Early warning system for thermal runaway of lithium ion battery in ultra-early stage

Abstract

The invention relates to the field of safety monitoring of power equipment, and provides an ultra-early-stage lithium ion battery thermal runaway early warning system which is applied to the interior of a battery prefabricated cabin. The data acquisition module is used for acquiring initial temperature rise change curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data and initial particle number data in the charging process of the lithium ion battery; the data processing module is used for carrying out sorting and dimension reduction on the initial temperature rise change curve data, the initial combustible gas concentration data, the initial electrolyte volatile liquid data and the initial particle number data through a principal component analysis method to obtain lithium ion battery characteristic data; the fire-fighting prediction module is used for inputting the lithium ion battery characteristic data into the fire-fighting state evaluation model to obtain the current fire-fighting state of the battery prefabricated cabin; the early warning generation module is used for judging the thermal runaway state of the corresponding lithium ion battery based on the current fire-fighting state and generating early warning signals of different levels based on the thermal runaway state; and the early warning response module is used for taking corresponding safety measures according to the early warning signals of different levels. According to the invention, ultra-early warning of thermal runaway of the container type lithium ion battery is realized, corresponding safety measures can be taken in time, and occurrence of false alarms and missing alarms is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the field of power equipment safety monitoring, and in particular to an ultra-early lithium-ion battery thermal runaway early warning system. Background Art

[0002] Containerized lithium-ion batteries are energy storage systems that use standard containers as carriers. They integrate multiple components such as lithium-ion batteries, battery management systems, and fire protection systems. They are highly integrated, large-capacity, and portable energy storage devices. Compared with traditional lithium batteries, containerized lithium-ion batteries have the characteristics of high integration, large capacity, strong mobility, and flexible interface design. Therefore, they are widely used in many fields. However, due to the special properties of lithium-ion batteries and the complexity of containerized energy storage systems, this type of energy storage system also has certain safety risks. Battery energy storage systems may experience thermal runaway in the event of overcharging, over-discharging, short circuiting, mechanical collision, etc., generating high temperatures, smoke, and even fire. Thermal runaway will not only damage the battery itself, but may also pose a threat to personal safety and equipment safety.

[0003] At present, traditional early warning methods are mainly aimed at ordinary lithium batteries, such as early warning methods for battery energy storage systems based on data-model hybrid drive, early warning methods for battery energy storage systems based on gas-liquid effusion image recognition, and early warning methods for battery energy storage systems based on gas analysis. However, when the above methods are applied to the fire safety early warning work of container-type lithium-ion battery energy storage systems, due to the large amount of complex data and the closed internal environment of container-type lithium-ion batteries, traditional data processing methods cannot effectively process the data of container-type lithium-ion batteries. In addition, the closed internal environment makes the gas diffusion characteristics difficult to predict and the light is insufficient to support the accuracy of image recognition. There are obvious phenomena of erroneous alarms and missed alarms. Summary of the invention

[0004] In view of this, the present invention proposes an ultra-early thermal runaway warning system for lithium-ion batteries, which adopts the principal component analysis method to perform dimensionality reduction processing on a variety of initial data, uses a fire status assessment model to predict the current fire status, and generates warning signals of different levels based on the thermal runaway status, thereby realizing an ultra-early warning of thermal runaway of container-type lithium-ion batteries, and being able to take corresponding safety measures in time, solving the limitations of traditional warning methods in processing large amounts of complex data of container-type lithium-ion batteries, as well as the problems of difficult-to-predict gas diffusion characteristics due to the closed internal environment and insufficient light affecting image recognition accuracy, and effectively reducing the occurrence of false alarms and missed alarms.

[0005] The technical solution of the present invention is implemented as follows: The present invention provides an ultra-early lithium-ion battery thermal runaway warning system, which is applied inside a battery prefabricated cabin, comprising:

[0006] A data acquisition module is used to collect initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data and initial particle number data during the charging process of the lithium-ion battery;

[0007] A data processing module is used to sort and reduce the dimension of the initial temperature rise curve data, the initial combustible gas concentration data, the initial electrolyte volatile liquid data and the initial particle number data by principal component analysis to obtain lithium-ion battery characteristic data;

[0008] A fire prediction module, used for inputting the characteristic data of the lithium-ion battery into a fire status assessment model to obtain the current fire status of the battery prefabricated cabin;

[0009] An early warning generation module, used to determine the thermal runaway state of the corresponding lithium-ion battery based on the current fire protection state, and generate early warning signals of different levels based on the thermal runaway state;

[0010] The early warning response module is used to take corresponding safety measures according to the early warning signals of different levels.

[0011] On the basis of the above technical solution, preferably, the sorting and dimension reduction of the initial temperature rise curve data, the initial combustible gas concentration data, the initial electrolyte volatile liquid data and the initial particle number data by the principal component analysis method comprises the following steps:

[0012] Standardize the initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data, and initial particle number data to obtain a lithium-ion battery standard data set;

[0013] Calculate the covariance matrix of the lithium-ion battery standard data set, and extract multiple eigenvalues ​​and multiple eigenvectors corresponding to the eigenvalues ​​based on the covariance matrix;

[0014] sorting the plurality of eigenvalues ​​according to their magnitude, selecting a plurality of principal component eigenvalues ​​from the plurality of eigenvalues, and constructing a projection matrix based on a plurality of principal component eigenvectors corresponding to the plurality of principal component eigenvalues;

[0015] Based on the projection matrix, the lithium-ion battery standard data set is subjected to dimensionality reduction processing to obtain lithium-ion battery characteristic data.

[0016] On the basis of the above technical solution, preferably, the calculation formula for the standardization process is:

[0017]

[0018] Wherein, z is the data of the lithium-ion battery standard data set, x is any one of the initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data, and initial particle number data, median(x) is the median of x, and IQR(x) is the interquartile range of x.

[0019] On the basis of the above technical solution, preferably, the calculation formula of the covariance matrix is:

[0020]

[0021] Among them, C is the covariance matrix, n is the data type of the lithium-ion battery standard data set, X i is the i-th type of data, is the data mean;

[0022] The calculation formula for extracting multiple eigenvalues ​​and multiple eigenvectors corresponding to the eigenvalues ​​based on the covariance matrix is:

[0023] Cv=λv;

[0024] Among them, λ is the eigenvalue and v is the eigenvector.

[0025] On the basis of the above technical solution, preferably, the calculation formula of the cumulative contribution rate is:

[0026]

[0027] Among them, P is the cumulative contribution rate, λ j is the jth eigenvalue, k is the number of principal component eigenvalues, and m is the number of eigenvalues.

[0028] On the basis of the above technical solution, preferably, the calculation formula of the projection matrix is:

[0029] W=[v1,v2,...,v k ];

[0030] Among them, v k is the kth principal component eigenvector, W is the projection matrix;

[0031] The calculation formula for performing dimensionality reduction processing on the lithium-ion battery standard data set based on the projection matrix is:

[0032] Y = z·W;

[0033] Among them, Y is the lithium-ion battery characteristic data, z is the data of the lithium-ion battery standard data set, and W is the projection matrix.

[0034] On the basis of the above technical solution, preferably, the calculation formula of the fire status assessment model is:

[0035]

[0036]

[0037] Where f(Y) is the current fire protection status of the battery prefabricated cabin, argmax(·) is the independent variable maximization function, is the weight of the a-th fire status, is the bias of the a-th fire state, P(y=a|Y) is the probability that the current fire state is the a-th class, exp(·) is the exponential function, is the weight of the c-th fire status, is the bias of the cth fire status, and d is the number of fire status categories.

[0038] On the basis of the above technical solution, preferably, judging the thermal runaway state of the corresponding lithium-ion battery based on the current fire protection state, and generating different levels of warning signals based on the thermal runaway state include:

[0039] The probability of various fire states of the current fire state is compared with the thermal runaway judgment threshold to obtain the thermal runaway state of the lithium-ion battery, and a corresponding warning signal is generated according to the thermal runaway state. The calculation formula of the thermal runaway state is:

[0040]

[0041] Among them, ZT R It is a thermal runaway state, ZT1 R 、ZT2 R 、ZT3 R are the first, second, and third thermal runaway states, respectively. P(y=1|Y) is the probability that the current firefighting state is the first firefighting state. P(y=2|Y) is the probability that the current firefighting state is the first firefighting state.

[0042] A red warning signal, an orange warning signal and a pink warning signal are generated according to different thermal runaway states. The red warning signal, the orange warning signal and the pink warning signal correspond to the first, second and third thermal runaway states respectively. The red warning signal, the orange warning signal and the pink warning signal are sorted from high to low according to the degree of urgency.

[0043] On the basis of the above technical solution, preferably, taking corresponding safety measures according to the warning signals of different levels includes:

[0044] For red warning signals, cut off the charging and discharging of lithium-ion batteries, start the emergency cooling and gas exhaust systems, notify staff to evacuate and contact the fire department;

[0045] For orange warning signals, reduce charging and discharging power, increase monitoring frequency, initiate preventive cooling measures, and notify staff to prepare for emergency response;

[0046] For pink warning signals, increase monitoring frequency, check battery management system parameters, prepare cooling equipment, and notify staff for further inspection.

[0047] On the basis of the above technical solution, preferably, it also includes: arranging a temperature sensor, a smoke sensor, a combustible gas detector and a particle counter inside the battery prefabricated cabin to monitor the state changes and environmental changes during the charging process of the lithium-ion battery.

[0048] Compared with the prior art, the ultra-early lithium-ion battery thermal runaway warning system of the present invention has the following advantages:

[0049] Beneficial effects:

[0050] (1) The principal component analysis method is used to reduce the dimensionality of various initial data, and the fire status assessment model is used to predict the current fire status. Different levels of warning signals are generated based on the thermal runaway status, achieving ultra-early warning of thermal runaway of containerized lithium-ion batteries, enabling timely adoption of corresponding safety measures, and effectively reducing the occurrence of false alarms and missed alarms;

[0051] (2) By collecting a variety of monitoring data during the charging process of lithium-ion batteries, including temperature, combustible gas concentration, electrolyte volatiles, and particle count, these heterogeneous data are fused and processed according to the principal component analysis method to extract key battery status characteristics, providing a comprehensive data basis for fire status assessment and improving the system's perception ability and diagnostic accuracy;

[0052] (3) Through the fire status assessment model based on machine learning, the current fire status of the battery prefabricated cabin is predicted according to the battery characteristic data, and the probability of each fire status is given, which helps to timely discover potential safety hazards;

[0053] (4) Based on the fire status assessment results, the thermal runaway state is divided into three levels, and corresponding red, orange and pink warning signals are generated. Corresponding measures are taken for different levels of warnings to effectively reduce the probability of thermal runaway accidents and the degree of losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 This is a structural diagram of an ultra-early lithium-ion battery thermal runaway warning system of the present invention. DETAILED DESCRIPTION

[0056] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] See also Figure 1 The present invention provides an ultra-early lithium-ion battery thermal runaway warning system, which is applied inside a battery prefabricated cabin, comprising:

[0058] A data acquisition module is used to collect initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data and initial particle number data during the charging process of the lithium-ion battery;

[0059] A data processing module is used to sort and reduce the dimension of the initial temperature rise curve data, the initial combustible gas concentration data, the initial electrolyte volatile liquid data and the initial particle number data by principal component analysis to obtain lithium-ion battery characteristic data;

[0060] A fire prediction module, used for inputting the characteristic data of the lithium-ion battery into a fire status assessment model to obtain the current fire status of the battery prefabricated cabin;

[0061] An early warning generation module, used to determine the thermal runaway state of the corresponding lithium-ion battery based on the current fire protection state, and generate early warning signals of different levels based on the thermal runaway state;

[0062] The early warning response module is used to take corresponding safety measures according to the early warning signals of different levels.

[0063] Specifically, an ultra-early warning system for thermal runaway of lithium-ion batteries in this embodiment uses principal component analysis to perform dimensionality reduction processing on a variety of initial data, uses a fire status assessment model to predict the current fire status, and generates warning signals of different levels based on the thermal runaway status, thereby achieving ultra-early warning of thermal runaway of containerized lithium-ion batteries, enabling timely adoption of corresponding safety measures, and effectively reducing the occurrence of false alarms and missed alarms.

[0064] Temperature sensors, smoke sensors, combustible gas detectors and particle counters are arranged inside the battery prefabricated cabin to monitor the state changes and environmental changes during the charging process of the lithium-ion battery.

[0065] Specifically, this embodiment can comprehensively monitor multiple parameters such as battery temperature changes, smoke generation, combustible gas concentration and particle quantity through the combination of multiple sensors, provide more comprehensive battery status information, and timely capture small abnormal changes in the battery charging process, such as abnormal temperature increase, trace gas release, etc., to achieve ultra-early warning and buy time for taking preventive measures.

[0066] Inside the closed battery prefabricated cabin, a multi-sensor arrangement is used to overcome the limitations of a single sensor in a specific environment, improve the reliability and accuracy of monitoring, and different types of data collected by multiple sensors can be fused and analyzed to provide a more comprehensive and accurate battery status assessment, helping to reduce the probability of false alarms and missed alarms.

[0067] The method of sorting and reducing the initial temperature rise curve data, the initial combustible gas concentration data, the initial electrolyte volatile liquid data and the initial particle number data by the principal component analysis method comprises the following steps:

[0068] Standardize the initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data, and initial particle number data to obtain a lithium-ion battery standard data set;

[0069] Calculate the covariance matrix of the lithium-ion battery standard data set, and extract multiple eigenvalues ​​and multiple eigenvectors corresponding to the eigenvalues ​​based on the covariance matrix;

[0070] sorting the plurality of eigenvalues ​​according to their magnitude, selecting a plurality of principal component eigenvalues ​​from the plurality of eigenvalues, and constructing a projection matrix based on a plurality of principal component eigenvectors corresponding to the plurality of principal component eigenvalues;

[0071] Based on the projection matrix, the lithium-ion battery standard data set is subjected to dimensionality reduction processing to obtain lithium-ion battery characteristic data.

[0072] Specifically, this embodiment greatly reduces the complexity of data processing and improves analysis efficiency by reducing high-dimensional raw data (including temperature rise curves, combustible gas concentrations, electrolyte volatiles and particle counts, etc.) to lower-dimensional feature data.

[0073] Through principal component analysis, the most representative and influential features can be extracted from a variety of raw data, retaining the most important information in the data, which helps to more accurately reflect the state of lithium-ion batteries.

[0074] In the process of dimensionality reduction, principal component analysis can effectively filter out noise and redundant information in the data and improve the signal-to-noise ratio of the data.

[0075] Standardizing the raw data eliminates the dimensional differences between different types of data, allowing data from different sources to be compared and analyzed on the same scale, thereby improving the accuracy of data processing.

[0076] The calculation formula for the standardization process is:

[0077]

[0078] Wherein, z is the data of the lithium-ion battery standard data set, x is any one of the initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data, and initial particle number data, median(x) is the median of x, and IQR(x) is the interquartile range of x.

[0079] Specifically, this embodiment uses the median and interquartile range for standardization, which is less sensitive to outliers and extreme data than the mean and standard deviation, improves the robustness of data processing, and is particularly suitable for processing battery monitoring data with outliers.

[0080] Converting data of different types and magnitudes (such as temperature, gas concentration, particle number, etc.) to the same scale eliminates the dimension effect and enables direct comparison and comprehensive analysis between different features.

[0081] By using the interquartile range as the denominator, the distribution characteristics of the original data are retained. Unlike z-score standardization, it does not assume that the data is normally distributed. It is more suitable for processing battery monitoring data with non-normal distribution.

[0082] The standardized data is more conducive to the training and prediction of subsequent machine learning models, which can accelerate the convergence of the model and improve the accuracy and generalization ability of the model.

[0083] The calculation formula of the covariance matrix is:

[0084]

[0085] Among them, C is the covariance matrix, n is the data type of the lithium-ion battery standard data set, X i is the i-th type of data, is the data mean;

[0086] The calculation formula for extracting multiple eigenvalues ​​and multiple eigenvectors corresponding to the eigenvalues ​​based on the covariance matrix is:

[0087] Cv=λv;

[0088] Among them, λ is the eigenvalue and v is the eigenvector.

[0089] Specifically, the covariance matrix calculation of this embodiment can reveal the correlation between different types of data (such as temperature, gas concentration, etc.), which is helpful to understand the mutual influence between various monitoring parameters.

[0090] By calculating the covariance matrix and its eigenvalues ​​and eigenvectors, a mathematical foundation is laid for the subsequent principal component analysis (PCA) dimensionality reduction, which enables high-dimensional data to be effectively compressed into a low-dimensional space while retaining the most important information.

[0091] The calculation of eigenvalues ​​and eigenvectors allows the identification of the most representative directions (principal components) in the data, which capture the main variations and patterns in the data and help extract the most critical battery state features.

[0092] Using matrix operation methods to calculate covariance and eigenvalues / eigenvectors greatly improves computational efficiency compared to calculating the covariance between each variable one by one, especially when processing large amounts of multidimensional battery monitoring data.

[0093] The calculation formula of the cumulative contribution rate is:

[0094]

[0095] Among them, P is the cumulative contribution rate, λ j is the jth eigenvalue, k is the number of principal component eigenvalues, and m is the number of eigenvalues.

[0096] Specifically, the cumulative contribution rate of this embodiment provides a quantitative indicator for evaluating how much proportion of information in the original data can be retained by the selected principal component, which helps to balance the trade-off between information retention and dimensionality reduction during the dimensionality reduction process.

[0097] By setting a threshold for the cumulative contribution rate (such as 95% or 99%), the degree of dimensionality reduction can be flexibly controlled. This enables the system to adaptively select an appropriate number of principal components based on specific application requirements, ensuring both full retention of information and effective dimensionality reduction.

[0098] The calculation process of the cumulative contribution rate is actually to sort and accumulate the eigenvalues ​​(representing the variance of each principal component). This sorting helps to identify the most important principal components.

[0099] The calculation method of the cumulative contribution rate is simple and intuitive, and only requires simple mathematical operations on the sorted eigenvalues. This simplicity makes the method easy to implement and understand, and is conducive to quickly evaluating the dimensionality reduction effect in practical applications.

[0100] The calculation formula of the projection matrix is:

[0101] W=[v1,v2,...,v k];

[0102] Among them, v k is the kth principal component eigenvector, W is the projection matrix;

[0103] The calculation formula for performing dimensionality reduction processing on the lithium-ion battery standard data set based on the projection matrix is:

[0104] Y = z·W;

[0105] Among them, Y is the lithium-ion battery characteristic data, z is the data of the lithium-ion battery standard data set, and W is the projection matrix.

[0106] Specifically, this embodiment constructs projection matrices W and z to perform matrix multiplication operations, which can quickly convert a high-dimensional standard data set z into low-dimensional feature data Y. This method has high computational efficiency and is suitable for processing large-scale lithium-ion battery monitoring data of this embodiment.

[0107] The projection matrix W contains the selected main eigenvectors, which can project the original data into a low-dimensional subspace containing the most important information, achieving effective data compression while retaining the most critical features for judging the status of lithium-ion batteries.

[0108] The originally correlated features are converted into new orthogonal features through projection transformation. This decoupling effect helps to avoid the multicollinearity problem between features.

[0109] Although dimensionality reduction is performed, since the projection matrix W is retained, the reduced-dimensional data can theoretically be reconstructed back to the original space through inverse operations, which provides the possibility for subsequent data reconstruction or further analysis and increases the flexibility of the system.

[0110] The calculation formula of the fire status assessment model is:

[0111]

[0112]

[0113] Where f(Y) is the current fire protection status of the battery prefabricated cabin, argmax(·) is the independent variable maximization function, is the weight of the a-th fire status, is the bias of the a-th fire state, P(y=a|Y) is the probability that the current fire state is the a-th class, exp(·) is the exponential function, is the weight of the c-th fire status, is the bias of the cth fire status, and d is the number of fire status categories.

[0114] Specifically, the fire status assessment model of this embodiment is suitable for complex battery prefabricated cabin environments by simultaneously evaluating the probabilities of multiple fire status categories, and can distinguish between a variety of different fire statuses, thereby improving the accuracy and comprehensiveness of status assessment.

[0115] The softmax function model outputs the probability of each fire status category instead of simple category division. This probabilistic output provides richer information and is helpful for risk assessment and decision making.

[0116] The use of exponential functions for nonlinear transformation enhances the model's ability to capture complex patterns and enables it to handle the nonlinear relationship between lithium-ion battery status and fire status.

[0117] Weights in the model and bias Used to reflect the impact of different characteristics on various fire conditions.

[0118] The determining the thermal runaway state of the corresponding lithium-ion battery based on the current fire protection state and generating different levels of warning signals based on the thermal runaway state includes:

[0119] The probability of various fire states of the current fire state is compared with the thermal runaway judgment threshold to obtain the thermal runaway state of the lithium-ion battery, and a corresponding warning signal is generated according to the thermal runaway state. The calculation formula of the thermal runaway state is:

[0120]

[0121] Among them, ZT R It is a thermal runaway state, ZT1 R 、ZT2 R 、ZT3 R are the first, second, and third thermal runaway states, respectively. P(y=1|Y) is the probability that the current firefighting state is the first firefighting state. P(y=2|Y) is the probability that the current firefighting state is the first firefighting state.

[0122] A red warning signal, an orange warning signal and a pink warning signal are generated according to different thermal runaway states. The red warning signal, the orange warning signal and the pink warning signal correspond to the first, second and third thermal runaway states respectively. The red warning signal, the orange warning signal and the pink warning signal are sorted from high to low according to the degree of urgency.

[0123] Specifically, in this embodiment, the thermal runaway state is divided into three levels: ZT1 R 、ZT2 R 、ZT3 R, and generates three warning signals of red, orange and pink accordingly, realizing a multi-level warning mechanism. Through this graded warning, the degree of danger of lithium-ion batteries can be more accurately reflected, which helps to take corresponding countermeasures.

[0124] The thermal runaway state is judged by comparing the fire state probability with a fixed threshold. This method not only takes into account the possibility of different fire states, but also sets clear judgment criteria, thereby improving the accuracy and reliability of the early warning.

[0125] By sorting the warning signals from high to low urgency (red>orange>pink), a clear warning priority system has been established, which helps to prioritize the most critical situations when multiple batteries experience abnormalities at the same time, thereby improving the efficiency of emergency response.

[0126] The taking of corresponding safety measures according to the warning signals of different levels includes:

[0127] For red warning signals, cut off the charging and discharging of lithium-ion batteries, start the emergency cooling and gas exhaust systems, notify staff to evacuate and contact the fire department;

[0128] For orange warning signals, reduce charging and discharging power, increase monitoring frequency, initiate preventive cooling measures, and notify staff to prepare for emergency response;

[0129] For pink warning signals, increase monitoring frequency, check battery management system parameters, prepare cooling equipment, and notify staff for further inspection.

[0130] Specifically, this embodiment adopts different levels of safety measures according to the severity of the warning signal (red, orange, pink), thereby achieving a refined emergency response. Through this hierarchical response mechanism, while ensuring safety, unnecessary losses caused by overreaction can be avoided.

[0131] The safety measures of this embodiment cover multiple aspects, including battery operation (such as cutting off charging and discharging, reducing power), environmental control (such as starting the cooling system, gas emission), personnel management (such as evacuation, preparing for emergency response) and equipment inspection, etc. This comprehensive protection strategy can effectively reduce the risk of accidents.

[0132] The system can automatically perform certain safety measures (such as cutting off charging and discharging, and starting the cooling system) based on early warning signals. It also includes manual intervention steps such as notifying staff. Through the combination of automation and manual intervention, it can respond quickly while retaining the possibility of human judgment and control.

[0133] Even for the lowest level of pink alert, preventive measures such as increasing monitoring frequency and checking system parameters are taken. Through the above preventive strategies, potential problems can be discovered early and prevented before they occur, thus improving the safety and reliability of the entire system.

[0134] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An ultra-early lithium-ion battery thermal runaway warning system, characterized in that: Applied to the interior of the battery prefabricated cabin, including: A data acquisition module is used to collect initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data and initial particle number data during the charging process of the lithium-ion battery; A data processing module is used to sort and reduce the dimension of the initial temperature rise curve data, the initial combustible gas concentration data, the initial electrolyte volatile liquid data and the initial particle number data by principal component analysis to obtain lithium-ion battery characteristic data; A fire prediction module, used for inputting the characteristic data of the lithium-ion battery into a fire status assessment model to obtain the current fire status of the battery prefabricated cabin; An early warning generation module, used to determine the thermal runaway state of the corresponding lithium-ion battery based on the current fire protection state, and generate early warning signals of different levels based on the thermal runaway state; The early warning response module is used to take corresponding safety measures according to the early warning signals of different levels.

2. The ultra-early lithium-ion battery thermal runaway warning system according to claim 1, characterized in that: The method of sorting and reducing the initial temperature rise curve data, the initial combustible gas concentration data, the initial electrolyte volatile liquid data and the initial particle number data by the principal component analysis method comprises the following steps: Standardize the initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data, and initial particle number data to obtain a lithium-ion battery standard data set; Calculate the covariance matrix of the lithium-ion battery standard data set, and extract multiple eigenvalues ​​and multiple eigenvectors corresponding to the eigenvalues ​​based on the covariance matrix; sorting the plurality of eigenvalues ​​according to their magnitude, selecting a plurality of principal component eigenvalues ​​from the plurality of eigenvalues, and constructing a projection matrix based on a plurality of principal component eigenvectors corresponding to the plurality of principal component eigenvalues; Based on the projection matrix, the lithium-ion battery standard data set is subjected to dimensionality reduction processing to obtain lithium-ion battery characteristic data.

3. The ultra-early lithium-ion battery thermal runaway warning system according to claim 2, characterized in that: The calculation formula for the standardization process is: Wherein, z is the data of the lithium-ion battery standard data set, x is any one of the initial temperature rise curve data, initial combustible gas concentration data, initial electrolyte volatile liquid data, and initial particle number data, median(x) is the median of x, and IQR(x) is the interquartile range of x.

4. The ultra-early lithium-ion battery thermal runaway warning system according to claim 3, characterized in that: The calculation formula of the covariance matrix is: Among them, C is the covariance matrix, n is the data type of the lithium-ion battery standard data set, X i is the i-th type of data, X is the data mean; The calculation formula for extracting multiple eigenvalues ​​and multiple eigenvectors corresponding to the eigenvalues ​​based on the covariance matrix is: Cv=λv; Among them, λ is the eigenvalue and v is the eigenvector.

5. The ultra-early lithium-ion battery thermal runaway warning system according to claim 4, characterized in that: The calculation formula of the cumulative contribution rate is: Among them, P is the cumulative contribution rate, λ j is the jth eigenvalue, k is the number of principal component eigenvalues, and m is the number of eigenvalues.

6. The ultra-early lithium-ion battery thermal runaway warning system according to claim 5, characterized in that: The calculation formula of the projection matrix is: <h2 style=";text-align:left;direction:ltr">W=[v1,v2,...,v<h2 style=";text-align:left;direction:ltr"> k <h2 style=";text-align:left;direction:ltr"> ]; Among them, v k is the kth principal component eigenvector, W is the projection matrix; The calculation formula for performing dimensionality reduction processing on the lithium-ion battery standard data set based on the projection matrix is: Y = z·W; Among them, Y is the lithium-ion battery characteristic data, z is the data of the lithium-ion battery standard data set, and W is the projection matrix.

7. An ultra-early lithium-ion battery thermal runaway warning system as claimed in claim 6, characterized in that: The calculation formula of the fire status assessment model is: Where f(Y) is the current fire protection status of the battery prefabricated cabin, argmax(·) is the independent variable maximization function, is the weight of the a-th fire status, is the bias of the a-th fire state, P(y=a|Y) is the probability that the current fire state is the a-th class, exp(·) is the exponential function, is the weight of the c-th fire status, is the bias of the cth fire status, and d is the number of fire status categories.

8. The ultra-early lithium-ion battery thermal runaway warning system according to claim 7, characterized in that: The determining the thermal runaway state of the corresponding lithium-ion battery based on the current fire protection state and generating different levels of warning signals based on the thermal runaway state includes: The probability of various fire states of the current fire state is compared with the thermal runaway judgment threshold to obtain the thermal runaway state of the lithium-ion battery, and a corresponding warning signal is generated according to the thermal runaway state. The calculation formula of the thermal runaway state is: Among them, ZT R It is a thermal runaway state, ZT1 R , are the first, second, and third thermal runaway states, respectively. P(y=1|Y) is the probability that the current firefighting state is the first firefighting state. P(y=2|Y) is the probability that the current firefighting state is the first firefighting state. A red warning signal, an orange warning signal and a pink warning signal are generated according to different thermal runaway states. The red warning signal, the orange warning signal and the pink warning signal correspond to the first, second and third thermal runaway states respectively. The red warning signal, the orange warning signal and the pink warning signal are sorted from high to low according to the degree of urgency.

9. The ultra-early lithium-ion battery thermal runaway warning system according to claim 8, characterized in that: The taking of corresponding safety measures according to the warning signals of different levels includes: For red warning signals, cut off the charging and discharging of lithium-ion batteries, start the emergency cooling and gas exhaust systems, notify staff to evacuate and contact the fire department; For orange warning signals, reduce charging and discharging power, increase monitoring frequency, initiate preventive cooling measures, and notify staff to prepare for emergency response; For pink warning signals, increase monitoring frequency, check battery management system parameters, prepare cooling equipment, and notify staff for further inspection.

10. The ultra-early lithium-ion battery thermal runaway warning system according to claim 9, characterized in that: Also includes: Temperature sensors, smoke sensors, combustible gas detectors and particle counters are arranged inside the battery prefabricated cabin to monitor the state changes and environmental changes during the charging process of the lithium-ion battery.

Images