Energy storage monitoring method of energy storage battery

Abstract

The invention discloses an energy storage monitoring method of an energy storage battery. The method comprises the following steps: acquiring voltage, current, temperature, internal impedance spectrum, gas composition and mechanical stress data of the battery in real time; constructing an electrochemical-thermal-mechanical coupling model based on a physical information neural network, embedding an electrochemical equation into a neural network hidden layer, training the model by using collected data to optimize network weight, and dynamically updating lithium ion concentration distribution and internal resistance in the battery; and calculating the growth rate of the lithium dendrites and the combined gas concentration and mechanical stress by utilizing the trained model and data acquired in real time, and triggering a corresponding alarm mechanism when the growth rate of the lithium dendrites exceeds a threshold value or the gas concentration and the mechanical stress meet preset conditions. Through multi-modal data fusion and physical information embedding, the monitoring accuracy and reliability are improved, meanwhile, the characteristics of dynamic updating and efficient calculation are achieved, the safety risk of the battery can be effectively prevented, and the service life of the battery is prolonged.

Description

Technical Field

[0001] This invention belongs to the field of intelligent monitoring technology for energy storage batteries, and specifically relates to an energy storage monitoring method for energy storage batteries. Background Technology

[0002] With the rapid development of energy storage battery technology, its importance in energy storage and utilization is becoming increasingly prominent. However, various problems may arise during battery use, such as lithium dendrite growth, gas accumulation, and excessive mechanical stress. These problems not only affect battery performance but may also lead to safety accidents. Traditional battery monitoring methods mainly rely on basic parameters such as voltage, current, and temperature. While these methods can provide some information about the battery's state, they are significantly insufficient in early fault detection and safety warnings. In recent years, researchers have begun to explore more comprehensive and refined battery monitoring technologies. For example, electrochemical impedance spectroscopy (EIS) is used to obtain the impedance characteristics inside the battery, gas sensors are used to monitor changes in the gas composition around the battery, and strain gauges or fiber optic sensors are used to measure the mechanical stress distribution of the battery. These technologies provide more multi-dimensional data support for battery state assessment. However, how to effectively integrate and utilize this multi-modal data to achieve accurate assessment and early warning of battery health remains a challenging problem.

[0003] Physical Information Neural Networks (PINNs), as an emerging computational method, embed physical equations into neural networks, thereby satisfying the constraints of physical laws while being data-driven. This method has shown unique advantages in handling complex physical systems. In the field of energy storage battery monitoring, the application of PINNs is expected to improve the model's understanding and prediction capabilities of the battery's internal physicochemical processes, enabling dynamic updates and accurate monitoring of the battery's state. Summary of the Invention

[0004] The present invention aims to at least partially solve one of the technical problems in the related art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: This invention provides an energy storage monitoring method for an energy storage battery, comprising: Step S1: Real-time acquisition of battery voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress data; Step S2: Construct an electrochemical-thermal-mechanical coupling model based on a physical information neural network, embed the electrochemical equation into the hidden layer of the neural network, use the collected data to train the model and optimize the network weights, and dynamically update the lithium ion concentration distribution and internal resistance inside the battery. Step S3: Using the trained model and real-time collected data, calculate the lithium dendrite growth rate and the combined gas concentration and mechanical stress. When the lithium dendrite growth rate exceeds the threshold or the gas concentration and mechanical stress meet the preset conditions, trigger the corresponding alarm mechanism.

[0006] Preferably, step S1 includes: The battery's voltage, current, and temperature data are acquired in real time using high-precision voltage, current, and temperature sensors. High-frequency impedance spectrum data of the battery were acquired using an electrochemical impedance spectroscopy (EIS) instrument in the frequency range of 0.1 Hz to 10 kHz. The impedance spectrum data were then converted from the time domain to the time-frequency domain using a short-time Fourier transform (STFT) to extract the time-frequency characteristics of the impedance spectrum. Gas sensors detect the composition of gases in the environment surrounding the battery, including the concentrations of carbon dioxide and hydrogen. The mechanical stress distribution of the battery is measured using strain gauges or fiber optic sensors.

[0007] Preferably, high-frequency impedance spectrum data of the battery are acquired using an electrochemical impedance spectroscopy (EIS) instrument within the frequency range of 0.1 Hz to 10 kHz. The impedance spectrum data is then transformed from the time domain to the time-frequency domain using a short-time Fourier transform (STFT), and the time-frequency characteristics of the impedance spectrum are extracted, including: High-frequency impedance spectral data of the battery were acquired using an electrochemical impedance spectroscopy (EIS) instrument in the frequency range of 0.1 Hz to 10 kHz. The acquired impedance data undergoes preprocessing, including noise removal and data smoothing, to improve data quality. The Hanning window function is selected to divide the impedance data into multiple short time periods. The length of each frame is set to 256 to 1024 sampling points, and the frames maintain an overlap of 50% to 75%. Perform a Short-Time Fourier Transform (STFT) on each frame of data to obtain the time-frequency representation of that frame. The calculation formula is as follows:

[0008] in, Indicates the input signal. For window functions, For time shift, For frequency; Extract the time-frequency characteristics of amplitude spectrum, phase spectrum and energy spectrum from the impedance spectrum from the short-time Fourier transform (STFT) calculation results.

[0009] Preferably, step S2 includes: A multi-scale neural network model is constructed, and physical equations are embedded into the hidden layer of the multi-scale neural network model to obtain an electrochemical-thermal-mechanical coupling model based on physical information neural network, so that the network can learn data while also satisfying the constraints of physical laws. Electrochemical equations are embedded in the hidden layer of a neural network, including a simplified diffusion equation for lithium-ion concentration:

[0010] in, Indicates lithium-ion concentration. Represents a time variable. This represents the diffusion coefficient of lithium ions. Battery current, It is Faraday's constant. The electrode area of ​​the battery; The internal resistance update equation is embedded in the hidden layer of the neural network to model the change in internal resistance:

[0011] in This represents the internal resistance at time t. The initial internal resistance, is the internal resistance variation coefficient, and k is the thermal conductivity of the battery; The model is trained using collected data on voltage, current, temperature, impedance spectrum, gas composition, and mechanical stress. The network weights are optimized using a loss function to dynamically update the lithium-ion concentration distribution and internal resistance within the battery. Represented as:

[0012] in, , and These are the weighting coefficients. The lithium-ion concentration predicted by the model. To measure the lithium-ion concentration, The internal resistance is predicted by the model. This is the measured internal resistance.

[0013] Preferably, the steps for constructing a multi-scale neural network model specifically include: Acquire massive amounts of voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress data from similar energy storage batteries in advance, and use them as supporting data; After preprocessing the supporting data, data features are extracted from the supporting data at different scales; Principal component analysis or t-SNE dimensionality reduction techniques are used to fuse multi-scale data features into a unified feature space. Construct a multi-scale neural network architecture model to process data of different scales separately, and then fuse them in the intermediate or output layers of the network. Use convolutional neural networks to process spatial features and recurrent neural networks or long short-term memory networks to process time series data. An attention mechanism is introduced into the neural network architecture model to adjust the weights of features at different scales, enabling the model to focus on key features during prediction. By utilizing the fused data features, a multi-scale neural network architecture model is trained to obtain a multi-scale neural network model.

[0014] Preferably, using a trained model and real-time acquired data, the calculation of lithium dendrite growth rate and combined gas concentration and mechanical stress includes: The lithium-ion concentration distribution and ohmic internal resistance at the current time and location are determined using the lithium-ion concentration distribution and internal resistance data predicted by the model. The reference high-frequency imaginary part and sensitivity factor under the current state are determined by looking up a table based on the lithium-ion concentration and ohmic internal resistance, and the internal resistance offset factor is calculated. The dendrite growth rate calibration coefficient is determined based on the sensitivity factor and internal resistance offset factor. Combined with real-time acquired impedance spectral data, the lithium dendrite growth rate is calculated.

[0015] in, Indicates the lithium dendrite growth rate. This represents the change in the imaginary part of impedance in the high-frequency band. This indicates the selected high-frequency band. For time intervals, These are the preset calibration coefficients; Calculate the rate of change of carbon dioxide concentration To detect abnormal changes in gas concentration:

[0016] in, This represents the current carbon dioxide concentration. The carbon dioxide concentration at the previous moment; Calculate the rate of change of maximum mechanical stress To detect abnormal changes in mechanical stress:

[0017] in The maximum mechanical stress at the current moment. This represents the maximum mechanical stress at the previous moment.

[0018] Preferably, when the lithium dendrite growth rate exceeds a threshold or the gas concentration and mechanical stress meet preset conditions, a corresponding alarm mechanism is triggered, specifically including: Pre-set conditions considering the rate of change of gas concentration and the rate of change of mechanical stress will trigger a thermal runaway warning when the rate of change of carbon dioxide concentration exceeds a certain threshold and the rate of change of maximum mechanical stress meets the following conditions:

[0019] in, The stress change rate threshold; Alternatively, a thermal runaway precursor warning may be triggered when the lithium dendrite growth rate exceeds a threshold or when the gas concentration and mechanical stress simultaneously meet the following conditions:

[0020] in, Indicates carbon dioxide concentration. Indicates the maximum mechanical stress. This is the stress threshold.

[0021] The present invention also proposes an energy storage monitoring device for an energy storage battery, comprising: The first execution module is used to acquire real-time data on battery voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress. The second execution module is used to construct an electrochemical-thermal-mechanical coupling model based on a physical information neural network. It embeds the electrochemical equation into the hidden layer of the neural network, uses the collected data to train the model, optimizes the network weights, and dynamically updates the lithium-ion concentration distribution and internal resistance inside the battery. The third execution module is used to calculate the lithium dendrite growth rate and the combined gas concentration and mechanical stress using the trained model and real-time collected data. When the lithium dendrite growth rate exceeds the threshold or the gas concentration and mechanical stress meet the preset conditions, the corresponding alarm mechanism is triggered.

[0022] The present invention has achieved at least the following beneficial effects: 1. This invention improves the accuracy and reliability of monitoring through multimodal data fusion and physical information embedding, while also featuring dynamic updates and efficient calculation, effectively preventing battery safety risks and extending battery life.

[0023] 2. By combining a physical information neural network with a coupled model, the health status of the battery can be predicted and assessed more accurately, effectively preventing potential safety risks and extending the battery's lifespan.

[0024] 3. Combining the advantages of data-driven approaches and physical models, by embedding physical equations into neural networks, the model can not only learn from large amounts of data but also follow known physical laws, improving its accuracy and reliability. Furthermore, the multi-scale neural network architecture can handle data features at different scales, further enhancing the model's performance.

[0025] 4. It can more accurately predict and assess the health status of batteries, and promptly detect potential safety hazards, such as the growth of lithium dendrites. By dynamically updating lithium-ion concentration and internal resistance, the model can reflect the battery's operating status in real time, providing more accurate information to the battery management system, thereby improving battery safety and lifespan.

[0026] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0027] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating the steps of an energy storage monitoring method for an energy storage battery according to an embodiment of the present invention. Figure 2 This is a flowchart illustrating the steps involved in constructing a multi-scale neural network model in an embodiment of the present invention. Figure 3 This is a schematic diagram of the structure of an energy storage monitoring device for an energy storage battery according to an embodiment of the present invention; Detailed Implementation Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0028] This invention provides an energy storage monitoring method for an energy storage battery, referring to... Figure 1 ,include: Step S1: Real-time acquisition of battery voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress data; Step S2: Construct an electrochemical-thermal-mechanical coupling model based on a physical information neural network, embed the electrochemical equation into the hidden layer of the neural network, use the collected data to train the model and optimize the network weights, and dynamically update the lithium ion concentration distribution and internal resistance inside the battery. Step S3: Using the trained model and real-time collected data, calculate the lithium dendrite growth rate and the combined gas concentration and mechanical stress. When the lithium dendrite growth rate exceeds the threshold or the gas concentration and mechanical stress meet the preset conditions, trigger the corresponding alarm mechanism.

[0029] The working principle and beneficial effects of the above technical solution are as follows: First, it acquires real-time data on battery voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress, providing comprehensive raw data for subsequent analysis. Second, it constructs an electrochemical-thermal-mechanical coupling model based on a physical information neural network, embedding electrochemical equations into the hidden layer of the neural network. The model is trained using the acquired data to optimize network weights, dynamically updating the lithium-ion concentration distribution and internal resistance within the battery, thus improving the model's prediction accuracy. Finally, using the trained model and real-time acquired data, it calculates the lithium dendrite growth rate and the combined gas concentration and mechanical stress. When the lithium dendrite growth rate exceeds a set threshold or the gas concentration and mechanical stress meet preset conditions, a corresponding alarm mechanism is triggered, achieving real-time monitoring and early warning of the battery's health status. This invention improves the accuracy and reliability of monitoring through multimodal data fusion and physical information embedding, while also possessing the characteristics of dynamic updating and efficient calculation, effectively preventing battery safety risks and extending battery life.

[0030] In a preferred embodiment, step S1 includes: The battery's voltage, current, and temperature data are acquired in real time using high-precision voltage, current, and temperature sensors. High-frequency impedance spectrum data of the battery were acquired using an electrochemical impedance spectroscopy (EIS) instrument in the frequency range of 0.1 Hz to 10 kHz. The impedance spectrum data were then converted from the time domain to the time-frequency domain using a short-time Fourier transform (STFT) to extract the time-frequency characteristics of the impedance spectrum. Gas sensors detect the composition of gases in the environment surrounding the battery, including the concentrations of carbon dioxide and hydrogen. The mechanical stress distribution of the battery is measured using strain gauges or fiber optic sensors.

[0031] The working principle and beneficial effects of the above technical solution are as follows: High-precision sensors are used to collect real-time voltage, current, and temperature data of the battery. Electrochemical impedance spectroscopy is used to acquire high-frequency impedance spectrum data within the frequency range of 0.1Hz to 10kHz. Short-time Fourier transform is then used to convert the data from the time domain to the time-frequency domain, extracting time-frequency features. Simultaneously, gas sensors detect the concentrations of carbon dioxide and hydrogen in the environment surrounding the battery, and strain gauges or fiber optic sensors measure the mechanical stress distribution of the battery. This multimodal data acquisition method can comprehensively capture various state information of the battery during operation, providing a rich data foundation for subsequent battery status assessment and fault early warning. Combined with a physical information neural network coupling model, the health status of the battery can be predicted and assessed more accurately, effectively preventing potential safety risks and extending battery life.

[0032] In a preferred embodiment, high-frequency impedance spectrum data of the battery is acquired using an electrochemical impedance spectroscopy (EIS) instrument within the frequency range of 0.1 Hz to 10 kHz. The impedance spectrum data is then transformed from the time domain to the time-frequency domain using a short-time Fourier transform (STFT), and the time-frequency characteristics of the impedance spectrum are extracted, including: High-frequency impedance spectral data of the battery were acquired using an electrochemical impedance spectroscopy (EIS) instrument in the frequency range of 0.1 Hz to 10 kHz. The acquired impedance data undergoes preprocessing, including noise removal and data smoothing, to improve data quality. The Hanning window function is selected to divide the impedance data into multiple short time periods. The length of each frame is set to 256 to 1024 sampling points, and the frames maintain an overlap of 50% to 75%. Perform a Short-Time Fourier Transform (STFT) on each frame of data to obtain the time-frequency representation of that frame. The calculation formula is as follows:

[0033] in, Indicates the input signal. For window functions, For time shift, For frequency; Extract the time-frequency characteristics of amplitude spectrum, phase spectrum and energy spectrum from the impedance spectrum from the short-time Fourier transform (STFT) calculation results.

[0034] The working principle and beneficial effects of the above technical solution are as follows: This preferred embodiment acquires high-frequency impedance spectrum data of the battery within a frequency range of 0.1 Hz to 10 kHz using an electrochemical impedance spectroscopy (EIS). First, the acquired data is preprocessed, including noise removal and data smoothing, to improve data quality. Next, a Hanning window function is selected to divide the data into multiple short time intervals, each frame containing 256 to 1024 sampling points, with 50% to 75% overlap between frames. Then, a short-time Fourier transform (STFT) is performed on each frame to obtain its time-frequency representation, thereby extracting the time-frequency characteristics of the impedance spectrum, including amplitude, phase, and energy spectra. This method can comprehensively capture the electrochemical characteristics of the battery during operation. By converting the impedance data from the time domain to the time-frequency domain using STFT, it provides more detailed information on the battery's internal state, offering strong support for battery state assessment and fault early warning. The wide frequency range of EIS data acquisition enables the capture of the battery's electrochemical behavior at different frequencies, providing a rich data foundation for subsequent analysis. Data preprocessing effectively removes noise and smooths the data, improving its accuracy and reliability. Using the Hanning window function for data framing reduces spectral leakage and improves the accuracy of the STFT. The extracted time-frequency features reflect the battery's electrochemical characteristics at different frequencies, providing comprehensive information for battery status assessment and fault early warning.

[0035] In a preferred embodiment, step S2 includes: A multi-scale neural network model is constructed, and physical equations are embedded into the hidden layer of the multi-scale neural network model to obtain an electrochemical-thermal-mechanical coupling model based on physical information neural network, so that the network can learn data while also satisfying the constraints of physical laws. Electrochemical equations are embedded in the hidden layer of a neural network, including a simplified diffusion equation for lithium-ion concentration:

[0036] in, Indicates lithium-ion concentration. Represents a time variable. This represents the diffusion coefficient of lithium ions. Battery current, It is Faraday's constant. The electrode area of ​​the battery; The internal resistance update equation is embedded in the hidden layer of the neural network to model the change in internal resistance:

[0037] in This represents the internal resistance at time t. The initial internal resistance, is the internal resistance variation coefficient, and k is the thermal conductivity of the battery; The model is trained using collected data on voltage, current, temperature, impedance spectrum, gas composition, and mechanical stress. The network weights are optimized using a loss function to dynamically update the lithium-ion concentration distribution and internal resistance within the battery. Represented as:

[0038] in, , and These are the weighting coefficients. The lithium-ion concentration predicted by the model. To measure the lithium-ion concentration, The internal resistance is predicted by the model. This is the measured internal resistance.

[0039] The working principle and beneficial effects of the above technical solution are as follows: First, a multi-scale neural network architecture is constructed to handle data at different scales, and physical equations are embedded in the hidden layers to ensure that the model meets the constraints of physical laws while learning from the data. The embedded physical equations include a simplified diffusion equation for lithium-ion concentration and an internal resistance update equation. The diffusion equation describes the diffusion process of lithium ions in the battery, while the internal resistance update equation models the change of internal resistance over time. In this way, the model can dynamically update the lithium-ion concentration distribution and internal resistance inside the battery. Next, the model is trained using collected data on voltage, current, temperature, impedance spectrum, gas composition, and mechanical stress. During training, a loss function that comprehensively considers the prediction errors of lithium-ion concentration and internal resistance is defined to optimize the network weights. The weight coefficients in the loss function can balance the contributions of different error terms, thus enabling the model to simultaneously focus on the prediction accuracy of lithium-ion concentration and internal resistance during training. This technical solution combines the advantages of data-driven and physical models. By embedding physical equations into the neural network, the model can not only learn from a large amount of data but also follow known physical laws, improving the accuracy and reliability of the model. Meanwhile, the multi-scale neural network architecture can handle data features at different scales, further improving model performance. This method can more accurately predict and assess battery health status, and promptly detect potential safety hazards, such as lithium dendrite growth. By dynamically updating lithium-ion concentration and internal resistance, the model can reflect the battery's operating status in real time, providing more accurate information to the battery management system, thereby improving battery safety and lifespan. Furthermore, this method also helps optimize battery design and operating strategies, reducing safety risks during battery use.

[0040] In a preferred embodiment, refer to Figure 2 The specific steps for constructing a multi-scale neural network model include: Acquire massive amounts of voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress data from similar energy storage batteries in advance, and use them as supporting data; After preprocessing the supporting data, data features are extracted from the supporting data at different scales; Principal component analysis or t-SNE dimensionality reduction techniques are used to fuse multi-scale data features into a unified feature space. Construct a multi-scale neural network architecture model to process data of different scales separately, and then fuse them in the intermediate or output layers of the network. Use convolutional neural networks to process spatial features and recurrent neural networks or long short-term memory networks to process time series data. An attention mechanism is introduced into the neural network architecture model to adjust the weights of features at different scales, enabling the model to focus on key features during prediction. By utilizing the fused data features, a multi-scale neural network architecture model is trained to obtain a multi-scale neural network model.

[0041] The working principle and beneficial effects of the above technical solution are as follows: First, massive amounts of voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress data from similar energy storage batteries are acquired as supporting data. Next, these data are preprocessed, and features are extracted from data at different scales. Then, principal component analysis or t-SNE dimensionality reduction techniques are used to fuse multi-scale features into a unified feature space. Subsequently, a multi-scale neural network architecture is constructed, using convolutional neural networks and recurrent neural networks or long short-term memory networks to process spatial features and time-series data, and fusing them in the intermediate or output layers of the network. Afterward, an attention mechanism is introduced to adjust the weights of features at different scales, enabling the model to focus on key features during prediction. Finally, the multi-scale neural network architecture is trained using the fused data features to obtain a multi-scale neural network model. This technical solution, through multi-scale data fusion and an attention mechanism, improves the accuracy and reliability of the model's assessment of the battery's internal state, achieving a comprehensive evaluation and early warning of the battery's health status, effectively preventing potential safety risks, and extending the battery's lifespan.

[0042] In a preferred embodiment, the calculation of lithium dendrite growth rate, combined gas concentration, and mechanical stress using a trained model and real-time acquired data includes: The lithium-ion concentration distribution and ohmic internal resistance at the current time and location are determined using the lithium-ion concentration distribution and internal resistance data predicted by the model. The reference high-frequency imaginary part and sensitivity factor under the current state are determined by looking up a table based on the lithium-ion concentration and ohmic internal resistance, and the internal resistance offset factor is calculated. The dendrite growth rate calibration coefficient is determined based on the sensitivity factor and internal resistance offset factor. Combined with real-time acquired impedance spectral data, the lithium dendrite growth rate is calculated.

[0043] in, Indicates the lithium dendrite growth rate. This represents the change in the imaginary part of impedance in the high-frequency band. This indicates the selected high-frequency band. For time intervals, This is the calibration coefficient for dendrite growth rate. Calculate the rate of change of carbon dioxide concentration To detect abnormal changes in gas concentration:

[0044] in, This represents the current carbon dioxide concentration. The carbon dioxide concentration at the previous moment; Calculate the rate of change of maximum mechanical stress To detect abnormal changes in mechanical stress:

[0045] in The maximum mechanical stress at the current moment. This represents the maximum mechanical stress at the previous moment.

[0046] The working principle and beneficial effects of the above technical solution are as follows: High-precision sensors are used to acquire real-time voltage, current, and temperature data of the battery. Electrochemical impedance spectroscopy (EIS) is used to acquire high-frequency impedance spectrum data over a wide frequency range, and short-time Fourier transform is employed to convert the data from the time domain to the time-frequency domain, extracting time-frequency features. Simultaneously, gas sensors detect the concentrations of carbon dioxide and hydrogen in the environment surrounding the battery, and strain gauges or fiber optic sensors measure the mechanical stress distribution of the battery. Features at different scales are extracted from the massive amounts of collected data, such as microscopic features like lithium-ion concentration, charge transfer impedance, battery capacity, and internal resistance, as well as macroscopic features like voltage, current, and temperature. Principal component analysis (PCA) or t-SNE dimensionality reduction techniques are used to fuse these multi-scale features into a unified feature space for subsequent model processing. A multi-scale neural network architecture is constructed, using convolutional neural networks to process spatial features, such as the frequency domain features of the impedance spectrum data; and recurrent neural networks (RNNs) or long short-term memory networks to process time-series data, such as the changes in voltage and current over time. An attention mechanism is introduced into the intermediate or output layers of the network to adjust the weights of features at different scales, enabling the model to focus on key features during prediction. A multi-scale neural network architecture is trained using fused data features. A reasonable loss function is designed, and network weights are optimized by combining the residuals of physical equations and the errors in data fitting. An adaptive optimization algorithm is used to improve the model's convergence speed and stability. The lithium dendrite growth rate is calculated by combining the model's predicted lithium-ion concentration distribution and internal resistance data with real-time impedance spectrum data. When the lithium dendrite growth rate exceeds a set threshold, a corresponding alarm mechanism is triggered to promptly detect abnormal lithium dendrite growth and prevent safety hazards such as internal battery short circuits. The rate of change of carbon dioxide concentration and maximum mechanical stress is calculated to monitor abnormal changes. When the gas concentration or mechanical stress exceeds a preset threshold, a corresponding alarm mechanism is triggered to promptly detect potential side reactions and structural stability issues within the battery. A comprehensive assessment of the battery's health status is achieved through the model's predicted lithium-ion concentration distribution, internal resistance changes, lithium dendrite growth rate, and real-time collected gas concentration and mechanical stress data. When any indicator exceeds a preset safety threshold, the system will automatically trigger the corresponding alarm mechanism to remind maintenance personnel to take timely measures to ensure the safe operation of the battery. This technical solution achieves comprehensive and accurate monitoring of the internal state of energy storage batteries through multimodal data fusion, multi-scale feature extraction and fusion, multi-scale neural network model construction, and multi-dimensional health assessment and early warning. This method not only improves the safety and reliability of battery use but also effectively prevents potential safety risks, extends battery life, and provides strong support for the efficient management and maintenance of energy storage batteries.

[0047] In a preferred embodiment, when the lithium dendrite growth rate exceeds a threshold or the gas concentration and mechanical stress meet preset conditions, a corresponding alarm mechanism is triggered, specifically including: Pre-set conditions considering the rate of change of gas concentration and the rate of change of mechanical stress will trigger a thermal runaway warning when the rate of change of carbon dioxide concentration exceeds a certain threshold and the rate of change of maximum mechanical stress meets the following conditions:

[0048] in, The stress change rate threshold; Alternatively, a thermal runaway precursor warning may be triggered when the lithium dendrite growth rate exceeds a threshold or when the gas concentration and mechanical stress simultaneously meet the following conditions:

[0049] in, Indicates carbon dioxide concentration. Indicates the maximum mechanical stress. This is the stress threshold.

[0050] The working principle and beneficial effects of the above technical solution are as follows: When the lithium dendrite growth rate exceeds a threshold or the gas concentration and mechanical stress meet preset conditions, a corresponding alarm mechanism is triggered. Specifically, conditions considering the rate of change of gas concentration and the rate of change of mechanical stress are first preset, such as a stress change rate threshold. When the lithium dendrite growth rate exceeds the threshold, or when the gas concentration and mechanical stress simultaneously meet the following conditions, a thermal runaway precursor warning is triggered: the carbon dioxide concentration exceeds 50 ppm and the maximum mechanical stress exceeds the stress threshold, or the rate of change of carbon dioxide concentration exceeds 10 ppm / s and the rate of change of maximum mechanical stress exceeds the preset stress change rate threshold. This technical solution achieves comprehensive monitoring and early warning of battery health status through the fusion and comprehensive judgment of multi-dimensional data. Monitoring the lithium dendrite growth rate can promptly detect changes in the microstructure inside the battery, preventing safety hazards such as internal short circuits caused by lithium dendrite growth. Monitoring the gas concentration and mechanical stress reflects the battery's operating environment and physical state from a macroscopic perspective, and by setting reasonable thresholds and conditions, potential safety risks can be effectively identified. When any monitored indicator exceeds a preset safety threshold, the system will automatically trigger the corresponding alarm mechanism, reminding maintenance personnel to take timely measures to ensure the safe operation of the battery. Through comprehensive analysis of multi-dimensional data, the accuracy and timeliness of early warnings are improved, effectively reducing safety risks during battery use. Simultaneously, this method can comprehensively assess the battery's health status, providing a scientific basis for battery maintenance and management, helping to extend battery life and improve battery efficiency. Furthermore, this technical solution has strong versatility and adaptability, and can be applied to different types of energy storage batteries, providing strong support for the safe operation and health management of energy storage batteries.

[0051] To achieve the above embodiments, the present invention also proposes an energy storage monitoring device for an energy storage battery. Figure 3 This is a schematic diagram of the structure of an energy storage monitoring device for an energy storage battery provided in an embodiment of the present invention. Figure 3 As shown, the device includes: The first execution module is used to acquire real-time data on battery voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress. The second execution module is used to construct an electrochemical-thermal-mechanical coupling model based on a physical information neural network. It embeds the electrochemical equation into the hidden layer of the neural network, uses the collected data to train the model, optimizes the network weights, and dynamically updates the lithium-ion concentration distribution and internal resistance inside the battery. The third execution module is used to calculate the lithium dendrite growth rate and the combined gas concentration and mechanical stress using the trained model and real-time collected data. When the lithium dendrite growth rate exceeds the threshold or the gas concentration and mechanical stress meet the preset conditions, the corresponding alarm mechanism is triggered.

[0052] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0053] To implement the above embodiments, the present invention also proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the method provided in the foregoing embodiments.

[0054] To implement the above embodiments, the present invention also proposes a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the methods provided in the foregoing embodiments.

[0055] To implement the above embodiments, the present invention also proposes a computer program product, including a computer program that, when executed by a processor, implements the methods provided in the foregoing embodiments.

[0056] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in this invention all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0057] It should be noted that personal information collected from users should be used for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. Furthermore, such collection / sharing should only be conducted after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign an agreement / authorization that includes authorization of relevant user information before the user uses the function. In addition, any necessary steps must be taken to protect and safeguard access to such personal information data and ensure that others with access to personal information data comply with their privacy policies and procedures.

[0058] This invention is intended to provide implementation schemes for users to selectively prevent the use or access to personal information data. That is, this disclosure is intended to provide hardware and / or software to prevent or block access to such personal information data. Once personal information data is no longer needed, risks can be minimized by restricting data collection and deleting data. Furthermore, where applicable, such personal information can be de-identified to protect user privacy.

[0059] In the foregoing descriptions of the embodiments, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0060] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0061] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of the invention pertain.

[0062] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0063] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any of the following techniques known in the art, or a combination thereof: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0064] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0065] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0066] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

[0067] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0068] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for monitoring the energy storage of a battery, characterized in that, The method comprises the following steps: Step S1, real-time acquisition of voltage, current, temperature, internal impedance spectrum, gas composition and mechanical stress data of the battery; Step S2, construction of an electrochemical-thermal-mechanical coupling model based on a physical information neural network, embedding of an electrochemical equation into a hidden layer of the neural network, training and optimization of network weights by using the collected data, dynamic updating of lithium ion concentration distribution and internal resistance in the battery; Step S3, calculation of lithium dendrite growth rate and combined gas concentration and mechanical stress by using the trained model and real-time collected data, triggering of a corresponding alarm mechanism when the lithium dendrite growth rate exceeds a threshold or the gas concentration and mechanical stress meet preset conditions.

2. The energy storage monitoring method of claim 1, wherein, Step S1 comprises: Real-time acquisition of voltage, current and temperature data of the battery by using high-precision voltage sensors, current sensors and temperature sensors; Acquisition of high-frequency impedance spectrum data of the battery within a frequency range of 0.1 Hz to 10 kHz by using an electrochemical impedance spectrometer EIS, and conversion of the impedance spectrum data from time domain to time-frequency domain by using short-time Fourier transform STFT, and extraction of time-frequency features of the impedance spectrum; Detection of gas components in the environment around the battery by using a gas sensor, including concentrations of carbon dioxide and hydrogen; Measurement of mechanical stress distribution of the battery by using a strain gauge or an optical fiber sensor.

3. The energy storage monitoring method of claim 2, wherein, Acquisition of high-frequency impedance spectrum data of the battery within a frequency range of 0.1 Hz to 10 kHz by using an electrochemical impedance spectrometer EIS, and conversion of the impedance spectrum data from time domain to time-frequency domain by using short-time Fourier transform STFT, and extraction of time-frequency features of the impedance spectrum include: Acquisition of high-frequency impedance spectrum data of the battery within a frequency range of 0.1 Hz to 10 kHz by using an electrochemical impedance spectrometer EIS; Preprocessing of the collected impedance data, including noise removal and data smoothing, to improve data quality; Selection of a Hanning window function to frame the impedance data into multiple short time periods, with each frame having a length of 256 to 1024 sampling points and a 50% to 75% overlap between frames; Short-time Fourier transform STFT calculation of each frame of data to obtain a time-frequency representation of the frame, with a calculation formula as follows: wherein, represents an input signal, is a window function, is a time shift, is a frequency; Extraction of time-frequency features of amplitude spectrum, phase spectrum and energy spectrum in the impedance spectrum from the short-time Fourier transform STFT calculation results.

4. The energy storage monitoring method of claim 1, wherein, Step S2 comprises: Construction of a multi-scale neural network model, embedding of a physical equation into a hidden layer of the multi-scale neural network model to obtain an electrochemical-thermal-mechanical coupling model based on a physical information neural network, so that the network learns data while satisfying the constraints of physical laws; Embedding of an electrochemical equation into a hidden layer of the neural network, including a simple diffusion equation of lithium ion concentration: wherein, represents the lithium ion concentration, represents the time variable, represents the diffusion coefficient of lithium ions, is the battery current, is the Faraday constant, is the electrode area of the battery; Embedding of an internal resistance updating equation into a hidden layer of the neural network to model changes in internal resistance: wherein represents the internal resistance at time t, is the initial internal resistance, is the internal resistance change coefficient, k is the thermal conductivity of the battery; The model is trained using collected voltage, current, temperature, impedance spectrum, gas composition and mechanical stress data, network weights are optimized through a loss function, and lithium ion concentration distribution and internal resistance inside the battery are dynamically updated, and the loss function is represented as: wherein, , and are weight coefficients, is the model predicted lithium ion concentration, is the measured lithium ion concentration, is the model predicted internal resistance, is the measured internal resistance.

5. The energy storage monitoring method of claim 4, wherein, The steps of constructing the multi-scale neural network model specifically comprise: Obtaining a large amount of voltage, current, temperature, internal impedance spectrum, gas composition and mechanical stress data collected from the same type of energy storage battery in advance as support data; Extraction of data features from support data after preprocessing; Principal component analysis or t-SNE dimensionality reduction techniques are used to fuse multi-scale data features into a unified feature space. Construct a multi-scale neural network architecture model to process data of different scales separately, and then fuse them in the intermediate or output layers of the network. Use convolutional neural networks to process spatial features and recurrent neural networks or long short-term memory networks to process time series data. An attention mechanism is introduced into the neural network architecture model to adjust the weights of features at different scales, enabling the model to focus on key features during prediction. By utilizing the fused data features, a multi-scale neural network architecture model is trained to obtain a multi-scale neural network model.

6. The energy storage monitoring method of an energy storage battery according to claim 1, wherein, Using a trained model and real-time acquired data, the lithium dendrite growth rate, combined gas concentration, and mechanical stress are calculated, including: The lithium-ion concentration distribution and ohmic internal resistance at the current time and location are determined using the lithium-ion concentration distribution and internal resistance data predicted by the model. The reference high-frequency imaginary part and sensitivity factor under the current state are determined by looking up a table based on the lithium-ion concentration and ohmic internal resistance, and the internal resistance offset factor is calculated. The dendrite growth rate calibration coefficient is determined based on the sensitivity factor and internal resistance offset factor. Combined with real-time acquired impedance spectral data, the lithium dendrite growth rate is calculated. wherein, represents a lithium dendrite growth rate, represents a change amount of the imaginary part of the high-frequency band impedance, represents a selected high-frequency band frequency, is a time interval, is a preset calibration coefficient; Computing a rate of change of carbon dioxide concentration to detect abnormal changes in gas concentration: wherein, is the carbon dioxide concentration at the current time instant, is the carbon dioxide concentration at the previous time instant; calculating the rate of change of the maximum mechanical stress to detect abnormal changes in the mechanical stress: wherein is the maximum mechanical stress at the current time instant, is the maximum mechanical stress at the previous time instant.

7. The energy storage monitoring method of an energy storage battery according to claim 6, wherein, When the lithium dendrite growth rate exceeds a threshold or the gas concentration and mechanical stress meet preset conditions, a corresponding alarm mechanism is triggered, specifically including: Pre-set conditions considering the rate of change of gas concentration and the rate of change of mechanical stress will trigger a thermal runaway warning when the rate of change of carbon dioxide concentration exceeds a certain threshold and the rate of change of maximum mechanical stress meets the following conditions: wherein is a stress rate of change threshold value; Alternatively, a thermal runaway precursor warning may be triggered when the lithium dendrite growth rate exceeds a threshold or when the gas concentration and mechanical stress simultaneously meet the following conditions: wherein represents the carbon dioxide concentration, represents the maximum mechanical stress, is a stress threshold value.

8. An energy storage monitoring device for an energy storage battery, characterized in that, include: The first execution module is used to acquire real-time data on battery voltage, current, temperature, internal impedance spectrum, gas composition, and mechanical stress. The second execution module is used to construct an electrochemical-thermal-mechanical coupling model based on a physical information neural network. It embeds the electrochemical equation into the hidden layer of the neural network, uses the collected data to train the model, optimizes the network weights, and dynamically updates the lithium-ion concentration distribution and internal resistance inside the battery. The third execution module is used to calculate the lithium dendrite growth rate and the combined gas concentration and mechanical stress using the trained model and real-time collected data. When the lithium dendrite growth rate exceeds the threshold or the gas concentration and mechanical stress meet the preset conditions, the corresponding alarm mechanism is triggered.

Images