Method and system for united online estimation of electric quantity and health degree of battery

Abstract

The invention discloses a method and a system for united online estimation of the electric quantity and the health degree of a battery, belongs to the technical field of batteries, and solves the problems of low information utilization rate and mutual influence of estimation errors caused by the lack of synergy of united estimation when SOC and SOH are estimated by the existing method. The method comprises the steps that CC-CV charging stage data and working condition circulation discharging stage data of the lithium ion battery are extracted, a battery joint estimation model is constructed, the battery joint estimation model analyzes the CC-CV charging stage data and the working condition circulation discharging stage data of the lithium ion battery, and SOC and SOH estimation values of the lithium ion battery are output. According to the method, the SOC and the SOH of the lithium ion battery are jointly estimated through the battery joint estimation model, the estimation error can be effectively reduced, the accuracy of the battery management system on the battery state is improved, and the battery joint estimation model can better adapt to the change of the dynamic characteristics of the battery, so that the state of the battery is more comprehensively reflected.

Description

Technical Field

[0001] The present invention belongs to the field of battery technology, and particularly relates to a method and system for jointly estimating battery power and health online. Background Art

[0002] With the rapid development of electric vehicles, energy storage systems, and other fields, lithium-ion batteries, as core energy storage units, have become critical technologies for ensuring system safety, extending service life, and optimizing energy efficiency. Accurately estimating the battery's state of charge (SOC) and state of health (SOH) is a core function of battery management systems, directly impacting operational safety, range prediction accuracy, and the economic viability of the battery's entire lifecycle. SOC, representing the percentage of remaining available battery capacity, is the parameter of greatest user interest. Among traditional SOC estimation methods, the open-circuit voltage (OCV) method is widely used due to its simple principle. However, it relies on the battery remaining stationary for a long period of time to achieve a stable open-circuit voltage, a condition difficult to achieve in real-world operating conditions. The battery's state of health (SOH) reflects the degree of overall performance degradation and is typically quantified through metrics such as capacity decay or internal resistance growth. The capacity decay method, currently the most direct SOH assessment method, determines the state of health by comparing the ratio of current capacity to initial capacity. However, this method requires at least one complete charge-discharge cycle to obtain valid capacity data.

[0003] To improve SOC estimation accuracy under dynamic conditions, model-based methods have become a research hotspot. Equivalent circuit models (ECMs) simulate the dynamic characteristics of batteries using components such as resistors and capacitors. Combined with optimal estimation algorithms such as Kalman filtering, these methods can effectively compensate for model errors. However, ECM parameters are susceptible to factors such as temperature and aging, requiring frequent calibration. Furthermore, existing models lack dynamic adaptability to battery aging, resulting in SOC drift during long-term cycling.

[0004] Data-driven approaches achieve SOC estimation by mining implicit patterns in historical data, demonstrating excellent performance in controlled laboratory environments. However, these approaches rely heavily on high-quality, diverse training data, and their generalization capabilities are limited by the data coverage. Furthermore, the incompleteness and noise of data collection in real-world operating conditions significantly reduce their robustness, making them difficult to meet the reliability requirements of industrial scenarios.

[0005] The internal resistance growth method assesses SOH by measuring changes in the battery's internal resistance, which is directly related to battery aging. However, battery internal resistance consists of ohmic resistance and polarization resistance. Polarization resistance is significantly affected by factors such as current rate and temperature. On-site measurement requires specialized equipment and is susceptible to interference from operating conditions, making it difficult to obtain low-cost online data. In recent years, incremental capacity analysis (ICA) has emerged as a method for extracting SOH information by using differential capacity curve characteristics. While superior to traditional methods, it places extremely high demands on the sampling frequency and stability of the test equipment, making the hardware cost prohibitive for large-scale applications.

[0006] In summary, existing SOC and SOH estimation methods have significant limitations:

[0007] The model-driven method has poor adaptability to dynamic working conditions. It relies on parameter calibration and is difficult to adapt to dynamic factors such as temperature and aging. The data-driven method is limited by data quality and generalization capabilities and cannot maintain stability under complex working conditions.

[0008] The accuracy of full life cycle estimation is insufficient. Traditional methods have weak dynamic tracking capabilities for battery aging, making it difficult to achieve high-precision estimation throughout the entire life cycle (especially in the early aging stage).

[0009] The synergy of joint estimation is lacking. Existing studies mostly focus on estimating SOC or SOH separately, without fully utilizing the coupling relationship between the two, resulting in low information utilization and mutual influence of estimation errors.

[0010] High-precision algorithms are difficult to deploy. Although high-precision methods such as electrochemical models and deep learning can improve estimation performance, they have high computational complexity and high hardware costs, making them difficult to implement at low cost in vehicles or energy storage systems.

[0011] To address the limitations of existing technologies, we propose a method and system for jointly estimating battery charge and health online. Summary of the Invention

[0012] The purpose of the present invention is to address the shortcomings of the existing technology and provide a method and system for jointly estimating battery power and health online, which solves the problems of the existing methods in estimating SOC and SOH, such as the lack of synergy in the joint estimation and the failure to fully utilize the coupling relationship between the two, resulting in low information utilization and mutual influence of estimation errors.

[0013] The present invention is implemented as follows: a method for jointly estimating battery power and health online, the method comprising:

[0014] S10, performing a verification experiment on the lithium-ion battery, extracting CC-CV charging phase data and working cycle discharge phase data of the lithium-ion battery, and preprocessing the CC-CV charging phase data and working cycle discharge phase data of the lithium-ion battery;

[0015] S20, building a battery joint estimation model based on the preprocessed lithium-ion battery CC-CV charging phase data and the working cycle discharge phase data, and jointly training the battery joint estimation model to output a converged battery joint estimation model;

[0016] S30, obtaining the pre-processed CC-CV charging phase data and the operating cycle discharge phase data of the lithium-ion battery, analyzing the CC-CV charging phase data and the operating cycle discharge phase data of the lithium-ion battery by a battery joint estimation model, and outputting the SOC and SOH estimation values ​​of the lithium-ion battery.

[0017] The CC-CV charging phase data includes voltage, current, temperature, and charge amount data, and the operating cycle discharge phase data includes voltage, current, and temperature data. The CC-CV charging phase data is used for convolutional neural network (CNN) to estimate SOH, and the operating cycle discharge phase data is used for particle swarm optimization deep belief network to estimate SOC.

[0018] The battery joint estimation model uses a convolutional neural network model and a deep belief network as the initial model, and introduces an adaptive extended Kalman or adaptive filter fusion algorithm into the initial model. The pre-built battery joint estimation model includes a SOH estimation module, a SOC estimation module, an SOC fusion module, and an optimization and training module.

[0019] The SOH estimation module is a SOH-CNN architecture based on a convolutional neural network model CNN. The SOH-CNN architecture includes an input layer, convolution layer 1, pooling layer 1, convolution layer 2, pooling layer 2, a fully connected layer, and an output layer. The voltage, current, temperature, and charge capacity data of the CC-CV charging stage data are used as training inputs for the convolutional neural network model CNN. The SOC estimation module introduces an adaptive extended Kalman filter algorithm and an adaptive H-infinity algorithm for SOC estimation. The SOC fusion module is used to define evaluation indicators, and weightedly fuse the evaluation indicators based on adaptive weighting coefficients, and finally output an SOC fusion estimation value.

[0020] When jointly training the battery joint estimation model, an alternating training mechanism is adopted. During training, the parameters of the SOH-CNN architecture are fixed first and the SOC network is trained until convergence. Then the parameters of the SOC estimation module are adjusted and the SOH-CNN architecture is fine-tuned. After each round of training, the coupling performance is evaluated through cross-validation. When the RMSE of the SOC estimation error drops by less than 0.1% for three consecutive rounds, the training is terminated. The battery joint estimation model sets a temperature compensation subnetwork to automatically correct parameter drift in the range of -20℃-60℃.

[0021] The battery joint estimation model analyzes the data of the CC-CV charging stage and the working cycle discharge stage of the lithium-ion battery, including:

[0022] Obtain lithium-ion battery CC-CV charging stage data and working cycle discharge stage data, and extract lithium-ion battery CC-CV charging stage data;

[0023] Wavelet transform is used to denoise the data of the CC-CV charging phase of the lithium-ion battery. The denoised data is then input into the SOH estimation module, and the output SOC estimate is compared with the actual SOC estimate to verify the effectiveness of the joint estimation algorithm based on multi-method fusion.

[0024] Extract the working condition cycle discharge phase data, use the working condition cycle discharge phase data as input, execute the SOC estimation module and the SOC fusion module, the SOC estimation module and the SOC fusion module identify and analyze the working condition cycle discharge phase data and output the SOC estimation value.

[0025] On the other hand, the present invention also provides a system for jointly estimating battery power and health online, the system comprising:

[0026] The data preprocessing module is used to conduct verification experiments on lithium-ion batteries, extract the CC-CV charging phase data and the working cycle discharge phase data of lithium-ion batteries, and preprocess the CC-CV charging phase data and the working cycle discharge phase data of lithium-ion batteries;

[0027] The model training module builds a battery joint estimation model based on the pre-processed lithium-ion battery CC-CV charging phase data and operating cycle discharge phase data, and jointly trains the battery joint estimation model to output a converged battery joint estimation model;

[0028] The online estimation module is used to obtain the pre-processed CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery. The battery joint estimation model analyzes the CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery and outputs the SOC and SOH estimated values ​​of the lithium-ion battery.

[0029] The online estimation module includes:

[0030] A data extraction unit is used to obtain the CC-CV charging stage data and the working cycle discharge stage data of the lithium-ion battery, and extract the CC-CV charging stage data and the working cycle discharge stage data of the lithium-ion battery;

[0031] The anti-interference processing unit performs noise reduction processing on the data of the CC-CV charging phase of the lithium-ion battery based on the wavelet transform method, and performs anti-interference processing on the data of the working cycle discharge phase based on the adaptive extended Kalman or adaptive filter fusion algorithm;

[0032] The joint estimation unit uses the battery joint estimation model to analyze the CC-CV charging stage data and the working cycle discharge stage data of the lithium-ion battery, and outputs the SOC and SOH estimation values ​​of the lithium-ion battery.

[0033] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0034] In an embodiment of the present invention, the SOC and SOH of a lithium-ion battery are jointly estimated through a battery joint estimation model, which can effectively reduce the estimation error and improve the accuracy of the battery management system in determining the battery status. The battery joint estimation model can better adapt to changes in battery dynamic characteristics and fully integrate multiple information sources such as battery voltage, current, and temperature, thereby more comprehensively reflecting the battery status.

[0035] In an embodiment of the present invention, the battery joint estimation model adopts a hierarchical architecture design of an SOH estimation module, an SOC estimation module, and an SOC fusion module, thereby realizing periodic updates of the differences between SOC and SOH, and realizing aging adaptation through a dynamic noise covariance matrix. At the same time, the EKF algorithm is introduced into the model to process the nonlinear characteristic data of the battery system, which can better adapt to changes in the battery dynamic characteristics, thereby improving the accuracy of state estimation.

[0036] The combined online battery charge and health estimation system provided by this invention can improve the operating efficiency of energy storage systems and significantly optimize their performance by monitoring battery status parameters in real time. It can ensure precise control of the charging and discharging processes, thereby avoiding energy waste. The SOH tracking function enables real-time assessment of battery degradation, which can help the system dynamically adjust its operating strategy. At the same time, the combined online battery charge and health estimation system, based on accurate health status assessment, ensures that the battery management system can implement intelligent charging and discharging management. This allows the battery management system to effectively slow down battery degradation by dynamically adjusting the depth of discharge (DoD) and charge rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a schematic diagram of the implementation process of a method for jointly estimating battery power and health online provided by the present invention.

[0038] Figure 2 The architecture diagram of the battery joint estimation model is shown.

[0039] Figure 3 The figure shows the structural diagram of the joint online estimation system of battery power and health. DETAILED DESCRIPTION

[0040] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art to which this application belongs. The terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit this application. The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0041] When existing methods estimate SOC and SOH, the synergy of joint estimation is lacking, and the coupling relationship between the two is not fully utilized, resulting in low information utilization and mutual influence of estimation errors. To address the above problems, we propose a method and system for jointly estimating battery power and health online. In short, when implementing the method, a verification experiment is first conducted on the lithium-ion battery, and the CC-CV charging stage data and the operating cycle discharge stage data of the lithium-ion battery are extracted. The CC-CV charging stage data and the operating cycle discharge stage data of the lithium-ion battery are preprocessed, and then a battery joint estimation model is constructed based on the preprocessed lithium-ion battery CC-CV charging stage data and the operating cycle discharge stage data. Finally, the battery joint estimation model analyzes the CC-CV charging stage data and the operating cycle discharge stage data of the lithium-ion battery, and outputs the SOC and SOH estimation values ​​of the lithium-ion battery. In an embodiment of the present invention, the SOC and SOH of a lithium-ion battery are jointly estimated through a battery joint estimation model, which can effectively reduce the estimation error and improve the accuracy of the battery management system in determining the battery status. The battery joint estimation model can better adapt to changes in battery dynamic characteristics and fully integrate multiple information sources such as battery voltage, current, and temperature, thereby improving information utilization and avoiding the mutual influence of estimation errors, thereby more comprehensively reflecting the battery status.

[0042] The embodiment of the present invention provides a method for jointly estimating battery power and health online. Figure 1 A schematic diagram of the implementation process of the method for jointly estimating battery power and health online is shown. The method for jointly estimating battery power and health online specifically includes:

[0043] S10, performing a verification experiment on the lithium-ion battery, extracting CC-CV charging phase data and working cycle discharge phase data of the lithium-ion battery, and preprocessing the CC-CV charging phase data and working cycle discharge phase data of the lithium-ion battery;

[0044] It should be noted that the CC-CV charging phase data includes but is not limited to voltage, current, temperature, and charge capacity data, while the operating cycle discharge phase data includes voltage, current, and temperature data. The CC-CV charging phase data is used by the convolutional neural network (CNN) to estimate SOH, while the operating cycle discharge phase data is used by the particle swarm optimization (PSO) deep belief network (DBN) to estimate SOC. The training input data for SOC estimation is voltage, current, temperature, and the first-phase SOH estimate. Simultaneously, the charge capacity during the CC-CV charging phase is extracted, and the actual SOC value calculated using the ampere-hour integration method is used as the training output. For the current verification operating conditions, the SOH estimate used for lithium-ion battery SOC estimation at each moment remains unchanged.

[0045] S20, building a battery joint estimation model based on the preprocessed lithium-ion battery CC-CV charging phase data and the working cycle discharge phase data, and jointly training the battery joint estimation model to output a converged battery joint estimation model;

[0046] In the embodiment of the present invention, Figure 2 The architecture diagram of the battery joint estimation model is shown. The battery joint estimation model uses a convolutional neural network model and a deep belief network as the initial model, and introduces an adaptive extended Kalman or adaptive filter fusion algorithm into the initial model. The pre-built battery joint estimation model includes an SOH estimation module, an SOC estimation module, an SOC fusion module, and an optimization and training module.

[0047] The battery joint estimation model also includes an SOH filter and an SOC filter. The SOH filter introduces an SOC-dependent observation matrix, establishes a bidirectional correction mechanism, and outputs a dynamic noise covariance matrix for parameter adaptation. The SOC filter state equation embeds the SOH compensation term, and the observation equation integrates voltage, current, temperature, and SOH estimation values.

[0048] It should be noted that the SOH estimation module is a SOH-CNN architecture based on the convolutional neural network model CNN. The SOH-CNN architecture includes an input layer, a convolutional layer 1, a pooling layer 1, a convolutional layer 2, a pooling layer 2, a fully connected layer and an output layer. The input layer, the convolutional layer 1, the pooling layer 1, the convolutional layer 2, the pooling layer 2, the fully connected layer and the output layer are connected in sequence. The voltage, current, temperature and charging amount data of the CC-CV charging stage data are used as the training input of the convolutional neural network model CNN. The adaptive extended Kalman filter algorithm and the adaptive H-infinity algorithm are introduced into the SOC estimation module for SOC estimation. The SOC fusion module is used to define evaluation indicators and weightedly fuse the evaluation indicators based on adaptive weighting coefficients, and finally output the SOC fusion estimation value. In an embodiment of the present invention, the battery joint estimation model adopts a hierarchical architecture design of an SOH estimation module, an SOC estimation module, and an SOC fusion module, thereby realizing periodic updates of the differences between SOC and SOH, and realizing aging adaptation through a dynamic noise covariance matrix. At the same time, the EKF algorithm is introduced into the model to process the nonlinear characteristic data of the battery system, which can better adapt to changes in the battery dynamic characteristics, thereby improving the accuracy of state estimation.

[0049] S30, obtaining the pre-processed CC-CV charging phase data and the operating cycle discharge phase data of the lithium-ion battery, analyzing the CC-CV charging phase data and the operating cycle discharge phase data of the lithium-ion battery by a battery joint estimation model, and outputting the SOC and SOH estimation values ​​of the lithium-ion battery.

[0050] In an embodiment of the present invention, the SOC and SOH of a lithium-ion battery are jointly estimated through a battery joint estimation model, which can effectively reduce the estimation error and improve the accuracy of the battery management system in determining the battery status. The battery joint estimation model can better adapt to changes in battery dynamic characteristics and fully integrate multiple information sources such as battery voltage, current, and temperature, thereby more comprehensively reflecting the battery status.

[0051] In an embodiment of the present invention, when the battery joint estimation model is jointly trained, an alternating training mechanism is adopted. During training, the parameters of the SOH-CNN architecture are first fixed to train the SOC network until convergence, and then the parameters of the SOC estimation module are adjusted, and the SOH-CNN architecture is fine-tuned. The coupling performance is evaluated by cross-validation after each round of training. The training is terminated when the SOC estimation error RMSE decreases by less than 0.1% for three consecutive rounds. The battery joint estimation model sets a temperature compensation subnetwork to automatically correct parameter drift within the range of -20℃-60℃. The temperature compensation subnetwork establishes a temperature-parameter mapping model to perform real-time correction of the battery internal resistance and capacity attenuation coefficient within the range of -20℃-60℃, thereby ensuring estimation accuracy over a wide temperature range.

[0052] An embodiment of the present invention provides a method for analyzing CC-CV charging phase data and operating cycle discharge phase data of a lithium-ion battery using a battery joint estimation model. The method for analyzing CC-CV charging phase data and operating cycle discharge phase data of a lithium-ion battery using a battery joint estimation model specifically includes:

[0053] S101, obtaining CC-CV charging phase data and working cycle discharge phase data of the lithium-ion battery, and extracting CC-CV charging phase data of the lithium-ion battery;

[0054] S102, using a wavelet transform method to perform noise reduction on the CC-CV charging phase data of the lithium-ion battery, then inputting the noise-reduced CC-CV charging phase data into a SOH estimation module, outputting an estimated SOC value and comparing it with the actual SOC estimate value to verify the effectiveness of the joint estimation algorithm based on multi-method fusion;

[0055] For the CC-CV charging stage data of lithium-ion batteries, relevant data is first extracted, and the wavelet transform method is used to reduce noise on the data. Wavelet transform has good time-frequency localization characteristics, which can remove noise while retaining the characteristics of useful signals, thereby improving data quality and providing more accurate input for the subsequent SOH estimation module, which helps to improve the accuracy of SOH estimation, and then provide more reliable SOH information for the SOC estimation module. Through the identification and analysis of data by the SOC estimation module and the SOC fusion module, SOC estimation can be performed based on the battery characteristics under actual working conditions. This analysis method based on data from different stages fully considers the differences in battery characteristics under different working conditions. Compared with relying solely on one type of data, it can more comprehensively and accurately reflect the true state of the battery, thereby improving the accuracy of SOC estimation.

[0056] S103, extracting the working condition cycle discharge phase data, taking the working condition cycle discharge phase data as input, executing the SOC estimation module and the SOC fusion module, the SOC estimation module and the SOC fusion module identify and analyze the working condition cycle discharge phase data, and output the SOC estimation value.

[0057] In this embodiment, the battery joint estimation model analyzes lithium-ion battery CC-CV charging phase data and operating cycle discharge phase data, focusing not only on SOC estimation but also SOH estimation. By jointly analyzing and processing data from different phases, it achieves a comprehensive assessment of the battery state. This helps to more accurately understand the battery's health and remaining charge, providing more comprehensive and accurate information for battery management and use. For example, in a battery management system, this information can be used to formulate more reasonable charge and discharge control and maintenance strategies.

[0058] On the other hand, an embodiment of the present invention provides a system for jointly estimating battery power and health online. Figure 3 The structure diagram of the joint online estimation system of battery power and health is shown. The joint online estimation system of battery power and health specifically includes:

[0059] The data preprocessing module 100 is used to perform a verification experiment on the lithium-ion battery, extract the CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery, and preprocess the CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery;

[0060] The model training module 200 constructs a battery joint estimation model based on the pre-processed lithium-ion battery CC-CV charging phase data and the working cycle discharge phase data, and jointly trains the battery joint estimation model to output a converged battery joint estimation model;

[0061] The online estimation module 300 is used to obtain the pre-processed CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery. The battery joint estimation model analyzes the CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery and outputs the SOC and SOH estimated values ​​of the lithium-ion battery.

[0062] It should be noted that the data preprocessing module 100, model training module 200, and online estimation module 300 communicate via Bluetooth, 5G, or DTU. The online estimation module 300 also features state of charge (SOC) and state of health (SOH) tracking. The combined online battery charge and health estimation system provided by the present invention can improve the operating efficiency of energy storage systems by significantly optimizing their performance through real-time monitoring of battery state parameters. The SOC estimation accuracy reaches ±1.5%, ensuring precise control of the charging and discharging processes and thus avoiding energy waste. The SOH tracking function enables real-time assessment of battery degradation, enabling the system to dynamically adjust its operating strategy. Actual project data shows that it can increase the overall efficiency of the energy storage system by 2-3 percentage points, control inter-pack capacity variation within ±1.5%, and significantly enhance system output stability. Furthermore, the combined online battery charge and health estimation system, based on accurate SOH assessments, enables intelligent charge and discharge management within the battery management system. This allows the battery management system to effectively mitigate battery degradation by dynamically adjusting the depth of discharge (DoD) and charge rate. Actual operating data shows that the battery management system using this technology can extend the battery cycle life by 15-20%, significantly reduce the frequency of battery replacement and maintenance costs, and extend the battery service life.

[0063] In this embodiment, the online estimation module 300 includes:

[0064] The data extraction unit 310 is used to obtain the CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery, and extract the CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery;

[0065] The anti-interference processing unit 320 performs noise reduction processing on the data of the CC-CV charging phase of the lithium-ion battery based on the wavelet transform method, and performs anti-interference processing on the data of the working cycle discharge phase based on the adaptive extended Kalman or adaptive filter fusion algorithm;

[0066] The joint estimation unit 330 uses the battery joint estimation model to analyze the CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery, and outputs the SOC and SOH estimation values ​​of the lithium-ion battery.

[0067] It should be noted that the combined online estimation technology of the battery charge and health system establishes a multi-layered safety protection system. Specifically, real-time SOC monitoring can trigger overcharge / over-discharge protection within 50ms; SOH analysis can provide 40-60 minutes of advance warning of lithium deposition risks; and a temperature-SOC-SOH multi-parameter fusion algorithm improves the accuracy of thermal runaway prediction to 92%. The system maintains reliable performance even under extreme operating conditions, with an SOC error of less than 2% in low-temperature environments of -30°C. Furthermore, the combined online estimation technology of the battery charge and health system provides data support for operations and maintenance management. Automatically generated health reports include key indicators such as capacity decay thermograms and internal resistance growth curves; support precise location of faulty cells; and enable dynamic adjustment of predictive maintenance cycles. Practical applications have shown that this can improve operation and maintenance response speed by 60% and reduce operation and maintenance costs by over 35%.

[0068] In summary, the present invention provides a method and system for jointly estimating battery power and health online. In an embodiment of the present invention, the SOC and SOH of a lithium-ion battery are jointly estimated through a battery joint estimation model, which can effectively reduce the estimation error and improve the accuracy of the battery management system for the battery status. The battery joint estimation model can better adapt to changes in battery dynamic characteristics and fully integrate multiple information sources such as battery voltage, current, and temperature, thereby more comprehensively reflecting the battery status.

[0069] It should be noted that for the aforementioned embodiments, for simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present invention is not limited by the order of the actions described, because according to the present invention, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0070] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the scope of protection of the invention. Obviously, the embodiments described are only some embodiments of the present invention, rather than all embodiments. Based on these embodiments, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in this field can still combine, add, delete or make other adjustments to the features in the various embodiments of the present invention according to the circumstances without conflict, without making creative work, so as to obtain different other technical solutions that do not deviate from the concept of the present invention in essence, and these technical solutions also fall within the scope of protection of the present invention.

Claims

1. A method for jointly estimating battery charge and health online, characterized in that: The method for jointly estimating battery charge and health online includes: S10, performing a verification experiment on the lithium-ion battery, extracting CC-CV charging phase data and working cycle discharge phase data of the lithium-ion battery, and preprocessing the CC-CV charging phase data and working cycle discharge phase data of the lithium-ion battery; S20, building a battery joint estimation model based on the preprocessed lithium-ion battery CC-CV charging phase data and the working cycle discharge phase data, and jointly training the battery joint estimation model to output a converged battery joint estimation model; S30, obtaining the pre-processed CC-CV charging phase data and the operating cycle discharge phase data of the lithium-ion battery, analyzing the CC-CV charging phase data and the operating cycle discharge phase data of the lithium-ion battery by a battery joint estimation model, and outputting the SOC and SOH estimation values ​​of the lithium-ion battery.

2. The method for jointly estimating battery charge and health online according to claim 1, wherein: The CC-CV charging phase data includes voltage, current, temperature, and charge amount data, and the operating cycle discharge phase data includes voltage, current, and temperature data. The CC-CV charging phase data is used for convolutional neural network (CNN) to estimate SOH, and the operating cycle discharge phase data is used for particle swarm optimization deep belief network to estimate SOC.

3. The method for jointly estimating battery charge and health online according to claim 1, wherein: The battery joint estimation model uses a convolutional neural network model and a deep belief network as the initial model, and introduces an adaptive extended Kalman and adaptive filter fusion algorithm into the initial model. The pre-built battery joint estimation model includes a SOH estimation module, a SOC estimation module, an SOC fusion module, and an optimization and training module.

4. The method for jointly estimating battery charge and health online according to claim 3, wherein: The SOH estimation module is a SOH-CNN architecture based on a convolutional neural network model CNN. The SOH-CNN architecture includes an input layer, a convolutional layer 1, a pooling layer 1, a convolutional layer 2, a pooling layer 2, a fully connected layer, and an output layer. The voltage, current, temperature, and charge capacity data of the CC-CV charging stage data are used as training inputs for the convolutional neural network model CNN. The SOC estimation module introduces an adaptive extended Kalman filter algorithm and an adaptive H-infinity algorithm for joint estimation of SOC and SOH. The SOC fusion module is used to define evaluation indicators, and weightedly fuse the evaluation indicators based on adaptive weighting coefficients, and finally output an SOC fusion estimation value.

5. The method for jointly estimating battery charge and health online according to claim 4, wherein: When jointly training the battery joint estimation model, an alternating training mechanism is adopted. During training, the parameters of the SOH-CNN architecture are fixed first and the SOC network is trained until convergence. Then the parameters of the SOC estimation module are adjusted and the SOH-CNN architecture is fine-tuned. After each round of training, the coupling performance is evaluated through cross-validation. When the RMSE of the SOC estimation error drops by less than 0.1% for three consecutive rounds, the training is terminated. The battery joint estimation model sets a temperature compensation subnetwork to automatically correct parameter drift in the range of -20℃-60℃.

6. The method for jointly estimating battery charge and health online according to claim 4, wherein: The battery joint estimation model analyzes the data of the CC-CV charging stage and the working cycle discharge stage of the lithium-ion battery, including: Obtain lithium-ion battery CC-CV charging stage data and working cycle discharge stage data, and extract lithium-ion battery CC-CV charging stage data; Wavelet transform is used to denoise the data of the CC-CV charging phase of the lithium-ion battery. The denoised data is then input into the SOH estimation module, and the output SOC estimate is compared with the actual SOC estimate to verify the effectiveness of the joint estimation algorithm based on multi-method fusion. Extract the working condition cycle discharge phase data, use the working condition cycle discharge phase data as input, execute the SOC estimation module and the SOC fusion module, the SOC estimation module and the SOC fusion module identify and analyze the working condition cycle discharge phase data and output the SOC estimation value.

7. A system for jointly estimating battery charge and health online, configured to implement the method for jointly estimating battery charge and health online according to any one of claims 1 to 6, characterized in that: The joint online estimation system for battery power and health includes: The data preprocessing module is used to conduct verification experiments on lithium-ion batteries, extract the CC-CV charging phase data and the working cycle discharge phase data of lithium-ion batteries, and preprocess the CC-CV charging phase data and the working cycle discharge phase data of lithium-ion batteries; The model training module builds a battery joint estimation model based on the pre-processed lithium-ion battery CC-CV charging phase data and operating cycle discharge phase data, and jointly trains the battery joint estimation model to output a converged battery joint estimation model; The online estimation module is used to obtain the pre-processed CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery. The battery joint estimation model analyzes the CC-CV charging phase data and the working cycle discharge phase data of the lithium-ion battery and outputs the SOC and SOH estimated values ​​of the lithium-ion battery.

8. The system for jointly estimating battery charge and health online according to claim 7, wherein: The online estimation module includes: A data extraction unit is used to obtain the CC-CV charging stage data and the working cycle discharge stage data of the lithium-ion battery, and extract the CC-CV charging stage data and the working cycle discharge stage data of the lithium-ion battery; The anti-interference processing unit performs noise reduction processing on the data of the CC-CV charging phase of the lithium-ion battery based on the wavelet transform method, and performs anti-interference processing on the data of the working cycle discharge phase based on the adaptive extended Kalman or adaptive filter fusion algorithm; The joint estimation unit uses the battery joint estimation model to analyze the CC-CV charging stage data and the working cycle discharge stage data of the lithium-ion battery, and outputs the SOC and SOH estimation values ​​of the lithium-ion battery.

Images