Online evaluation method for digital energy storage battery pack

Through digital data acquisition and transmission systems and deep learning algorithms, real-time evaluation and intelligent management of energy storage battery packs are realized, real-time and accuracy problems of traditional evaluation methods are solved, and evaluation efficiency and reliability of power systems are improved.

CN119024195APending Publication Date: 2024-11-26LBATTERYCLOUD CO LTD +1
View PDF 0 Cites 12 Cited by

Patent Information

Application Number
CN202411112656.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-14
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The traditional energy storage battery pack evaluation methods have problems such as poor real-time, low accuracy and low evaluation efficiency. It is impossible to detect potential problems in a timely manner and deal with them, and cannot fully reflect the actual performance and health status of the energy storage battery pack.

Method used

The digital data acquisition and transmission system is used to collect the operating parameter data of the energy storage battery pack in real time, and build a data analysis model system through deep learning algorithms, automatically identify abnormal states, predict the change trends of future performance indicators, generate evaluation reports, and realize intelligent monitoring and manage.

Benefits of technology

It improves the accuracy and evaluation efficiency of the working status evaluation of the energy storage battery pack, reduces the complexity of manual operation, and significantly improves the operating efficiency and reliability of the power system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119024195A_ABST
    Figure CN119024195A_ABST
Patent Text Reader

Abstract

The invention provides a digital energy storage battery pack on-line evaluation method, which comprises the steps of establishing a digital data acquisition and transmission system, constructing a data analysis model system by adopting a deep learning algorithm, constructing a decision support system, acquiring information data of voltage, current, temperature and internal resistance of operation of an energy storage battery pack in real time, and evaluating an energy storage battery pack on-line basis. A data analysis model system established through a deep learning algorithm is combined for accurate analysis, the abnormal state of the energy storage battery pack is automatically identified, the variation trend of the future energy attenuation rate, power attenuation rate and cycle life is predicted, the working state of the energy storage battery pack is automatically adjusted, real-time and accurate evaluation of the performance and the state of the energy storage battery pack is achieved, and the energy storage battery pack is protected. The intelligent level of battery pack management is effectively improved, safe and stable operation of a power system is guaranteed, and meanwhile the evaluation efficiency and accuracy are greatly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of battery technology, and in particular to an online evaluation method for a digital energy storage battery pack. Background Art

[0002] With the gradual depletion of fossil energy and the increasing concern of human beings about environmental issues, clean renewable energy generation has expanded rapidly, and energy storage technology has been developed and used on a large scale. Existing energy storage technologies include: mechanical energy storage, electrochemical energy storage, electrical energy storage, thermal energy storage and chemical energy storage. Compared with other energy storage technologies, electrochemical energy storage has good portability and scalability, and can be widely used in application scenarios such as electric vehicles and grid energy storage.

[0003] With the rapid development of technologies such as electric vehicles and smart grids, the performance evaluation and management of energy storage battery packs as their core components are particularly important. The performance stability and safety of energy storage battery packs have become key factors restricting their large-scale application. The traditional energy storage battery pack evaluation method has the following problems:

[0004] (1) Poor real-time performance: Traditional evaluation methods cannot obtain the operating status and performance data of energy storage battery packs in real time, resulting in the inability to detect and deal with potential problems in a timely manner.

[0005] (2) Low accuracy: Traditional evaluation methods are based on simple physical parameters and cannot fully reflect the actual performance and health status of energy storage battery packs.

[0006] (3) Low evaluation efficiency: Traditional evaluation methods require a lot of manual participation. The evaluation process is cumbersome and time-consuming, and cannot meet the needs of modern power systems for efficient management. Summary of the invention

[0007] In view of this, the present invention aims to propose an online evaluation method for a digital energy storage battery pack, which collects the operating parameter data of the energy storage battery pack in real time through a digital data acquisition and transmission system, and uses a deep learning algorithm to build a data analysis model system, automatically identifies the abnormal state of the energy storage battery pack, predicts the change trend of its future performance indicators, and generates an evaluation report, thereby realizing intelligent monitoring and management of the energy storage battery pack and improving the operating efficiency and reliability of the power system.

[0008] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0009] A digital energy storage battery pack online evaluation method comprises the following steps:

[0010] S1: Establish a digital data acquisition and transmission system to collect the operating parameter data of the energy storage battery pack in real time through the sensor network, and transmit the data to the data processing center through the high-speed communication network. The collected operating parameter data includes information data of voltage, current, temperature, and internal resistance;

[0011] S2: Use deep learning algorithms to build a data analysis model system, automatically identify abnormal conditions of energy storage battery packs based on data collected by the digital data acquisition and transmission system, predict future trends in energy attenuation rate, power attenuation rate, and cycle life, and generate an assessment report containing current status, future predictions, performance risks, and maintenance recommendations;

[0012] S3: Build a decision support system, integrate the analysis results of the deep learning algorithm into the decision support system, and automatically adjust the working status of the energy storage battery pack based on the results of the maintenance recommendations in the evaluation report.

[0013] Furthermore, in step S1, the energy storage battery pack is a reconfigurable energy storage battery pack connection mode, and sensors for detecting operating parameter information data of voltage, current, temperature, and internal resistance are set on each energy storage battery pack. The relevant sensors can transmit the collected operating parameters to the cloud database according to the data transmission protocol and the high-speed communication network.

[0014] Furthermore, in step S1, the data transmission protocol adopts the standard data transmission protocol of TCP / IP, and the high-speed communication network is a communication network using wireless communication.

[0015] Furthermore, in step S1, the sensor network performs encryption and authentication when transmitting the operating parameter data of the energy storage battery pack to the data processing center, including the following specific steps:

[0016] S11: Each node in the sensor network regularly collects operating parameter data of the energy storage battery pack;

[0017] S12: Format the collected data and add timestamp and sensor ID data;

[0018] S13: Encrypt the data processed in step S12 using AES or RSA encryption algorithm;

[0019] S14: Generate a digital signature on the encrypted data to ensure that the data has not been tampered with during transmission and to verify the source of the data;

[0020] S15: attaching the generated digital signature to the encrypted data to form a complete data packet;

[0021] S16: The sensor node sends the encrypted data packet with the digital signature to the data processing center via the wireless communication network;

[0022] S17: The data processing center uses the public key to verify the digital signature in the data packet to confirm the integrity of the data and the authenticity of the source. If the signature fails, the data packet is discarded and the relevant data is recorded; if the signature succeeds, enter S18:

[0023] S18: The data processing center uses the corresponding key to decrypt the encrypted data, restores the original energy storage battery pack operating parameter data, and stores the relevant data in a database, which can be called by the data analysis model system in step S2.

[0024] Furthermore, in step S2, a hybrid model system combining a long short-term memory network and a convolutional neural network is used to learn and generate an evaluation report, which specifically includes the following steps:

[0025] S21: cleaning and standardizing the collected energy storage battery group operating parameters;

[0026] S22: Divide the continuous data stream into sequences of fixed length as input to the model system;

[0027] S23: Use convolutional neural networks to extract local features of the input sequence that can capture spatial and temporal patterns in the data;

[0028] S24: The local feature sequence extracted by the convolutional neural network is input into the long short-term memory network to capture the long-term dependencies in the sequence data. The output of the long short-term memory network is sent to the fully connected layer for further processing, and finally generates a prediction result or classification label through the output layer;

[0029] S25: Use the labeled historical data to train the model system, including data in normal and abnormal states, and use the mean square error loss function to optimize the training model system;

[0030] S26: The model system is configured as a regression task. The output layer directly outputs the working status of the energy storage battery pack and predicts future performance indicators. The future performance indicators include the change trend of energy decay rate, power decay rate, and cycle life.

[0031] S27: Based on the prediction results of the model system, an evaluation report including the current status, future predictions, performance risks, and maintenance recommendations is automatically generated.

[0032] Furthermore, in step S25, the Adam or RMSprop optimization algorithm is used to minimize the loss function and update the weights of the model system.

[0033] Furthermore, in step S26, the following steps are included:

[0034] S261: Define the number of neurons in the output layer to match the predicted change trend of the energy storage battery group;

[0035] S262: Set the last layer of the model system to a linear activation layer to ensure that the hidden layer can fully capture the dynamic changes in the time series;

[0036] S263: Collect real label data including energy decay rate, power decay rate and cycle life variation trend, wherein the energy decay rate is the calculated ratio of the energy output of the energy storage battery pack after multiple charge and discharge cycles to the energy output of the initial state, the power decay rate is based on the calculated ratio of the actual output power of the energy storage battery pack under different discharge conditions to the maximum output power in the initial state, and the cycle life variation trend uses the number of cycles required for the battery to degrade to a specific capacity threshold as a label;

[0037] S264: Select mean square error (MSE), root mean square error (RMSE) or mean absolute error (MAE) as the loss function to guide the model system to update parameters;

[0038] S265: Use historical data that contains input features and corresponding performance indicator labels and is annotated to perform model training, calculate the gradient of the loss function with respect to the model weights through a back-propagation algorithm, and use an optimization algorithm to update the model weights, and continuously iterate the training until the termination condition that the loss function value is lower than a preset threshold is met;

[0039] S266: Predict the performance indicators of new energy storage battery pack data. For the energy storage battery pack data at each time step, the model will output the predicted energy decay rate, power decay rate and cycle life change trend.

[0040] Furthermore, in step S26, step S260 is added:

[0041] The category with the highest probability among the voltage, current, temperature, and internal resistance parameters is selected as the prediction result of the model-related parameters. If the probability of the normal state of the relevant parameters exceeds 0.8, the energy storage battery pack is considered to be in a normal state; otherwise, the energy storage battery pack is considered to be in an abnormal state.

[0042] Furthermore, in step S26, the output layer directly outputs the predicted performance indicators. If the energy decay rate is compared with the initial energy, the energy decay exceeds 10%, and the marking state is entered; if the energy decay exceeds 20%, the warning state is entered; if the power decay rate is compared with the initial power, the power decay exceeds 5%, the marking state is entered; if the power decay exceeds 10%, the warning state is entered; the remaining cycle life is predicted by the historical number of cycles and the current performance parameters, and the marking state is entered when the predicted remaining cycle life is lower than 80% of the design life, and the warning state is entered when the predicted remaining cycle life is lower than 50% of the design life; wherein, when a single performance indicator is in the marking state, the first level warning is entered; if at least two performance indicators are in the marking state or one performance indicator is in the warning state, the second level warning is entered; if at least two performance indicators are in the warning state, the third level warning is entered; in step S26, if the energy storage battery pack is identified as abnormal, the third level warning state is directly entered.

[0043] Furthermore, in the first-level warning state, the monitoring frequency is strengthened and data changes are recorded; in the second-level warning state, the energy storage battery pack connection and cooling system are checked, the load distribution is adjusted, and balanced charging or discharging is performed to ensure that the energy storage battery pack operates within a safe range; in the third-level warning state, the output power of the energy storage battery pack is limited, the backup power supply is started, and preparations are made to isolate the faulty energy storage battery pack for professional maintenance.

[0044] Compared with the prior art, the digital energy storage battery pack online evaluation method of the present invention has the following advantages:

[0045] (1) The digital energy storage battery pack online evaluation method described in the present invention realizes the real-time collection, accurate analysis and intelligent decision-making of the operating parameters of the energy storage battery pack by integrating a digital data acquisition and transmission system, a deep learning algorithm to construct a data analysis model system and a decision support system. Through the deep learning algorithm, the deep mining of the collected data can more accurately identify the abnormal state of the energy storage battery pack and predict its future performance degradation trend, providing users with targeted maintenance suggestions, which not only improves the accuracy of the energy storage battery pack working status evaluation, but also reduces the complexity of manual operation through automation, and significantly improves the evaluation efficiency.

[0046] (2) The digital energy storage battery pack online evaluation method described in the present invention realizes real-time and accurate evaluation of the performance and status of the energy storage battery pack by collecting the energy storage battery pack operation data in real time and combining it with a deep learning algorithm for precise analysis. It effectively improves the intelligent level of battery pack management, ensures the safe and stable operation of the power system, and greatly improves the evaluation efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0048] Figure 1 It is a structural schematic diagram of a reconfigurable switch circuit of an energy storage battery pack according to an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of the structure of the energy storage battery array switch according to an embodiment of the present invention;

[0050] Figure 3 It is a logic block diagram of the online evaluation method of the digital energy storage battery pack according to the embodiment of the present invention;

[0051] Figure 4 A logic block diagram of a digital data acquisition and transmission system established in the online evaluation method for a digital energy storage battery pack according to an embodiment of the present invention;

[0052] Figure 5 A logic block diagram of a data analysis model system constructed in the online evaluation method for a digital energy storage battery pack according to an embodiment of the present invention;

[0053] Figure 6 A logical block diagram of the performance indicator output for prediction by constructing a data analysis model system according to an embodiment of the present invention;

[0054] Description of reference numerals:

[0055] Positive main line 1, negative main line 2, energy storage battery group 3, connecting branch line 4, access switch 5, isolation branch line 6, isolation switch 7, series-parallel switching switch 8. DETAILED DESCRIPTION

[0056] In order to make the technical means, objectives and effects of the present invention easy to understand, the embodiments of the present invention are described in detail below with reference to specific drawings.

[0057] It should be noted that all the terms used in the present invention for directional and positional indications, such as "up", "down", "left", "right", "front", "back", "vertical", "horizontal", "inside", "outside", "top", "low", "lateral", "longitudinal", "center", etc., are only used to explain the relative positional relationship and connection status between the components in a certain state (as shown in the accompanying drawings), and are only for the convenience of describing the present invention, rather than requiring the present invention to be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes, and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features.

[0058] In the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0059] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0060] In traditional energy storage battery packs, the fixed connection mode causes the performance of the energy storage battery pack to depend on the battery cell with the worst performance, which often leads to problems such as overcharge and overdischarge of the battery cell. Therefore, a reconfigurable energy storage battery pack is proposed to solve this problem.

[0061] As Figures 1-2 As shown, the reconfigurable energy storage battery pack includes a plurality of energy storage battery packs 3, each of which includes a plurality of battery cells. The plurality of energy storage battery packs 3 form a reconfigurable circuit with a DC bus through a battery module switch array. The DC bus includes a positive main line 1 and a negative main line 2. An energy storage battery pack 3 is arranged between the positive main line 1 and the negative main line 2. The energy storage battery pack 3 is connected between the positive main line 1 and the negative main line 2 through a connecting branch line 4. An access switch 5 is arranged on the connecting branch line 4. An isolation branch line 6 connected in parallel is arranged at opposite ends of each of the connecting branches 4 between the positive main line 1 and the negative main line 2. An isolating switch 7 is provided on the positive main line 1 and the negative main line 2 with a series-parallel switching switch 8. The battery module switch array changes the series-parallel relationship between the energy storage battery groups 3 according to the on-off of the switch array. The battery module switch array adopts multi-level, fine-grained discrete digital processing and control of battery energy flow to achieve a control system that is relatively independent of the physical and chemical properties of the battery, and dynamically adjusts the series-parallel connection between battery cells. By turning on and off different switches in the above-mentioned battery module switch array, a plurality of the energy storage battery groups 3 and the DC bus constitute a reconfigurable circuit. The specific connection relationship has been specifically recorded in the patent document previously disclosed by the applicant and will not be repeated here.

[0062] like Figures 1 to 6 As shown, the present application discloses a method for online evaluation of a digital energy storage battery pack, comprising the following steps:

[0063] S1: Establish a digital data acquisition and transmission system to collect the operating parameter data of the energy storage battery pack in real time through the sensor network, and transmit the data to the data processing center through the high-speed communication network. The collected operating parameter data includes information data of voltage, current, temperature, and internal resistance;

[0064] S2: Use deep learning algorithms to build a data analysis model system, automatically identify abnormal conditions of energy storage battery packs based on data collected by the digital data acquisition and transmission system, predict future trends in energy attenuation rate, power attenuation rate, and cycle life, and generate an assessment report containing current status, future predictions, performance risks, and maintenance recommendations;

[0065] S3: Build a decision support system, integrate the analysis results of the deep learning algorithm into the decision support system, and automatically adjust the working status of the energy storage battery pack based on the results of the maintenance recommendations in the evaluation report.

[0066] The digital energy storage battery pack online evaluation method disclosed in this application is based on a reconfigurable energy storage battery pack system. By establishing a digital data acquisition and transmission system and utilizing a constructed sensor network system, the voltage, current, temperature, internal resistance and other information data collected from the energy storage battery pack are transmitted to a data processing center through a high-speed communication network, thereby achieving real-time acquisition and transmission of the operating parameters of the energy storage battery pack, thereby enabling the timely discovery and processing of potential problems. Since traditional evaluation methods are based on simple physical parameters, they cannot fully reflect the actual performance and health status of the energy storage battery pack. This application uses a deep learning algorithm to build a data analysis model system, and uses a deep learning algorithm to build a data analysis model system to analyze and mine the collected data. Combined with the analysis of the operating data of the energy storage battery pack within a preset period of time in the past, including the changing trends of parameters such as voltage, current, temperature, and internal resistance, it can more accurately and automatically identify abnormal conditions, including overcharge, over-discharge, high temperature, low temperature, abnormal internal resistance, etc., and attach the time, location, and level of abnormal conditions, and predict the future energy decay rate, power decay rate, and cycle life. The trend can be displayed in a performance decay curve or table, which is convenient for users to understand intuitively, and specific maintenance suggestions are given based on its performance trend, including replacing battery cells, adjusting charging and discharging strategies, optimizing the operating environment, etc. Finally, the corresponding evaluation report is generated and output, thereby improving the accuracy of the evaluation of the working state of the energy storage battery pack. In addition, a decision support system is also constructed to integrate the analysis results of the deep learning algorithm into the decision support system, and automatically adjust the working state of the energy storage battery pack according to the maintenance suggestions in the evaluation report, including adjusting the charging and discharging strategy, optimizing energy distribution, etc., realizing the automation and intelligence of the evaluation process, and greatly improving the evaluation efficiency.

[0067] The digital energy storage battery pack online evaluation method described in the present application realizes real-time collection, accurate analysis and intelligent decision-making of the operating parameters of the energy storage battery pack by integrating a digital data acquisition and transmission system, a deep learning algorithm to build a data analysis model system and a decision support system. Through in-depth mining of the collected data by the deep learning algorithm, it can more accurately identify the abnormal state of the energy storage battery pack and predict its future performance degradation trend, providing users with targeted maintenance suggestions, which not only improves the accuracy of the working status evaluation of the energy storage battery pack, but also reduces the complexity of manual operation through automation, and significantly improves the evaluation efficiency.

[0068] As a preferred example of the present application, in step S1, the energy storage battery pack is a reconfigurable energy storage battery pack connection mode, and sensors for detecting operating parameter information data of voltage, current, temperature, and internal resistance are set on each energy storage battery pack. The relevant sensors can transmit the collected operating parameters to the cloud database according to the data transmission protocol and the high-speed communication network. In the example of the present application, a high-precision, low-internal resistance Hall current sensor is set on the positive and negative busbars of the energy storage battery pack to detect the charging and discharging current information of the entire energy storage battery pack; a high-precision, low-drift voltage sensor or voltage measurement module is set between the positive and negative poles of the energy storage battery pack to measure the voltage difference between the two ends of the energy storage battery pack and obtain the voltage information of the energy storage battery pack; a high-precision, fast-response thermistor sensor, thermocouple or digital temperature sensor is set on the surface of the energy storage battery pack or the inner surface of the energy storage battery pack cover to detect the temperature information of the energy storage battery pack; the internal resistance information data of the energy storage battery pack detects the voltage change at both ends of the energy storage battery pack through the DC discharge method, and then calculates the ohmic internal resistance information of the energy storage battery pack.

[0069] The above-mentioned setting realizes real-time monitoring and precise measurement of the operating parameters of the energy storage battery pack by setting sensors specifically for detecting voltage, current, temperature and internal resistance on the reconfigurable energy storage battery pack, and combining high-speed communication networks and cloud databases for data transmission and storage, providing rich and high-quality input data for subsequent deep learning algorithms, enabling the algorithms to more accurately identify abnormal states of energy storage battery packs, predict their future performance degradation trends, and provide a scientific basis for the maintenance and management of energy storage battery packs, providing strong support for the stable operation and optimization of energy storage systems.

[0070] As a preferred example of the present application, in step S1, the data transmission protocol adopts the standard data transmission protocol of TCP / IP, and the high-speed communication network is a communication network using wireless communication. In the present application, through the confirmation mechanism, flow control, error detection and retransmission mechanism of TCP / IP, the errors and packet loss in the data transmission process are reduced, and the accuracy and reliability of data transmission are guaranteed. At the same time, a wireless communication network that is not restricted by physical lines is used as a network carrier for data transmission, so that the data collection points of the energy storage battery pack can be flexibly arranged, without worrying about the laying and maintenance of cables, and can adapt to complex and changing environments to ensure real-time transmission of data. At the same time, the combination of TCP / IP protocol and wireless communication network ensures the real-time and accuracy of the operating parameter information of the energy storage battery pack. These data can be used as the input of the deep learning algorithm to help the data analysis model system better learn and understand the operating rules of the energy storage battery pack and improve the accuracy of prediction and decision-making. In addition, the input of real-time data also helps the algorithm to detect and handle abnormal situations in a timely manner, and improve the safety of the energy storage battery pack.

[0071] As a preferred example of the present application, in step S1, the sensor network performs encryption and authentication when transmitting the operating parameter data of the energy storage battery pack to the data processing center, including the following specific steps:

[0072] S11: Each node in the sensor network regularly collects operating parameter data of the energy storage battery pack;

[0073] S12: Format the collected data and add timestamp and sensor ID data;

[0074] S13: Encrypt the data processed in step S12 using AES or RSA encryption algorithm;

[0075] S14: Generate a digital signature on the encrypted data to ensure that the data has not been tampered with during transmission and to verify the source of the data;

[0076] S15: attaching the generated digital signature to the encrypted data to form a complete data packet;

[0077] S16: The sensor node sends the encrypted data packet with the digital signature to the data processing center via the wireless communication network;

[0078] S17: The data processing center uses the public key to verify the digital signature in the data packet to confirm the integrity of the data and the authenticity of the source. If the signature fails, the data packet is discarded and the relevant data is recorded; if the signature succeeds, enter S18:

[0079] S18: The data processing center uses the corresponding key to decrypt the encrypted data, restores the original energy storage battery pack operating parameter data, and stores the relevant data in a database, which can be called by the data analysis model system in step S2.

[0080] Through the above settings, the security of the sensor network when transmitting the operating parameter data of the energy storage battery group to the data processing center is ensured. The integrity and authenticity of the data are protected by encryption and authentication technology, which effectively improves the security of data transmission and reduces the risk of data tampering or forgery. At the same time, it supports the requirements of advanced data analysis model systems for data accuracy, providing more reliable and efficient technical support for the monitoring, analysis and management of energy storage battery groups.

[0081] As a preferred example of the present application, in step S2, a hybrid model system combining a long short-term memory network (LSTM) and a convolutional neural network (CNN) is used to learn and generate an evaluation report, which specifically includes the following steps:

[0082] S21: cleaning and standardizing the collected energy storage battery group operating parameters;

[0083] S22: Divide the continuous data stream into sequences of fixed length as input to the model system;

[0084] S23: Use CNN to extract local features of the input sequence, which can capture the spatial and temporal patterns in the data;

[0085] S24: The local feature sequence extracted by CNN is input into the long short-term memory network (LSTM) to capture the long-term dependencies in the sequence data. The output of LSTM is sent to the fully connected layer for further processing and finally generates the prediction result or classification label through the output layer;

[0086] S25: Use the labeled historical data to train the model system, including data in normal and abnormal states, and use the mean square error loss function to optimize the training model system;

[0087] S26: The model system is configured as a regression task. The output layer directly outputs the working status of the energy storage battery pack and predicts future performance indicators. The future performance indicators include the change trend of energy decay rate, power decay rate, and cycle life.

[0088] S27: Based on the prediction results of the model system, an evaluation report including the current status, future predictions, performance risks, and maintenance recommendations is automatically generated.

[0089] In the above implementation, the collected operating parameters (voltage, current, temperature, internal resistance) of the energy storage battery pack are cleaned and standardized, and the continuous data stream is divided into fixed-length sequences for input into the model system, so as to effectively remove noise and outliers, ensure the quality of input data, and significantly improve the efficiency and accuracy of subsequent model system training; in step S23, CNN is used to process each data sequence, extract local features of the input sequence, effectively capture the spatial and temporal patterns in the data, accurately identify early signs of performance degradation of the energy storage battery pack, and then process the feature sequence extracted by CNN through the LSTM network, so that it can capture long-term dependencies in the input sequence, and accurately predict the energy storage performance. Long-term performance change trend of the battery pack; by adopting a hybrid model system that combines long short-term memory network with convolutional neural network, when training the data analysis model system, the labeled historical data (including normal state data and abnormal state data) is used to train the model system, and the mean square error loss function is combined to ensure that the model system can achieve high prediction accuracy in the regression task. By configuring the regression task of the model system, the abnormal state of the energy storage battery pack can be accurately identified and the future performance change trend can be predicted. Finally, according to the prediction results, an evaluation report containing the current state, future prediction, performance risk and maintenance recommendations is automatically generated, which can help decision makers understand the health status of the energy storage battery pack in a timely manner and formulate effective maintenance strategies.

[0090] In the example of this application, by combining the advantages of CNN and LSTM, a hybrid model system combining the two is established for learning, and the operating parameters of the energy storage battery pack are deeply analyzed and predicted, which improves the data quality and model training efficiency. CNN is used to extract local features, and combined with LSTM to process long-term dependencies, it can accurately identify early signs of performance degradation of energy storage battery packs and predict long-term performance changes. It not only realizes accurate identification of abnormal states of energy storage battery packs, but also predicts future performance indicators, and automatically generates an evaluation report containing current status, future predictions, performance risks and maintenance recommendations, providing decision makers with timely and comprehensive health status information of energy storage battery packs, which helps to formulate effective maintenance strategies and improve the safety and reliability of energy storage systems.

[0091] As a preferred example of the present application, in step S25, the optimization algorithm of Adam or RMSprop is used to minimize the loss function and update the weight of the model system. Both Adam and RMSprop can adaptively adjust the learning rate, so that the model can converge faster during the training process and improve the training efficiency. In the deep learning process, the model system combines the initial weights, learning rates, momentum parameters or decay rates and other hyperparameters in the initialized model system, uses the current model weights and input data to calculate the value of the loss function, calculates the gradient of the loss function with respect to the model weight, updates the learning rate according to the first-order moment estimate and the second-order moment estimate of the gradient, or adjusts the learning rate according to the exponential weighted average of the square of the gradient to implement the optimizer update, uses the learning rate calculated by the optimizer to update the model weight, and repeats the above steps until the termination condition that the loss function value is lower than the preset threshold is met. As a preferred example of this application, the Adam optimization algorithm is used for optimization training. The initialization parameters include the learning rate α, the exponential decay rate β1 of the first-order moment estimate, the exponential decay rate β2 of the second-order moment estimate, a small positive number €, the initialization parameter vector θ, the first-order moment vector m0, and the second-order moment vector v0. In each iteration, the gradient g is calculated. t With respect to the parameter θ t-1 The value of the loss function at time (t) is updated to the first-order moment estimate:

[0092] m t =β1·m t-1 +(1-β1)·g t ;

[0093] Update the second-order moment estimate to:

[0094]

[0095] For m t 、v t Correct the deviation and update the parameter θ t

[0096]

[0097] In the example of this application, the value of α is 0.001, the value of β1 is 0.9, the value of β2 is 0.999, and the value of € is 10. -8 , m0 is 0, v0 is 0, m t is the updated first-order moment estimate at time t, v t is the updated second-order moment estimate at time t, θ t is the updated parameter vector at time t.

[0098] This setting enables the hybrid model system to adaptively adjust the learning rate, thereby helping the model system converge faster during training and reducing training time, enabling the model system to achieve higher prediction accuracy in a shorter time, providing strong support for health status assessment and maintenance strategy formulation of energy storage battery packs.

[0099] As a preferred example of the present application, in step S26, the following steps are included:

[0100] S261: Define the number of neurons in the output layer to match the predicted change trend of the energy storage battery group;

[0101] In the specific example of the present application, the number of neurons in the output layer is set to 3, corresponding to the three predicted performance indicators of energy decay rate, power decay rate and change trend of cycle life respectively.

[0102] S262: Set the last layer of the model system to a linear activation layer to ensure that the hidden layer can fully capture the dynamic changes in the time series;

[0103] S263: Collect real label data including energy decay rate, power decay rate and cycle life variation trend, wherein the energy decay rate is the calculated ratio of the energy output of the energy storage battery pack after multiple charge and discharge cycles to the energy output of the initial state, the power decay rate is based on the calculated ratio of the actual output power of the energy storage battery pack under different discharge conditions to the maximum output power in the initial state, and the cycle life variation trend uses the number of cycles required for the battery to degrade to a specific capacity threshold as a label;

[0104] In the example of the present application, by combining the sensor network to collect the initial data such as voltage, current, temperature, internal resistance of the energy storage battery pack and the measurement data in the subsequent multiple charge and discharge cycles in real time, the energy output of the energy storage battery pack in the initial state, the energy output after multiple charge and discharge cycles, the maximum output power in the initial state, and the actual output power under different discharge conditions are analyzed and calculated. For example, the initial energy output of the energy storage battery pack is calculated by integration using the voltage and current data in the initial state; during each charge and discharge cycle, the voltage, current, temperature and internal resistance data are continuously recorded, and after each charge and discharge cycle, the energy output of the cycle is calculated to obtain the energy output after multiple charge and discharge cycles; by comparing the initial energy output with the energy output after multiple cycles, the capacity decay rate is calculated, and the battery state of health (SOH) is evaluated in combination with the comparison of the initial internal resistance and the internal resistance value after multiple charge and discharge cycles. This part of the content is the prior art that has been disclosed and will not be elaborated on here.

[0105] S264: Select mean square error (MSE), root mean square error (RMSE) or mean absolute error (MAE) as the loss function to guide the model system to update parameters;

[0106] S265: Use historical data that contains input features and corresponding performance indicator labels and is annotated to perform model training, calculate the gradient of the loss function with respect to the model weights through a back-propagation algorithm, and use an optimization algorithm to update the model weights, and continuously iterate the training until the termination condition that the loss function value is lower than a preset threshold is met;

[0107] S266: Predict the performance indicators of new energy storage battery pack data. For the energy storage battery pack data at each time step, the model will output the predicted energy decay rate, power decay rate and cycle life change trend.

[0108] In step S265, the trained model system is evaluated using the validation set, and the model is optimized according to the evaluation results, including adjusting the model structure, adding or reducing features, adjusting hyperparameters, etc., and the optimized model is used to predict the performance indicators of new energy storage battery pack data. The data in the validation set are samples that are completely independent of the training set and the test set, and come from the same overall distribution as the training set, and have the same and correct labels.

[0109] As a preferred example of the present application, the specific capacity threshold recorded in step S263 is 80% of the initial capacity.

[0110] As a preferred example of the present application, for the energy storage battery pack data of each time step, in step S26, step S260 is added:

[0111] The category with the highest probability among the voltage, current, temperature and internal resistance parameters is selected as the prediction result of the model related parameters. If the probability of the normal state of the related parameters exceeds 0.8, the energy storage battery pack is considered to be in a normal state; otherwise, the energy storage battery pack is considered to be in an abnormal state;

[0112] Through the above-mentioned model system configuration, efficient training and accurate prediction of the model are achieved, which provides important support for the health management and maintenance of energy storage battery packs, effectively reduces the risks of false alarms and missed alarms, and ensures the safe and stable operation of the energy storage system.

[0113] In the example of the present application, in step S26, the output layer directly outputs the predicted performance indicators. If the energy decay rate is compared with the initial energy, the energy decay exceeds 10% and enters the marking state, and the energy decay exceeds 20% and enters the warning state; if the power decay rate is compared with the initial power, the power decay exceeds 5% and enters the marking state, and the power decay exceeds 10% and enters the warning state; the remaining cycle life is predicted by the historical number of cycles and the current performance parameters, and when the predicted remaining cycle life is lower than 80% of the design life, the marking state is entered, and when the predicted remaining cycle life is lower than 50% of the design life, the warning state is entered; wherein, a single When one performance indicator is in the marked state, it enters the first-level warning; when at least two performance indicators are in the marked state or one performance indicator is in the warning state, it enters the second-level warning; when at least two performance indicators are in the warning state, it enters the third-level warning; in the first-level warning state, strengthen the monitoring frequency and record data changes; in the second-level warning state, check the energy storage battery group connection and cooling system, adjust the load distribution, and perform balanced charging or discharging to ensure that the energy storage battery group operates within a safe range; in the third-level warning state, limit the output power of the energy storage battery group, start the backup power supply, prepare to isolate the faulty energy storage battery group, and perform professional maintenance. As a preferred example of the present application, in step S26, if the energy storage battery group is identified as abnormal, it directly enters the third-level warning state. Only when the energy storage battery group is detected to be in a normal state in step S260, it enters the prediction of future performance indicators.

[0114] This setting provides a specific maintenance recommendation strategy based on the online evaluation method for digital energy storage battery packs described in this application. By refining the threshold settings of key performance indicators such as energy decay rate, power decay rate and cycle life, and combining a graded early warning mechanism, the health status and performance change trends of the energy storage battery pack can be monitored more accurately, which can not only timely mark and warn of potential performance degradation problems, but also avoid the operation of the energy storage battery pack under extreme conditions, thereby effectively extending the service life of the energy storage battery pack and improving the overall safety and reliability of the system; at the same time, the graded early warning mechanism can take targeted maintenance measures according to different warning levels, which not only avoids unnecessary excessive maintenance, but also ensures that key performance problems are dealt with in a timely manner, providing efficient and accurate decision-making support for operation and maintenance personnel, which not only improves the accuracy and timeliness of online evaluation of energy storage battery packs, but also significantly optimizes the maintenance management strategy of energy storage battery packs, and has significant technical effects and practical application value.

[0115] In the example of the present application, the establishment of a digital data acquisition and transmission system in step S1, the construction of a data analysis model system using a deep learning algorithm in step S2, and the decision support system in step S3 adopt a modular independent design and can be customized according to the information selected by the user. In step S3, a decision support system module is established by combining the automated monitoring system and remote control technology. Various operating parameters of the energy storage battery pack are collected in real time through the sensor network. The collected data is processed and analyzed in real time using the intelligent evaluation system deployed in the data center or the cloud. The data analysis model system constructed using the deep learning algorithm is combined to calculate key performance indicators such as energy decay rate and power decay rate, predict the remaining cycle life, and evaluate the health status of the energy storage battery pack according to the preset threshold. According to the health status evaluation results, the system automatically triggers different levels of warnings and generates corresponding operation and maintenance suggestions. The energy storage battery pack management system needs to have a remote control interface to receive and execute control instructions from the data analysis model system. When the data analysis model system generates operation and maintenance suggestions, the corresponding control instructions are sent to the energy storage battery pack through the remote control interface, including operation instructions such as adjusting load distribution, starting balanced charge and discharge procedures, and isolating faulty energy storage battery packs, so as to automatically adjust the working status of the energy storage battery pack.

[0116] In the example of the present application, the data analysis model system can automatically record detailed information of each operation and maintenance operation, including operation time, operation content, execution results, etc., for subsequent analysis and tracing, and regularly conduct in-depth analysis of operation and maintenance data, evaluate operation and maintenance effects, identify potential problems, and continuously optimize evaluation models and control strategies to improve the system's adaptability and intelligence level.

[0117] 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, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A digital energy storage battery pack online evaluation method, characterized in that: The steps include: S1: Establish a digital data acquisition and transmission system to collect the operating parameter data of the energy storage battery pack in real time through the sensor network, and transmit the data to the data processing center through the high-speed communication network. The collected operating parameter data includes information data of voltage, current, temperature, and internal resistance; S2: Use deep learning algorithms to build a data analysis model system, automatically identify abnormal conditions of energy storage battery packs based on data collected by the digital data acquisition and transmission system, predict future trends in energy attenuation rate, power attenuation rate, and cycle life, and generate an assessment report containing current status, future predictions, performance risks, and maintenance recommendations; S3: Build a decision support system, integrate the analysis results of the deep learning algorithm into the decision support system, and automatically adjust the working status of the energy storage battery pack based on the results of the maintenance recommendations in the evaluation report.

2. The method for online evaluation of a digital energy storage battery pack according to claim 1, characterized in that: In step S1, the energy storage battery pack is a reconfigurable energy storage battery pack connection mode, and sensors for detecting operating parameter information data of voltage, current, temperature, and internal resistance are set on each energy storage battery pack. The relevant sensors can transmit the collected operating parameters to the cloud database according to the data transmission protocol and the high-speed communication network.

3. The method for online evaluation of a digital energy storage battery pack according to claim 2, characterized in that: In step S1, the data transmission protocol adopts the standard data transmission protocol of TCP / IP, and the high-speed communication network is a communication network using wireless communication.

4. The method for online evaluation of a digital energy storage battery pack according to claim 1, 2 or 3, characterized in that: In step S1, the sensor network encrypts and authenticates the operating parameter data of the energy storage battery pack when transmitting it to the data processing center, including the following specific steps: S11: Each node in the sensor network regularly collects operating parameter data of the energy storage battery pack; S12: Format the collected data and add timestamp and sensor ID data; S13: Encrypt the data processed in step S12 using AES or RSA encryption algorithm; S14: Generate a digital signature on the encrypted data to ensure that the data has not been tampered with during transmission and to verify the source of the data; S15: attaching the generated digital signature to the encrypted data to form a complete data packet; S16: The sensor node sends the encrypted data packet with the digital signature to the data processing center via the wireless communication network; S17: The data processing center uses the public key to verify the digital signature in the data packet to confirm the integrity of the data and the authenticity of the source. If the signature fails, the data packet is discarded and the relevant data is recorded; if the signature succeeds, enter S18: S18: The data processing center uses the corresponding key to decrypt the encrypted data, restores the original energy storage battery pack operating parameter data, and stores the relevant data in a database, which can be called by the data analysis model system in step S2.

5. The method for online evaluation of a digital energy storage battery pack according to claim 4, characterized in that: In step S2, a hybrid model system combining a long short-term memory network and a convolutional neural network is used to learn and generate an evaluation report, which specifically includes the following steps: S21: cleaning and standardizing the collected energy storage battery group operating parameters; S22: Divide the continuous data stream into sequences of fixed length as input to the model system; S23: Use convolutional neural networks to extract local features of the input sequence that can capture spatial and temporal patterns in the data; S24: The local feature sequence extracted by the convolutional neural network is input into the long short-term memory network to capture the long-term dependencies in the sequence data. The output of the long short-term memory network is sent to the fully connected layer for further processing, and finally generates a prediction result or classification label through the output layer; S25: Use the labeled historical data to train the model system, including data in normal and abnormal states, and use the mean square error loss function to optimize the training model system; S26: The model system is configured as a regression task. The output layer directly outputs the working status of the energy storage battery pack and predicts future performance indicators. The future performance indicators include the change trend of energy decay rate, power decay rate, and cycle life. S27: Based on the prediction results of the model system, an evaluation report including the current status, future predictions, performance risks, and maintenance recommendations is automatically generated.

6. The method for online evaluation of a digital energy storage battery pack according to claim 5, characterized in that: In step S25, the Adam or RMSprop optimization algorithm is used to minimize the loss function and update the weights of the model system.

7. The method for online evaluation of a digital energy storage battery pack according to claim 5, characterized in that: In step S26, the following steps are included: S261: Define the number of neurons in the output layer to match the predicted change trend of the energy storage battery group; S262: Set the last layer of the model system to a linear activation layer to ensure that the hidden layer can fully capture the dynamic changes in the time series; S263: Collect real label data including energy decay rate, power decay rate and cycle life variation trend, wherein the energy decay rate is the calculated ratio of the energy output of the energy storage battery pack after multiple charge and discharge cycles to the energy output of the initial state, the power decay rate is based on the calculated ratio of the actual output power of the energy storage battery pack under different discharge conditions to the maximum output power in the initial state, and the cycle life variation trend uses the number of cycles required for the battery to degrade to a specific capacity threshold as a label; S264: Select mean square error (MSE), root mean square error (RMSE) or mean absolute error (MAE) as the loss function to guide the model system to update parameters; S265: Use historical data that contains input features and corresponding performance indicator labels and is annotated to perform model training, calculate the gradient of the loss function with respect to the model weights through a back-propagation algorithm, and use an optimization algorithm to update the model weights, and continuously iterate the training until the termination condition that the loss function value is lower than a preset threshold is met; S266: Predict the performance indicators of new energy storage battery pack data. For the energy storage battery pack data at each time step, the model will output the predicted energy decay rate, power decay rate and cycle life change trend.

8. The method for online evaluation of digital energy storage battery pack according to claim 7, characterized in that: In step S26, step S260 is added: The category with the highest probability among the voltage, current, temperature, and internal resistance parameters is selected as the prediction result of the model-related parameters. If the probability of the normal state of the relevant parameters exceeds 0.8, the energy storage battery pack is considered to be in a normal state; otherwise, the energy storage battery pack is considered to be in an abnormal state.

9. The method for online evaluation of a digital energy storage battery pack according to claim 8, characterized in that: In step S26, the output layer directly outputs the predicted performance indicators. If the energy decay rate is compared with the initial energy, the energy decay exceeds 10% and enters the marking state, and the energy decay exceeds 20%, it enters the warning state; if the power decay rate is compared with the initial power, the power decay exceeds 5% and enters the marking state, and the power decay exceeds 10%, it enters the warning state; the remaining cycle life is predicted by the historical number of cycles and the current performance parameters. When the predicted remaining cycle life is lower than 80% of the design life, it enters the marking state, and when the predicted remaining cycle life is lower than 50% of the design life, it enters the warning state; among them, when a single performance indicator is in the marking state, it enters the first-level warning; when at least two performance indicators are in the marking state or one performance indicator is in the warning state, it enters the second-level warning; when at least two performance indicators are in the warning state, it enters the third-level warning; in step S26, if the energy storage battery pack is identified as abnormal, it directly enters the third-level warning state.

10. The method for online evaluation of a digital energy storage battery pack according to claim 9, characterized in that: In the first-level warning state, strengthen the monitoring frequency and record data changes; in the second-level warning state, check the energy storage battery pack connection and cooling system, adjust the load distribution, and perform balanced charging or discharging to ensure that the energy storage battery pack operates within a safe range; in the third-level warning state, limit the output power of the energy storage battery pack, start the backup power supply, prepare to isolate the faulty energy storage battery pack, and conduct professional maintenance.

Citation Information

Cited By

  • Lithium ion battery health state estimation method based on open-circuit voltage curve reconstruction and computer readable storage medium

    CN119936713A

  • A method for estimating the state of health of lithium-ion batteries based on open-circuit voltage curve reconstruction and a computer-readable storage medium.

    CN119936713B

  • Communication and optimization method and system of energy storage system

    CN119946097A

  • Battery energy storage power station state evaluation and intelligent sensing method and system

    CN120065000A

  • Battery energy storage power station status assessment and intelligent perception method and system

    CN120065000B