecture, and the functions of collecting, storing, transmitting, and calculating battery operation data are realized through hardware devices such as sensors, storage, and servers. Among them, data collection is realized through collection sensing devices, and the devices used for different operating parameters are different, such as pressure sensors, current sensors, temperature sensors, etc., to ensure that the average response time of data collection is <100ms, and to realize real-time data collection. Among them, data transmission can be carried out by wired optical fiber or wireless WIFI according to the actual object situation, to ensure that the average completion time of data transmission is <100ms; the average network delay is <200ms.

[0069] It should be noted that data is stored through storage hardware devices, and the storage capacity is customized according to the actual needs of the hardware devices, and the system is guaranteed to operate normally when the disk (database) is full.

[0070] In some embodiments, the platform layer is an intermediate core computing layer, which realizes functions such as data analysis of battery operation data, edge computing, and real-time interaction of digital twins. Among them, data analysis realizes functions such as operation data cleaning, specification, verification, feature analysis, and rule extraction, and the data is transmitted to the edge computing module after preprocessing. The edge computing module mainly realizes functional calculations, and the server required for calculation is customized according to the amount of data and the complexity of algorithm calculation, mainly considering computing power, CPU, GPU, memory, communication and other capabilities. For example, R-600 has a computing power of 275Tops, a 12-core CPU, a 2048-core GPU, 64GB of memory, and supports 4G / 5G / WIFI communication.

[0071] It should be noted that edge computing has the advantages of high-speed data interaction, massive and heterogeneous connection, security and privacy protection, giving full play to the value of data, helping to break down industry barriers, enhancing the real-time nature of data processing, and reducing the computing pressure of cloud platforms. The cloud platform receives key parameters transmitted by encryption technology, which can be uploaded to the cloud platform's data center through wireless communication technology, Wi-Fi or cellular network, and conducts data mining.

[0072] Specifically, the data analysis and edge computing modules can detect the characteristic parameters of battery cell performance in real time, locate degraded units online by analyzing the changes in key performance parameters during battery aging, and provide services such as battery health status assessment, battery status correction, and battery remaining life prediction.

[0073] Reference Figure 5 As shown, Figure 5 It is a schematic diagram of the hardware device structure of the energy storage system provided in one embodiment of the present application.

[0074] In some embodiments, the hardware device structure of the energy storage system includes a local device network, an edge device network and a cloud platform.

[0075] Among them, local devices and networks realize local data collection through sensors and instruments, and ensure data network transmission through serial port servers and switches; edge devices and networks mainly involve edge database storage, data transmission and edge computing devices that meet edge computing power requirements, as well as switches and firewall devices that ensure data network security; cloud platforms mainly involve cloud-based large data storage databases, intensive data transmission, complex computing power cloud computing, as well as result display and remote control.

[0076] In order to more clearly illustrate the energy storage battery health management system and energy storage system based on digital twins in the embodiment of the present application, specific examples are provided below for illustration.

[0077] Example 1:

[0078] refer to Figure 1 The detailed process of battery health management based on digital twin energy storage battery health management system is as follows Figure 1 As shown, battery health management includes steps such as demand analysis in multiple operating scenarios of the energy storage battery health management system, digital twin architecture of the battery health management system, and real-time application of the energy storage battery health management system.

[0079] First of all, this example takes wind and solar storage, thermal storage, and industrial and commercial storage as the multiple operating scenarios of the energy storage battery health management system, and constructs the demand factor framework of the energy storage battery health management system according to three demand categories: equipment operation status prediction, rapid response to health status, and remote monitoring. Specifically, the Kano demand analysis model is first selected to classify the attributes of multi-scenario demand factors, and the hierarchical analysis method is used to rank the importance of demand factors. Then, the importance of data management is adjusted based on the GQFD quality function deployment theory. Demand analysis is carried out by combining demand classification and importance analysis to solve the problem of more subjective factors in demand data, so as to obtain objective demand classification and the importance of similar demand factors. Afterwards, the demand factor framework is improved based on the results of demand analysis, and a classification method suitable for the requirements of energy storage battery health management system is proposed.

[0080] It should be noted that in the above steps, the operation scenarios may include but are not limited to wind and solar storage, thermal storage, and industrial and commercial storage. Demand analysis is to construct a demand factor framework for the energy storage battery health management system based on three demand categories: equipment operation status prediction, rapid response to health status, and remote monitoring. Specifically, demand analysis is achieved by combining demand classification and importance analysis. The Kano model is used to classify the attributes of multi-scenario demand factors, and AHP (Analytic Hierarchy Process) is used to rank the importance of demand factors; the GQFD quality function deployment theory is used to adjust the importance of data management.

[0081] Secondly, based on the demand factor framework and combined with digital twin technology, the monitoring, diagnosis and optimization functions of the energy equipment of the energy storage battery health management system are designed, so as to realize the full life cycle management of the energy storage battery health management system. Specifically, the data required for battery health management is determined according to the mechanism model and mathematical model of the energy storage battery, and the corresponding data collection equipment is installed to complete the construction of the physical layer of the digital twin. Then the system is integrated and standardized, and the data layer of the digital twin is built to provide a data basis for the operation and optimization of the battery health management system. At the mechanism layer, from the evolution law model of the "electricity-heat-force-gas" characteristics of the battery throughout its life cycle, a mathematical or mechanism model for the complex reaction and performance degradation mechanism inside the battery is established, providing a theoretical basis for the prediction of the equipment's operating life and status; at the presentation layer, the digital twin realizes a high-fidelity mirroring of the energy storage system, so that on-site personnel can quickly locate the battery fault point and implement repairs based on the information provided by the digital twin. Finally, at the interaction layer, the real-time operating status information of the real system is integrated into the forward-looking prediction of the digital twin, and the refined virtual-reality interaction method of the operation control strategy is explored, and finally a standardized multi-level digital twin architecture for different application scenarios is proposed.

[0082] refer to Figure 3 The core of the battery health management architecture is the prediction, perception, monitoring and diagnosis of the battery health status, which is mainly based on data-driven and physical mechanism fusion modeling. The SOC prediction model is built through the Kalman filter and the timing algorithm. The Kalman filter performs state estimation, and the timing algorithm realizes online SOC prediction. Specifically, SOC prediction monitoring is divided into offline model construction and online calculation. The offline modeling mainly uses current, voltage, temperature and historical SOC values ​​to build the SOC prediction model through timing prediction. The historical aging data such as current, voltage and ambient temperature of the battery under different working conditions are used as the training data set. The training data is time-series input into the Kalman filter to obtain a set of Kalman filter state estimation values; then the battery historical aging data and the Kalman filter state estimation values ​​are normalized and other data preprocessing is performed, and the timing prediction model is built offline as a data set. Online calculation includes data acquisition module, state estimation module and prediction module. The acquisition module acquires characteristic parameters such as current, voltage and ambient temperature that affect battery SOC. The state estimation module realizes the state estimation of the characteristic parameter input Kalman filter, in which the current and voltage are adjusted through the state prediction equation, covariance prediction equation and Kalman gain equation, and the Kalman filter state estimation value is finally obtained through the state update equation and covariance update equation. The state equation of the Kalman filter is obtained by the battery kinetic model based on the Thevenin equivalent circuit. The prediction module adopts the time series prediction method, and uses the current, voltage, ambient temperature measurement value and Kalman filter state estimation value as input characteristic parameters to obtain the battery SOC prediction value. The time series algorithm comes from the constructed algorithm library, including RNN, LSTM, GRU and other algorithms, and introduces memory and gating mechanisms to process and model sequence data. The energy storage battery health management system built is suitable for multiple operating scenarios through two-layer optimization. The first layer has a built-in model library and algorithm library, which improves the adaptability of multiple scenarios through scenario adaptation; the second layer builds an agent to automatically select the model and the corresponding hyperparameter combination to achieve self-learning adaptation of the scenario model. The agent aims to maximize the accuracy of the model, with the accuracy as the reward value (reward), and continuously learns through the reinforcement learning algorithm until the optimal model and hyperparameter combination are found. The battery health status monitoring and prediction method is also applied to different levels such as battery cells, modules, and clusters. The battery health management model library is constructed by offline training modeling of historical data of multiple scenarios and multiple objects.

[0083] Finally, through the established data analysis platform and data interactive sharing model, the panoramic perception of energy storage battery health management is realized, driving the continuous improvement of digital twin models and parameters and the synchronous evolution of virtual-real systems. First, data anomaly cleaning and correction, time and space adaptive repair are performed for various basic data, so as to provide more accurate intelligent panoramic situation awareness analysis. Secondly, the value information of the data is extracted, the valuable information is mined and extracted, and distributed storage and real-time extraction are used to reduce the scale and complexity of the data. Finally, in view of the problems of high data interaction threshold, long response time, and only quasi-real-time data service in the traditional cross-sectional data access method, the sharing and sharing data publishing and receiving rules are designed based on the wide-area service agent form, and the data interactive sharing model in the form of wide-area service agent is established based on the semantic mapping relationship of system data in the description layer, data association layer and interactive sharing layer. The message bus is used as the information transmission channel to realize data transmission, the data flow model is defined using a standardized data structure, and real-time data delivery is performed according to the data flow definition, so as to realize the real-time interaction of the operation data of the digital twin of the energy storage battery health management system, and finally improve the overall operation and control level of the energy storage battery health management system.

[0084] Specifically, the data analysis platform built first cleans and corrects data anomalies and performs adaptive repairs in time and space for various basic data, thereby providing more accurate intelligent panoramic situation awareness analysis. It also extracts valuable information from the data, mines and extracts valuable information, and uses distributed storage and real-time extraction to reduce data size and complexity.

[0085] Furthermore, the energy storage system designs sharing and sharing data publishing and receiving rules based on the wide area service agent form, and establishes a data interactive sharing model in the form of a wide area service agent based on the semantic mapping relationship of system data in the description layer, data association layer and interactive sharing layer.

[0086] It is worth noting that the energy storage system also uses a message bus as an information transmission channel to realize data transmission, uses standardized data structures to define data flow models, and performs real-time data delivery according to the data flow definition, realizing real-time interaction of operating data of the digital twin of the energy storage battery health management system.

[0087] It is worth noting that after the demand analysis is carried out in step 1 to determine the elements, the health assessment is generated through battery data collection, uploading, storage, and analysis in step 2, and finally executed on the physical entity of the energy storage power station in step 3, forming a complete closed-loop process.

[0088] Example 2:

[0089] refer to Figure 2, based on the five-layer system architecture of the digital twin platform, using the high-speed data interaction capability of the edge computing platform, establish a virtual-real information link mechanism for the real-time operation mirror of the digital twin, aiming at the consistency of the real-time operation status of the digital twin and the real system, establish the information link between the digital twin and the real measurement, build the virtual mirror boundary constraints based on historical data and real-time measurement, ensure the consistency of the digital mirror state and the real measurement state, realize the real-time mapping of the physical space, and coordinate and control each interface in a digital form, and support the operation of the system digital twin platform through the "data transmission chain". The multi-scenario model library and algorithm library built solve the model solidification problem of the traditional twin method, improve the self-selection ability of the twin system and the flexible application ability of the model. Through the evolvable reinforcement learning method, the core parameters of the scene model are self-learned, the ability to cope with random uncertainty is improved, and the dynamic correction and self-evolution of key models and parameters are realized.

[0090] In some embodiments, the embodiments of the present application build the physical layer of the digital twin by installing corresponding data collection equipment, and the three-dimensional model can be established through three-dimensional geometric modeling software. Sensors and other sensing devices are used to collect the actual operating parameters of the single battery of the physical entity of the energy storage power station, and the system is integrated and standardized to build the data layer of the digital twin, providing a data basis for the operation and optimization of the battery health management system. Among them, the actual operating parameters of the battery include: one or more of the charge and discharge power, charge and discharge voltage, charge and discharge current, charge and discharge quantity, temperature and charge and discharge time. Among them, the charge and discharge power includes: charging power or discharge power. The charge and discharge voltage includes: charging voltage or discharge voltage. The charge and discharge time includes: charging time or discharge time. The charge and discharge time is the actual time used for the energy storage battery to complete a round of charging or a round of discharging. The charge and discharge quantity includes: charging quantity or discharge quantity.

[0091] In some embodiments, the mechanism layer is used to analyze the mapping relationship between the mechanism model of the physical object and the data-driven model of the digital twin. The embodiment of the present application establishes a mathematical or mechanism model for the complex reaction and performance degradation mechanism inside the battery at the mechanism layer, providing a theoretical basis for the prediction of the equipment's operating life and state; the mechanism model is obtained by modeling based on the failure mechanism of the single cell, the evolution law of the "electric-heat-force-gas" characteristics, the battery pack electrical balancing mechanism, etc.

[0092] In some embodiments, the presentation layer analyzes the output presentation form of the digital twin model. The embodiment of the present application depicts the coupling interaction characteristics at the presentation layer to achieve high-fidelity mirroring of the energy storage system by the digital twin, so that on-site personnel can quickly locate the battery fault point and implement repairs based on the information provided by the digital twin.

[0093] In some embodiments, the interaction layer analyzes the interaction mechanism between the digital twin system and the actual system and the user. In the interaction layer, the embodiment of the present application integrates the real-time operating status information of the real system into the forward-looking prediction of the digital twin, and explores the refined virtual-real interaction method of the operation control strategy.

[0094] Finally, a standardized multi-level digital twin architecture for different application scenarios is constructed. And there is a certain connection relationship between the levels. Through the connection relationship between the levels, a digital twin model of the energy storage battery is established to achieve graphical, visual, and interactive virtual presentation, and provide a precisely mapped virtual environment for testing the correctness and reliability of the control and protection functions of the battery energy storage system.

[0095] Example 3:

[0096] refer to Figure 3 , SOC prediction is achieved by data-driven and mechanism fusion methods. This process is divided into offline model construction and online calculation; the SOC prediction model is built through Kalman filter and time series algorithm, Kalman filter performs state estimation, and time series algorithm realizes online SOC prediction.

[0097] In some embodiments, the algorithm used comes from the RNN, LSTM, GRU and other timing algorithms in the algorithm library, which utilizes the timing algorithm's ability to remember the battery's charging and discharging behavior under different working conditions and time and space, improves its ability to cope with uncertain environments and behaviors, and improves the accuracy of its state of charge SOC prediction.

[0098] Specifically, the state estimation value of the Kalman filter and parameters such as the battery SOC current, voltage, and ambient temperature are used as input feature parameters, which reduces the open-loop risk of the neural network to errors and improves the robustness of the model.

[0099] Offline modeling mainly uses current, voltage, temperature and historical SOC values ​​to build an SOC prediction model through time series prediction. The historical aging data such as current, voltage and ambient temperature of the battery under different working conditions are used as the training data set. The training data is time-serialized and input into the Kalman filter to obtain a set of Kalman filter state estimates; then the battery historical aging data and Kalman filter state estimates are normalized and other data preprocessed, and used as the data set to build a time series prediction model offline.

[0100] In some embodiments, the online computing includes a data acquisition module, a state estimation module, and a prediction module.

[0101] Among them, the acquisition module collects characteristic parameters such as current, voltage and ambient temperature that affect the battery SOC. The state estimation module realizes the state estimation of the characteristic parameter input Kalman filter, where the current and voltage are adjusted through the state prediction equation, covariance prediction equation and Kalman gain equation, and the Kalman filter state estimation value is finally obtained through the state update equation and covariance update equation, where the state equation of the Kalman filter is obtained by the battery kinetic model based on the Thevenin equivalent circuit. The prediction module adopts the time series prediction method, using the current, voltage, ambient temperature measurement values ​​and the Kalman filter state estimation value as input characteristic parameters to obtain the battery SOC prediction value.

[0102] In some embodiments, the constructed energy storage battery health management system is suitable for multiple operating scenarios through two-layer optimization. The first layer has a built-in model library and algorithm library, and improves the multi-scenario adaptability through scene adaptation; the second layer builds an intelligent agent (Agent) to automatically select the model and the corresponding hyperparameter combination to achieve self-learning adaptation of the scene model.

[0103] It should be noted that the intelligent agent aims to maximize the accuracy of the model, with the accuracy as the reward value (reward), and continuously learns through the reinforcement learning algorithm until the optimal model and hyperparameter combination are found. The model correction is achieved through the optimization of the model hyperparameter combination, and the SOC online prediction is realized. At the same time, the model library model is adaptively updated through feedback, and the model closed loop is realized.

[0104] It is understandable that the battery health status monitoring and prediction method is also applied to different levels such as battery cells, modules, and clusters. The battery health management model library is constructed by offline training and modeling of historical data from multiple scenarios and multiple objects.

[0105] Specifically, state diagnosis includes SOH analysis and loss analysis. Among them, SOH analysis involves the calculation of energy storage SOH decay trend and battery equivalent cycle number. According to the battery failure mechanism, the loss analysis includes reversible capacity loss and irreversible capacity loss, among which the evaluation of irreversible capacity loss is based on data supervised learning of the evaluation analysis model, and the main indicator parameters include loss data size, cycle stability performance index, etc.

[0106] Predictive maintenance uses intelligent algorithms to predict status parameters. Based on system operation data, real-time equipment operation parameters and other information, it predicts the future development trend of the equipment, predicts the battery life and health status, and supports more accurate predictive maintenance.

[0107] In some embodiments, the battery health assessment implements consistency analysis and assessment of voltage, temperature, current, SOC, etc.

[0108] The voltage consistency analysis is mainly aimed at cluster-level voltage. The fuzzy comprehensive evaluation model is used to evaluate the voltage consistency. The actual single-cell voltage in the operating condition before and after the charging cutoff and the discharging cutoff is selected for extreme consistency analysis. The consistency evaluation is performed from the dimensions of voltage extreme difference and standard deviation after voltage normalization. For example, the excellent stability parameter is set to a range of 50mv and a standard deviation of 0.002. The parameter settings need to be set according to the actual object.

[0109] For the battery cluster temperature consistency analysis, the operating period used in the voltage consistency analysis is selected, and the temperature extremes and standard deviations of the acquisition nodes close to the end conditions are evaluated for consistency.

[0110] For current consistency analysis, the battery stack is selected as the basic unit, the maximum current difference in each stack is counted, the maximum current difference of each stack is visualized, and the current difference of each stack is large or exceeds the set threshold, and then the analysis is carried out at the cluster level.

[0111] SOC consistency analysis enables automatic selection of clusters with large SOC differences as key focus clusters. Based on inter-cluster SOC monitoring analysis, key cluster verification analysis, and comparative analysis, the average error is within 3%.

[0112] Example 4:

[0113] refer to Figure 2 It can be seen that the embodiment of the present application is implemented based on the three parts of the digital twin five-level architecture, the collaborative optimization decision-making architecture and the interactive interface 400 design.

[0114] Specifically, the collaborative optimization architecture is built using edge computing technology and cloud network, and is mainly composed of three parts: offline training module, online decision-making module and effect evaluation module.

[0115] The offline training module collects historical data or simulation data based on a variety of machine learning algorithms to conduct offline training and optimization of the digital twin model, thereby improving the expressiveness and adaptability of the digital twin model. Through the offline module, a multi-scenario model library and algorithm library are built to solve the model solidification problem of the traditional twin method, thereby improving the adaptability of the twin system and the flexible application capability of the model.

[0116] The online decision-making module is based on the offline training module, which monitors and analyzes the battery health status in real time online and formulates the optimal decision plan through rapid response.

[0117] The effect evaluation module optimizes and evaluates the results of the online decision-making module by comparing and evaluating with the actual operation data, thereby improving the accuracy of decision-making.

[0118] Through the evolvable reinforcement learning method, self-learning of the core parameters of the scene model can be achieved, the ability to cope with random uncertainties can be improved, and dynamic correction and self-evolution of key models and parameters can be realized.

[0119] The design of interactive interface 400 generally follows the design principles of segmentation, high cohesion, and low coupling, minimizing the degree of integration between systems and modules within a system and reducing operational complexity.

[0120] Interface design follows industry interface specifications to ensure universality, reusability, scalability and consistency.

[0121] Reference Figure 6 As shown, Figure 6 It is a schematic diagram of the system data flow provided by an example of the present application.

[0122] Depend on Figure 6 It can be seen that the embodiments of the present application are composed of processes such as data collection, data transmission, data storage, data calculation and display of calculation results.

[0123] Data transmission supports wired optical fiber and wireless network transmission. The communication hardware supports RS232 and RS485 to network port, RS232 and RS485 to WIFI, and access to multiple platforms. It supports DI input acquisition, long-distance communication, multi-node deployment, data transparent transmission, transmission of real data content, data with address and encrypted transmission to ensure data traceability and security.

[0124] Data storage includes local data storage, edge data storage and cloud data storage.

[0125] Data computing supports cloud-edge collaborative computing. The edge side performs simple algorithm calculations and computing tasks with high real-time requirements, such as SOC computing; the cloud side performs computing tasks with higher algorithm complexity, such as SOH computing. At the same time, the cloud side is responsible for the training of complex models.

[0126] Based on the above architecture and analysis principles, the digital twin virtual model continuously receives the operating parameters of the physical entity of the energy storage battery, timely feedbacks the battery operation status through real-time analysis of the battery health status, and feeds back relevant decision-making strategies to the energy storage battery entity for corresponding health management operations.

[0127] In some embodiments, the embodiments of the present application improve computing power and data storage capacity by combining cloud computing and edge computing. Battery-related data will be measured and transmitted to the cloud to establish a digital twin system of the battery system. The battery health management model will calculate the battery state of charge and evaluate the health state in real time in the digital twin system, and accurately estimate and predict the battery state; the self-selection capability of the twin system and the flexible application capability of the model are improved through the constructed multi-scenario model library and algorithm library, and the ability to cope with random uncertainty is improved through the evolvable reinforcement learning method; this invention is conducive to improving the overall service life and operational stability of large-scale energy storage systems, thereby significantly improving the overall reliability, safety and economic benefits of large-scale energy storage systems.

[0128] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0129] In several embodiments provided in the present application, it should be understood that the disclosed systems, apparatuses and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, 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 apparatuses or units, which can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0130] It should also be understood that the various implementation methods provided in the embodiments of the present application can be combined arbitrarily to achieve different technical effects.

[0131] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions without violating the spirit of the present application. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. The energy storage battery health management system based on digital twin is characterized by: include: A demand analysis module, used to determine the operation and maintenance demand parameters of the equipment in the energy storage system, and determine the demand factor framework according to the operation and maintenance demand parameters of the energy storage system; The demand factor framework represents the result of classifying the importance of the operation and maintenance demand parameters; A digital twin model construction module is used to determine the operation scenario of the energy storage system according to the demand factor framework, and match the device simulation model that meets the operation scenario and the hyperparameter combination of the device simulation model from the model library to obtain the digital twin model of the energy storage system by fusion; A data analysis module is used to receive real-time operation feedback data of the energy storage system, and output virtual simulation data and health assessment results corresponding to the real-time operation feedback data based on the digital twin model.

2. The energy storage battery health management system according to claim 1, characterized in that: The energy storage battery health management system also includes a model optimization module, which is used to optimize the device simulation model and the hyperparameter combination through a reinforcement learning algorithm based on the difference between the real-time operation feedback data and the virtual simulation data, with the goal of maximizing the accuracy of the model, and update the model library according to the optimized results.

3. The energy storage battery health management system according to claim 1, characterized in that: The digital twin model construction module is also used to build a new equipment simulation model based on the mechanism model of the energy storage battery and the operation and maintenance demand parameters corresponding to the demand factor framework if a device simulation model that meets the operation scenario cannot be matched from the model library, and add the new equipment simulation model to the model library.

4. The energy storage battery health management system according to claim 3, characterized in that: The digital twin model construction module is also used to add an operation scenario label to the new device simulation model, and the operation scenario label corresponds to the operation scenario of the energy storage system; The digital twin model construction module is also used to determine the operation scenario label according to the demand factor framework, automatically match the equipment simulation model that meets the operation scenario from the model library according to the operation scenario label, and automatically set the hyperparameter combination of the equipment simulation model.

5. The energy storage battery health management system according to claim 3, characterized in that: The architecture of the digital twin model includes a physical layer, a data layer, a mechanism layer, a presentation layer and an interaction layer; the physical layer is a physical bottom layer of the digital twin built according to the data acquisition components installed in the energy storage system, the data layer is a data processing layer that performs data integration and normalization on the real-time operation feedback data, the mechanism layer is a model layer that analyzes the mapping relationship between the mechanism model of the energy storage battery and the data-driven model of the digital twin, the presentation layer is a display interface of the digital twin model for the mirror data of the energy storage system, and the interaction layer is a user interaction interface for the health management of the energy storage battery.

6. The energy storage battery health management system according to claim 1, characterized in that: The demand analysis module is specifically used to convert the operation and maintenance demand parameters into demand elements of a preset demand analysis model, classify the attributes of the demand elements according to the demand analysis model, and sort the importance of the demand elements through a hierarchical analysis method, and then adjust the importance of the demand elements after importance sorting according to the quality function deployment method to obtain a demand element framework.

7. The energy storage battery health management system according to claim 1, characterized in that: The digital twin model includes an SOC prediction model, which is used to obtain characteristic parameters that affect the SOC of the energy storage battery in the real-time operation feedback data, input the characteristic parameters into a Kalman filter for state estimation, obtain a state estimation value, and then use the characteristic parameters and the state estimation value as input parameters for prediction of a timing algorithm to obtain a SOC prediction value of the energy storage battery; wherein the timing algorithm is selected from a preset algorithm library.

8. The energy storage battery health management system according to claim 7, characterized in that: The SOC prediction model is constructed by offline modeling; the construction method is to obtain historical aging data of the energy storage battery, use the historical aging data to construct a training data set, and after time series processing, the training data set is input into a Kalman filter, and adjusted through a state prediction equation, a covariance prediction equation and a Kalman gain equation, and a state estimation value of the training data set is obtained through a state update equation and a covariance update equation, and then the SOC prediction model is constructed according to the historical aging data and the state estimation value of the training data set; wherein the historical aging data includes the current, voltage and historical SOC value of the energy storage battery, and the state equation of the Kalman filter is obtained by a battery electrical model based on a Thevenin equivalent circuit.

9. The energy storage battery health management system according to claim 1, characterized in that: The energy storage battery health management system also includes a data interactive sharing model and an interactive interface. The interactive interface is connected to the data acquisition component of the energy storage system to collect data. The data interactive sharing model is used to adjust the data received by the interactive interface based on a predefined data structure to form standardized real-time operation feedback data. The data interactive sharing model is also used to forward the standardized real-time operation feedback data to the digital twin model in real time.

10. An energy storage system, characterized in that: Comprising the energy storage battery health management system as described in any one of claims 1 to 9.

Images