Ternary lithium battery thermal runaway early warning method and system for time sequence prediction

Abstract

The invention relates to a ternary lithium battery thermal runaway early warning method and system based on time sequence prediction, and relates to the field of intelligent early warning, battery characteristic parameters are monitored and divided according to time sequence, the thermal runaway probability and the characteristic parameter mutation time are predicted in combination with environment characteristic parameters, then the unit mutation time is calculated, thermal runaway early warning identification resources are optimized accordingly, and the early warning efficiency is improved. And finally, performing thermal runaway identification early warning on each battery characteristic parameter section, thereby solving the technical problems of insufficient judgment accuracy and low timeliness of thermal runaway early warning of the ternary lithium battery, effectively improving the accuracy and timeliness of thermal runaway early warning, and providing powerful guarantee for safe use of the ternary lithium battery.

Description

Technical Field

[0001] The present invention relates to the field of intelligent early warning, and in particular to a time series prediction method and system for thermal runaway early warning of a ternary lithium battery. Background Art

[0002] With the rapid development of electric vehicles, energy storage systems, and other fields, ternary lithium batteries have been widely used in these fields due to their high energy density and long cycle life. However, the safety risk of ternary lithium batteries due to thermal runaway has become increasingly prominent. Thermal runaway occurs when a battery's internal chemical reactions run away under overheating conditions, causing a sharp rise in battery temperature and even fire or explosion, posing a serious threat to the safety of people and property.

[0003] Currently, thermal runaway early warning technologies for ternary lithium batteries primarily include simple threshold discrimination and model analysis. The simple threshold discrimination method sets thresholds for battery characteristic parameters, such as temperature and voltage, and triggers an alert when the monitored value exceeds the threshold. However, this method has significant limitations. First, the thresholds are often based on experience or limited data, making them inadequate for fully reflecting the thermal runaway characteristics of batteries under different operating conditions. Second, when battery characteristic parameters approach but have not yet exceeded the threshold, this method cannot provide early warning of potential thermal runaway risks, resulting in insufficient timeliness of the warning. Model analysis, on the other hand, uses mathematical or machine learning models to monitor and analyze battery characteristic parameters in real time to predict the probability of thermal runaway. While this method improves early warning accuracy to a certain extent, it also has some drawbacks. For example, model training requires a large amount of historical data, and the model's generalization ability is limited by the diversity and representativeness of the training data. Furthermore, model analysis often requires high computing resources, which impacts the timeliness of early warnings, making it difficult to meet the requirements of applications with high real-time requirements.

[0004] In summary, the existing thermal runaway warning system for ternary lithium batteries has problems of insufficient judgment accuracy and low timeliness, which makes it difficult to meet the safety requirements in practical applications. Summary of the Invention

[0005] The present invention aims to solve the technical problems of insufficient accuracy and low timeliness of thermal runaway warning judgment of ternary lithium batteries in the prior art, and provides a time series prediction thermal runaway warning method and system for ternary lithium batteries to solve the problems.

[0006] The technical solution of the present invention to solve the above technical problems is as follows:

[0007] In a first aspect, the present invention provides a time series prediction method for thermal runaway warning of a ternary lithium battery, the method comprising: monitoring battery characteristic parameters of the ternary lithium battery to obtain a battery characteristic parameter sequence, dividing the battery characteristic parameter segment sequence according to the time series; collecting environmental characteristic parameters of the ternary lithium battery operation, predicting the thermal runaway probability of the lithium battery to obtain the thermal runaway probability, and predicting the battery characteristic parameter mutation to obtain the predicted mutation time; calculating the actual mutation time of the battery characteristic parameter mutation based on the battery characteristic parameter sequence, and obtaining the unit mutation time in combination with the predicted mutation time; optimizing the thermal runaway warning identification resources based on the unit mutation time and the thermal runaway probability to obtain the optimal thermal runaway warning discrimination resources, performing thermal runaway identification and warning on each battery characteristic parameter segment, and obtaining a thermal runaway warning result.

[0008] In the second aspect, the present invention provides a ternary lithium battery thermal runaway warning system with time series prediction, the system comprising: a data monitoring module, used to monitor the battery characteristic parameters of the ternary lithium battery, obtain a battery characteristic parameter sequence, and obtain a battery characteristic parameter segment sequence according to the time series; a parameter acquisition module, used to collect the environmental characteristic parameters of the ternary lithium battery operation, predict the thermal runaway probability of the lithium battery, obtain the thermal runaway probability, and predict the battery characteristic parameter mutation to obtain the predicted mutation time; a time calculation module, used to calculate the actual mutation time of the battery characteristic parameter mutation based on the battery characteristic parameter sequence, and obtain the unit mutation time in combination with the predicted mutation time; an identification and warning module, used to optimize the thermal runaway warning identification resource based on the unit mutation time and the thermal runaway probability, obtain the optimal thermal runaway warning discrimination resource, perform thermal runaway identification and warning on each battery characteristic parameter segment, and obtain a thermal runaway warning result.

[0009] The beneficial effects of the present invention are: by monitoring the battery characteristic parameters and dividing them according to time series, combining the environmental characteristic parameters to predict the thermal runaway probability and the characteristic parameter mutation time, and then calculating the unit mutation time, and optimizing the thermal runaway warning identification resources accordingly, finally performing thermal runaway identification and warning for each battery characteristic parameter segment, effectively improving the accuracy and timeliness of the thermal runaway warning, and providing a strong guarantee for the safe use of ternary lithium batteries. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic flow chart of a time series prediction method for thermal runaway warning of a ternary lithium battery provided by the present invention.

[0011] Figure 2 This is a structural schematic diagram of a time series prediction thermal runaway warning system for ternary lithium batteries provided by the present invention.

[0012] Description of the accompanying drawings: data monitoring module 11, parameter acquisition module 12, time calculation module 13, identification and warning module 14. DETAILED DESCRIPTION

[0013] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0014] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0015] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0016] Example 1:

[0017] like Figure 1 As shown, an embodiment of the present invention provides a time series prediction method for thermal runaway warning of a ternary lithium battery, the method comprising:

[0018] S10: Monitor the battery characteristic parameters of the ternary lithium battery to obtain a battery characteristic parameter sequence, and divide the battery characteristic parameter segment sequence according to the time sequence.

[0019] For example, the basic principle of time series prediction is to use the patterns and trends in historical time series data to predict the data values at a certain point in the future or within a time period through mathematical models or machine learning algorithms. By collecting, organizing and analyzing data from a period of time, the periodic, trending or random change patterns in the data are identified, and then a prediction model is built based on these patterns to infer and predict future data. In the context of thermal runaway warning for ternary lithium batteries, time series prediction can be applied to the change patterns of battery characteristic parameters such as temperature and voltage over time, thereby predicting the risk of thermal runaway that may occur in the future.

[0020] Specifically, in the process of thermal runaway warning for ternary lithium batteries, the battery characteristic parameters are first continuously monitored. These characteristic parameters include but are not limited to key indicators such as battery temperature, voltage, and current, which can directly reflect the working status and health of the battery. Preferably, temperature monitoring includes the surface temperature of the battery cell (single-point accuracy ±0.5°C) and the average temperature of the module (global accuracy ±1°C), covering a range of -30°C to 80°C; voltage monitoring includes single cell voltage (resolution 0.001V), battery pack terminal voltage (accuracy 0.1% FS), and monitoring overcharge / over-discharge thresholds (such as the overcharge threshold of 4.3V for ternary lithium cells and the over-discharge threshold of 2.5V); current monitoring includes charge and discharge current (accuracy 0.5% FS), distinguishing between positive charging current and negative discharging current, and monitoring the short-circuit current threshold (>5C for 10ms is considered abnormal). In addition to the core parameters listed above, SOC (state of charge) monitoring is also an option. This is estimated using the ampere-hour integration method combined with a Kalman filter, with an accuracy of ±3%. SOH (state of health) monitoring is based on a comprehensive calculation of capacity decay (<80% is considered aging) and internal resistance growth (>1.5 times the initial value is considered abnormal). The selection of battery characteristic parameters can be adjusted based on actual needs.

[0021] In a preferred embodiment, the specific sensor selection and sampling frequency include: a thin-film platinum resistor (Pt100) installed at the contact surface between the battery cell tab and the shell to monitor the battery cell temperature, with a sampling frequency set to 10Hz, and data preprocessing using median filtering (window 3 points) combined with normalization; a thermocouple (K type) installed on the central heat sink of the module to monitor the module temperature, with a sampling frequency set to 5Hz, and data preprocessing using sliding average filtering (window 5 points) combined with normalization; an isolated voltage sensor installed on the positive and negative poles of the battery cell to monitor the single cell voltage, with a sampling frequency set to 1Hz, and data preprocessing using Butterworth low-pass filtering (cutoff frequency 0.1Hz); a Hall current sensor installed in the main circuit of the battery pack to monitor the charge and discharge current, with a sampling frequency set to 10Hz, and data preprocessing using DC offset removal combined with Z-score normalization; and a digital temperature and pressure sensor installed at the vent of the battery compartment to monitor the ambient temperature / pressure, with a sampling frequency set to 1Hz, and data preprocessing using exponential smoothing filtering combined with normalization. The filtering algorithm used includes: 3-point median filtering for temperature data to eliminate random noise (such as sensor glitches); 5-point sliding average for voltage / current data to suppress high-frequency interference (such as switching power supply ripple). The normalization formula is: , where μ is the historical mean and σ is the historical standard deviation. It is applicable to continuous parameters such as voltage, current, and temperature.

[0022] Furthermore, through high-precision sensors and data acquisition, the values of the above parameters can be obtained in real time and arranged in chronological order, thus forming a complete battery characteristic parameter sequence. In order to more finely analyze the changing trend of the battery status, this continuous parameter sequence needs to be divided according to a certain time interval, for example, cutting it in units of one second, so that multiple battery characteristic parameter segments are obtained. Each parameter segment contains the specific value of the battery characteristic parameter within the time period, and together constitutes a battery characteristic parameter segment sequence. This time segmentation method helps to more accurately capture the subtle changes in the battery status in subsequent steps, and provide strong data support for thermal runaway warning.

[0023] S20: Collect the environmental characteristic parameters of the ternary lithium battery operation, predict the probability of thermal runaway of the lithium battery, obtain the thermal runaway probability, and predict the mutation of the battery characteristic parameters to obtain the predicted mutation time.

[0024] Preferably, the characteristic parameters of the operating environment of the ternary lithium battery are collected. These environmental characteristic parameters include ambient temperature, pressure, humidity, etc., which have a significant impact on the thermal behavior of the battery. By deploying high-precision environmental sensors around the battery, these parameter data can be obtained in real time and accurately. Subsequently, these environmental characteristic parameters are used as input using a trained thermal runaway probability prediction model. Based on the correlation between historical data and environmental factors, the model outputs the probability value of thermal runaway of the lithium battery under current environmental conditions.

[0025] At the same time, to proactively predict potential battery anomalies, a prediction of battery parameter mutations is also required. This prediction process relies on another prediction model, the battery parameter mutation prediction model. This model also uses environmental characteristic parameters as input, but outputs the time span required for a battery parameter (such as temperature) to undergo a preset mutation amplitude (for example, a temperature change amplitude greater than or equal to 10 degrees Celsius), i.e., the predicted mutation time. The preset mutation amplitude can be set based on specific circumstances. In a preferred embodiment, the temperature mutation threshold is set to ΔT ≥ 5°C / min (referring to the temperature rise characteristics before thermal runaway in the national standard GB / T 39222-2020, the measured average temperature rise rate in the five minutes before thermal runaway is greater than 3°C / min, and a safety margin of 5°C / min is adopted); the voltage mutation threshold is set to ΔV ≥ 0.3V / second (corresponding to the voltage drop rate at the initial stage of an internal short circuit, based on 18650 battery abuse test data); and the current mutation threshold is set to ΔI ≥ 2C / second (the normal current change rate under fast charging conditions is less than 1C / second; exceeding 2C / second is considered an abnormal surge).

[0026] Through these two prediction processes, the system can comprehensively evaluate the safety status of the battery under the current environment, providing an important basis for subsequent thermal runaway warning.

[0027] S30: Calculate the actual mutation time of the battery characteristic parameter according to the battery characteristic parameter sequence, and combine it with the predicted mutation time to obtain the unit mutation time.

[0028] Furthermore, after obtaining the battery characteristic parameter sequence, in order to accurately grasp the actual time length of the battery characteristic parameter mutation, the sequence needs to be deeply analyzed. By identifying the moments when the characteristic parameters in the sequence (such as temperature, voltage, etc.) significantly deviate from the normal range and calculating the time span from the normal state to the mutation state, the actual mutation time of the battery characteristic parameter mutation can be obtained.

[0029] However, due to factors such as sensor accuracy, data acquisition frequency, and the complex dynamics within the battery, the calculation of the actual mutation time may contain certain errors. To improve the accuracy of the mutation time estimate, the actual mutation time is fused with the predicted mutation time previously obtained through environmental characteristic parameter prediction. This fusion method can use weighted averaging or dynamically adjust weights based on historical data to fully leverage the complementarity between the prior knowledge of the prediction model and the actual observation data, thereby obtaining a more accurate unit mutation time. This time metric comprehensively considers the results of prediction and actual observation, providing a more reliable time benchmark for subsequent thermal runaway warnings. For example, if the predicted mutation time is 10 seconds, but the actual mutation time is measured to be 12 seconds due to factors such as sensor delay, a weighted average (for example, each weight is 0.5) can be used to obtain a unit mutation time of 11 seconds. This fusion result is closer to the actual situation and helps improve the timeliness and accuracy of thermal runaway warnings.

[0030] Among them, the weight determination method of the unit mutation time fusion algorithm dynamically adjusts the weight based on the historical prediction error. The specific formula is ,in, =|Predicted mutation time t-1 -Actual mutation time t-1 |, which is the error at the previous moment. k=0.5 is a hyperparameter for adjusting the rate of change of weights. α t The value range is (0,1). The larger the error, the higher the actual mutation time weight. For example, when error>5 seconds, α t is 0.8, focusing on trusting real-time data. Furthermore, the unit mutation time calculation formula is: ,in, To predict the time of mutation, is the actual mutation time.

[0031] S40: Optimize thermal runaway warning identification resources based on the unit mutation time and thermal runaway probability to obtain optimal thermal runaway warning discrimination resources, perform thermal runaway identification and warning on each battery characteristic parameter segment, and obtain a thermal runaway warning result.

[0032] Specifically, after obtaining the unit mutation time and thermal runaway probability, in order to improve the response speed of the thermal runaway warning system and ensure the accuracy of the warning, the thermal runaway warning identification resources need to be optimized. Specifically, multiple thermal runaway warning identifier models are first constructed, each of which has the ability to independently identify thermal runaway risks. Subsequently, the number of models involved in the warning identification is dynamically adjusted based on the unit mutation time (i.e., the time required for the battery characteristic parameters to change from a normal state to a mutation state) and the thermal runaway probability (i.e., the possibility of thermal runaway in the battery under current environmental conditions). For example, if the unit mutation time is short and the thermal runaway probability is high, the system will increase the number of warning identifiers to process more battery characteristic parameter segments in parallel, thereby shortening the warning response time; conversely, if the unit mutation time is long and the thermal runaway probability is low, the system will appropriately reduce the number of warning identifiers to save computing resources.

[0033] This optimization process finds the optimal thermal runaway warning resource configuration, maximizing the warning response speed while ensuring warning accuracy. Ultimately, this optimally configured warning identifier is used to perform thermal runaway warnings for each battery characteristic parameter segment. By combining the outputs of each identifier, the final thermal runaway warning result is derived, enabling timely and accurate warnings of thermal runaway risks in ternary lithium batteries.

[0034] In a preferred embodiment, battery characteristic parameters of a ternary lithium battery are monitored to obtain a battery characteristic parameter sequence, which is then divided into segments according to time sequence to obtain a battery characteristic parameter segment sequence, including: monitoring battery characteristic parameters of a ternary lithium battery to obtain a battery characteristic parameter sequence, wherein the battery characteristic parameters include battery temperature; dividing the battery characteristic parameter sequence into segments according to a preset time window to obtain multiple battery characteristic parameter segments, and arranging them to obtain a battery characteristic parameter segment sequence.

[0035] Optionally, in the initial phase of thermal runaway warning for ternary lithium batteries, comprehensive, multi-dimensional monitoring of characteristic parameters is required. Battery temperature, as one of the core parameters, is directly related to the battery's internal chemical reaction rate, internal resistance changes, and heat accumulation effects, making it a key indicator in the monitoring system. Voltage, current, SOC (state of charge) and other parameters are also included in the monitoring scope to fully reflect the battery's operating status.

[0036] Through a high-precision sensor array and a real-time data acquisition system, the instantaneous values of battery characteristic parameters can be continuously obtained and arranged in chronological order to construct a battery characteristic parameter sequence, which fully records the dynamic process of the evolution of battery characteristic parameters over time.

[0037] In order to improve the accuracy and efficiency of subsequent analysis, it is necessary to use a preset time window to divide the battery characteristic parameter sequence into time series. For example, if the time window is set to 1 second, all battery characteristic parameters collected per second will constitute a battery characteristic parameter segment. Multiple consecutive parameter segments are arranged in chronological order to eventually form a battery characteristic parameter segment sequence. This division method not only helps to capture short-term fluctuations in battery status, but also provides a standardized data input format for subsequent thermal runaway warning algorithms based on time series pattern recognition. For example, when the battery temperature shows an abnormal upward trend in several consecutive parameter segments, potential thermal runaway risks can be quickly identified based on sequence analysis.

[0038] In a preferred embodiment, the environmental characteristic parameters of the operation of the ternary lithium battery are collected, the thermal runaway probability of the lithium battery is predicted, the thermal runaway probability is obtained, and the battery characteristic parameter mutation prediction is performed to obtain the predicted mutation time, including: collecting the environmental characteristic parameters in the operating environment of the ternary lithium battery, wherein the environmental characteristic parameters include ambient temperature and pressure; calling a thermal runaway probability predictor and a battery parameter mutation predictor; inputting the environmental characteristic parameters into the thermal runaway probability predictor and the battery parameter mutation predictor respectively, and predicting the output to obtain the thermal runaway probability and the predicted mutation time.

[0039] Specifically, the collection and utilization of environmental characteristic parameters is a key step in building an accurate prediction model. Through the deployment of high-precision sensors in the battery operating environment, core characteristic parameters such as ambient temperature and pressure are collected in real time. Among them, the ambient temperature directly affects the heat dissipation efficiency and internal chemical reaction rate of the battery, while the ambient pressure may affect its thermal stability by changing the gas state inside the battery. After preprocessing, the collected environmental characteristic parameter data is input into the trained thermal runaway probability predictor and battery parameter mutation predictor respectively.

[0040] The thermal runaway probability predictor is based on machine learning algorithms, such as neural networks or support vector machines. It integrates the correlation between parameters such as ambient temperature and pressure and historical thermal runaway cases, and outputs the probability value of thermal runaway of the battery in the current environment. For example, if the ambient temperature is near the upper limit of the optimal operating temperature of the battery for a long time and the pressure fluctuates greatly, the predictor may output a higher probability of thermal runaway. Specifically, the algorithm architecture is constructed using a neural network as an example. Its network structure includes 3 layers of fully connected neural networks, namely input layer-hidden layer 1-hidden layer 2-output layer. Its input layer is set to include 4-dimensional features including ambient temperature, ambient pressure, humidity, and charge and discharge rate. Hidden layer 1 has 20 neurons, activation function ReLU, and Dropout rate 0.2; hidden layer 2 has 10 neurons, activation function ReLU, and Dropout rate 0.1; the output layer is set to 1 neuron, activation function Sigmoid, and outputs a probability value between 0 and 1. During the model training process, the optimizer uses Adam, and the initial value of the learning rate is set to 10−3 , decaying by 0.9 every 50 epochs. The batch size was 32, and the number of iterations was 200. The loss function used was binary cross entropy. The model converged when the validation set loss decreased by less than 0.001 for 10 consecutive epochs and the accuracy was ≥ 95%.

[0041] At the same time, the battery parameter mutation predictor focuses on abnormal changes in battery characteristic parameters, such as temperature and voltage. The time span required for the parameters to undergo a preset mutation amplitude (such as a sudden temperature rise of ≥15°C) is predicted through time series analysis or dynamic threshold detection methods, that is, the predicted mutation time. For example, if the ambient temperature rises sharply and the pressure increases abnormally, the predictor may output a shorter predicted mutation time, indicating that the battery is about to enter a critical state of thermal runaway. Specifically, the algorithm architecture is constructed by combining a single-layer LSTM network (processing time series features) with a fully connected layer. Its network structure includes: input layer, environmental characteristic parameter sequence (time step length is 5, that is, input the environmental data of the last 5 seconds; LSTM layer, 16 memory units, activation function tanh, return the output of the last time step; output layer, 1 neuron, activation function linear, output predicted mutation time, unit: second). During the model training process, the optimizer uses RMSprop, and the learning rate is set to 5×10 −4 The batch size is 16, the number of iterations is 150, and the mean squared error (MSE) loss function is used. The model convergence criteria are when the validation set MSE is less than 1.0 (i.e., the prediction time error is less than 1 second) and R² ≥ 0.9.

[0042] Through the above process, a dynamic correlation analysis between environmental characteristic parameters and battery thermal runaway risks is achieved, providing a scientific basis for subsequent thermal runaway warning and prevention.

[0043] In a preferred embodiment, the training steps of the thermal runaway probability predictor and the battery parameter mutation predictor include: collecting a set of sample environmental characteristic parameters based on the operation data of the ternary lithium battery in the historical time; collecting the average probability of thermal runaway of the ternary lithium battery under different sample environmental characteristic parameters, and marking to obtain a sample thermal runaway probability set; collecting the minimum time for the battery characteristic parameters of the ternary lithium battery to undergo a preset mutation amplitude under different local environmental characteristic parameters, and marking to obtain a sample predicted mutation time set; using machine learning to construct a thermal runaway probability predictor and a battery parameter mutation predictor, wherein the input features of the thermal runaway probability predictor and the battery parameter mutation predictor are environmental characteristic parameters, and the output features are thermal runaway probability and predicted mutation time, respectively; using the sample environmental characteristic parameter set, combined with the sample thermal runaway probability set and the sample predicted mutation time set, the thermal runaway probability predictor and the battery parameter mutation predictor are supervised training until convergence.

[0044] Furthermore, in the process of building a thermal runaway probability predictor and a battery parameter mutation predictor, it is necessary to first extract the operating data of the ternary lithium battery within a specific time period from the historical database, covering environmental characteristic parameters such as ambient temperature, pressure, and humidity. These parameters constitute the sample environmental characteristic parameter set. For example, data collected in a high temperature and high humidity environment in the summer, where the ambient temperature is generally above 35°C and the humidity is greater than 70%, will serve as the basic input dimensions for subsequent model training.

[0045] Subsequently, by analyzing historical data on battery thermal runaway cases under different combinations of environmental characteristic parameters, the average probability of thermal runaway for each sample was calculated and annotated, forming a sample thermal runaway probability set. For example, when the ambient temperature exceeds 45°C and the pressure fluctuates abnormally, the probability of battery thermal runaway may be as high as 30%, and this probability value will be clearly marked in the corresponding sample.

[0046] At the same time, for situations where battery characteristic parameters (such as temperature) experience a preset mutation amplitude (e.g., a temperature surge of ≥20°C), the shortest time span for the mutation to occur under different environmental characteristic parameters is recorded and annotated as the sample predicted mutation time set. For example, under extreme conditions of a rapid increase in ambient temperature and a sudden increase in pressure, the battery temperature may exceed the safety threshold in just 10 seconds. This time value will serve as key annotation information.

[0047] Based on this labeled data, machine learning frameworks such as deep neural networks or gradient boosting trees are used to construct thermal runaway probability predictors and battery parameter mutation predictors, respectively. The core design of these two predictors is to use environmental characteristic parameters as input features and output the two target features of thermal runaway probability and predicted mutation time through nonlinear mapping relationships within the model.

[0048] During the model training phase, the sample environmental characteristic parameter set is paired with the sample thermal runaway probability set and fed into the thermal runaway probability predictor. Using cross-validation and a loss function minimization strategy, such as mean squared error or logarithmic loss, the model parameters are iteratively optimized until the model's prediction performance on the validation set converges to a stable state. Similarly, the sample environmental characteristic parameter set is paired with the sample predicted mutation time set and fed into the battery parameter mutation predictor. Supervised training is performed using a loss function commonly used in regression tasks (such as Huber loss) to ensure the model's prediction accuracy for mutation time.

[0049] Through this process, we ultimately obtain an intelligent predictor that can quickly and accurately output the probability of thermal runaway and predict mutation time based on real-time environmental characteristic parameters, providing decision support for the safe operation of ternary lithium batteries.

[0050] In a preferred embodiment, according to the battery characteristic parameter sequence, the actual mutation time of the battery characteristic parameter mutation is calculated, and combined with the predicted mutation time, the unit mutation time is obtained, including: according to the battery characteristic parameter sequence, the minimum time for the battery characteristic parameter to undergo a preset mutation amplitude is calculated to obtain the actual mutation time; according to the predicted mutation time and the actual mutation time, the unit mutation time is calculated.

[0051] For example, the actual mutation time is quantitatively calculated based on a sequence of collected battery characteristic parameters (covering continuous records of changes in key parameters such as temperature and voltage over time). Specifically, a preset mutation amplitude threshold is set (for example, a sudden temperature rise of ≥18°C or a voltage drop of ≥0.5V). The parameter sequence is scanned to identify the moment when the characteristic parameter first crosses the threshold. The time span from the starting moment to the mutation moment is calculated, which is the actual mutation time. For example, if the battery temperature sequence shows a steady rise from 25°C to 42°C over 100 seconds of continuous monitoring, and then a sudden jump to 45°C at the 103rd second, the actual mutation time is recorded as 3 seconds (i.e., the time difference from the 100th to the 103rd second). The system then combines the actual mutation time with the predicted mutation time obtained earlier through environmental characteristic parameter prediction to generate a unit mutation time. The unit mutation time can be calculated based on a weighted average or a dynamic weight adjustment strategy. For example, if the predicted mutation time is 5 seconds and the actual mutation time is 3 seconds, and based on historical data and current environmental conditions, the predicted and actual values are weighted at 0.4 and 0.6, respectively, then the unit mutation time is calculated as (5 × 0.4 + 3 × 0.6) = 3.8 seconds. This fusion result not only retains the prediction model's prior judgment of environmental factors, but also integrates the dynamic feedback of real-time monitoring data, thereby more accurately reflecting the true time scale of battery characteristic parameter mutations. For example, in an extremely high temperature environment, if the prediction model overestimates the mutation time due to abnormal ambient temperature, but the actual monitoring data shows that the battery's internal thermal management mechanism has effectively delayed the occurrence of mutations, the unit mutation time will be significantly biased towards the actual value, providing a more reliable basis for subsequent early warning decisions.

[0052] In a preferred embodiment, thermal runaway warning identification resources are optimized based on the unit mutation time and thermal runaway probability to obtain optimal thermal runaway warning discrimination resources, and thermal runaway identification and warning are performed for each battery characteristic parameter segment, including: constructing a thermal runaway warning identifier array; randomly selecting and configuring a first number of thermal runaway warning identifiers; obtaining the time for the thermal runaway warning identifiers under the first number to perform thermal runaway identification and warning calculations on the battery characteristic parameter segment to obtain a first warning time; calculating and obtaining a first thermal runaway warning score based on the first warning time, the first number, the unit mutation time and the thermal runaway probability; continuing to randomly configure a random number of thermal runaway warning identifiers, performing quantity optimization, and obtaining an optimal number with the largest warning score; randomly selecting the optimal number of thermal runaway warning identifiers, inputting each battery characteristic parameter segment, and outputting an optimal number of thermal runaway identification results, screening the thermal runaway identification results with the largest occurrence ratio, and obtaining a thermal runaway warning result, wherein the thermal runaway identification result includes yes or no, and warning processing is performed when the thermal runaway warning result is yes.

[0053] Preferably, an array of thermal runaway warning identifiers is constructed, comprising multiple types or parameter configurations. These identifiers are built based on algorithms such as machine learning and deep learning, and are capable of extracting risk signatures from characteristic battery parameter segments and determining whether thermal runaway has occurred. Next, a first number of thermal runaway warning identifiers are randomly selected to participate in the warning calculation, for example, 5 are randomly selected from 20 trained identifiers. After the identifiers are selected, the real-time acquired battery characteristic parameter segments (e.g., a time series consisting of parameters such as temperature and voltage) are input into these identifiers. The time required for each identifier to complete the thermal runaway identification and warning calculation and output a result ("yes" or "no") is recorded, and this time is used as the first warning time. For example, if the majority of the five identifiers complete the calculation and output the result within 1 second, the first warning time is recorded as 1 second.

[0054] Subsequently, a thermal runaway warning scoring function is constructed based on the first warning time, the first number of identifiers, the unit mutation time (characterizing the urgency of the battery characteristic parameter mutation), and the thermal runaway probability (characterizing the possibility of thermal runaway under the current environment and battery state). For example, the function can comprehensively consider the ratio of the warning time to the mutation time, the consistency of the warning result and the thermal runaway probability, and other dimensions to quantitatively evaluate the warning effectiveness under the current identifier configuration and calculate the first thermal runaway warning score. Specifically, the two-dimensional scoring formula is: time score = , the larger the value, the faster the response, the ideal value is ≥1; Accuracy score = , where N is the number of recognizers, P risk is the probability of thermal runaway. Finally, the comprehensive score = λ×time score + (1−λ)×accuracy score, where λ=0.6, the time priority weight, can be adjusted according to the scenario. alertis the identifier array processing time (seconds), which must satisfy T alert ≤T unit ×0.8, (reserve 20% safety time).

[0055] To further optimize early warning resources, different numbers of thermal runaway warning identifiers are randomly configured iteratively, and the aforementioned warning time recording and scoring calculation process is repeated. For example, in subsequent iterations, 8 and 12 identifiers are selected for testing, respectively. The optimal number of identifiers with the highest warning score is ultimately determined, achieving the best balance between warning timeliness and resource consumption. After determining the optimal number, the same number of thermal runaway warning identifiers is randomly selected again, and the real-time battery characteristic parameter segments are input into these identifiers in parallel to obtain multiple thermal runaway identification results. A majority voting mechanism is then used to select the identification result with the highest occurrence rate as the final thermal runaway warning result. For example, if 9 out of 12 identifiers output "yes," the thermal runaway warning result is determined to be "yes." If the thermal runaway warning result is "yes," the system immediately triggers the warning process, such as activating the battery cooling system, disconnecting the power supply, or sending an alert to maintenance personnel. This effectively reduces the risk of thermal runaway accidents and ensures the safe operation of the battery system.

[0056] In a preferred embodiment, a thermal runaway warning identifier array is constructed, including: collecting a set of sample battery characteristic parameter segments based on thermal management data of a ternary lithium battery over a historical period, and marking sample thermal runaway identification results based on whether the ternary lithium battery has thermal runaway under different sample battery characteristic parameter segments, to obtain a set of sample thermal runaway identification results; randomly dividing the sample battery characteristic parameter segment set and the sample thermal runaway identification result set to obtain multiple groups of thermal runaway warning training data, wherein there is a data intersection between every two groups of thermal runaway warning training data; using machine learning to construct multiple thermal runaway warning identifiers, wherein the input feature of each thermal runaway warning identifier is a battery characteristic parameter segment, and the output feature is a thermal runaway identification result; using the multiple groups of thermal runaway warning training data respectively, the multiple thermal runaway warning identifiers are supervised trained until convergence to obtain a thermal runaway warning identifier array.

[0057] Furthermore, in the process of constructing a thermal runaway early warning identifier array, it is necessary to extract a set of sample battery characteristic parameter segments covering key parameters such as temperature, voltage, and current based on the historical thermal management data of ternary lithium batteries. These parameter segments record the operating status of the battery under different working conditions in the form of a time series. Subsequently, the samples are labeled according to whether the battery has experienced thermal runaway during the corresponding period of each parameter segment in the historical records. For example, if the battery temperature soars abnormally and is accompanied by a sudden drop in voltage during the recording of a certain parameter segment, which eventually leads to a thermal runaway event, the parameter segment is labeled as "yes" (thermal runaway occurs); conversely, if the battery state is stable and there is no abnormality, it is labeled as "no" (thermal runaway does not occur), thereby forming a sample thermal runaway identification result set.

[0058] To improve model generalization and training efficiency, an overlapping sampling strategy was used to randomly partition the set of sample battery characteristic parameter segments and the set of sample thermal runaway identification results. This generated multiple sets of thermal runaway warning training data, each containing partially overlapping parameter segments. For example, a sliding window method was used to extract multiple subsets from the complete dataset, with a certain proportion of duplicate samples retained between adjacent subsets. This ensured that the training data covered a wide range of operating conditions and enhanced the model's ability to recognize similar scenarios through data intersection.

[0059] Furthermore, based on the above training data, a machine learning framework (such as random forests in ensemble learning, gradient boosting trees, or convolutional neural networks in deep learning) is used to construct multiple thermal runaway warning identifiers. Each identifier uses the battery characteristic parameter segment as input features and outputs the thermal runaway identification result ("yes" or "no") through nonlinear mapping.

[0060] During the training phase, supervised training is performed on each recognizer using multiple sets of training data. For example, cross-validation techniques are used to divide the data into a training set and a validation set. Model parameters (such as the number of trees in a random forest or the weights in a deep learning network) are iteratively optimized to minimize the loss function (such as cross-entropy loss) on the validation set until the model's performance metrics (such as accuracy and recall) on the validation set reach a convergence threshold. For example, training is considered converged when the validation set accuracy fluctuates by less than 0.5% over five consecutive iterations.

[0061] Finally, by integrating multiple independently trained and stable thermal runaway warning identifiers in parallel, a thermal runaway warning identifier array is formed. This array can comprehensively utilize the advantages of each identifier and output the final warning result through voting or weighted fusion mechanism, effectively improving the reliability and robustness of thermal runaway warning.

[0062] In a preferred embodiment, a first thermal runaway warning score is calculated based on the first warning time, the first quantity, the unit mutation time and the thermal runaway probability, including: calculating the ratio of the unit mutation time to the first warning time to obtain a first time warning score; calculating the ratio of the first quantity to the thermal runaway probability to obtain a first accurate warning score; and calculating the first thermal runaway warning score based on the first time warning score and the first accurate warning score.

[0063] Specifically, in the thermal runaway warning effectiveness evaluation system, a first-time warning score is first constructed based on the unit mutation time (the shortest time required for a battery characteristic parameter to mutate to a critical state) and the first warning time (the time required for a specific number of thermal runaway warning identifiers to complete the identification task). This score quantifies the response speed by calculating the ratio of the unit mutation time to the first warning time. For example, if the unit mutation time is 10 seconds and the first warning time is only 1 second, the first-time warning score is 10. A higher score indicates a faster system response to potential thermal runaway risks. Conversely, if the first warning time is extended to 5 seconds due to redundant identifier configuration or inefficient algorithm, the score drops to 2, reflecting a decrease in response timeliness.

[0064] At the same time, to assess the accuracy of the early warning, a first accurate early warning score is calculated. This score is constructed by the ratio of the first number (the number of identifiers currently participating in the early warning) to the thermal runaway probability (characterizing the objective possibility of thermal runaway occurring in the current battery state). For example, when the first number is 20 identifiers and the thermal runaway probability is 2%, the first accurate early warning score is 10. This score design is based on probability compensation logic, that is, by increasing the number of identifiers, the risk of false positives / missed negatives is reduced, thereby indirectly improving the accuracy of the early warning. However, it should be noted that if the probability of thermal runaway suddenly increases to 10% due to changes in the environment or battery state, the score will be adjusted to 2, indicating that the early warning results need to be evaluated more cautiously in high-risk scenarios.

[0065] Ultimately, the first thermal runaway warning score is obtained by weightedly combining the first-time warning score and the first-accuracy warning score. For example, to balance response speed and accuracy, a normalization coefficient of 0.1 is applied to the first-accuracy warning score (because its value is typically large, such as the first number, which may reach tens or hundreds). The first-time warning score retains its original weight, and the weighted sum of the two is used to obtain a comprehensive score. For example, if the first-time warning score is 8 (unit mutation time is 12 seconds, first warning time is 1.5 seconds) and the first-accuracy warning score is 50 (first number of 50 identifiers, thermal runaway probability is 1%), then the first thermal runaway warning score is 8 + 50 × 0.1 = 13. This scoring system achieves a joint optimization of the warning system's timeliness, accuracy, and resource consumption through dynamic weight adjustment and normalization strategies, providing a quantitative basis for subsequent optimization of the number of warning identifier arrays.

[0066] The embodiment of the present invention provides a time series prediction method for thermal runaway warning of a ternary lithium battery, which has at least the following technical effects:

[0067] 1. By monitoring battery characteristic parameters in real time and dividing the parameter segment sequence according to time series, and combining environmental characteristic parameters to construct a two-dimensional input feature system, the joint prediction of thermal runaway probability and parameter mutation time is achieved. This breaks through the limitations of traditional single parameter or static threshold warning. The dynamic evolution of battery status is captured through the time series parameter sequence. Combined with the nonlinear influence of environmental factors on thermal runaway risk, the spatiotemporal resolution and accuracy of the warning model are significantly improved.

[0068] 2. The concept of unit mutation time is proposed. By calculating the ratio of actual mutation time to predicted mutation time, the timeliness of the early warning system's response to battery state mutations is quantitatively evaluated. This indicator is combined with the thermal runaway probability to construct a dual-objective optimization model, which drives the dynamic adjustment of the number of early warning identifier arrays, achieving a dynamic balance between warning timeliness and computational cost.

[0069] 3. Construct a thermal runaway warning identifier array based on an overlapping sampling strategy. By randomly dividing the training data and retaining the data intersection, each identifier learns the characteristic distribution under different operating conditions. Combined with the majority voting mechanism to screen high-frequency recognition results, it effectively suppresses the misjudgment caused by data noise or overfitting of a single model, and significantly improves the robustness and interpretability of the warning results.

[0070] Example 2:

[0071] like Figure 2 As shown, based on the same inventive concept as the method for thermal runaway warning of a ternary lithium battery using time series prediction provided in Example 1, an embodiment of the present invention further provides a thermal runaway warning system for a ternary lithium battery using time series prediction, the system comprising:

[0072] The data monitoring module 11 is used to monitor the battery characteristic parameters of the ternary lithium battery, obtain a battery characteristic parameter sequence, and obtain a battery characteristic parameter segment sequence according to time sequence.

[0073] The parameter acquisition module 12 is used to collect the environmental characteristic parameters of the ternary lithium battery operation, predict the probability of thermal runaway of the lithium battery, obtain the thermal runaway probability, and predict the sudden change of the battery characteristic parameters to obtain the predicted mutation time.

[0074] The time calculation module 13 is used to calculate the actual mutation time of the battery characteristic parameter mutation according to the battery characteristic parameter sequence, and obtain the unit mutation time by combining the predicted mutation time.

[0075] The identification and warning module 14 is used to optimize the thermal runaway warning identification resources based on the unit mutation time and the thermal runaway probability, obtain the optimal thermal runaway warning discrimination resources, perform thermal runaway identification and warning for each battery characteristic parameter segment, and obtain a thermal runaway warning result.

[0076] Furthermore, the data monitoring module 11 is further configured to perform the following steps:

[0077] The battery characteristic parameters of the ternary lithium battery are monitored to obtain a battery characteristic parameter sequence, wherein the battery characteristic parameters include battery temperature; the battery characteristic parameter sequence is divided into time series according to a preset time window to obtain multiple battery characteristic parameter segments, which are arranged to obtain a battery characteristic parameter segment sequence.

[0078] Furthermore, the parameter acquisition module 12 is further configured to perform the following steps:

[0079] The environmental characteristic parameters in the operating environment of the ternary lithium battery are collected, wherein the environmental characteristic parameters include ambient temperature and pressure; a thermal runaway probability predictor and a battery parameter mutation predictor are called; the environmental characteristic parameters are input into the thermal runaway probability predictor and the battery parameter mutation predictor respectively, and the prediction output obtains the thermal runaway probability and the predicted mutation time.

[0080] Furthermore, the parameter acquisition module 12 is further configured to perform the following steps:

[0081] According to the operation data of the ternary lithium battery in the historical time, a set of sample environmental characteristic parameters is collected; the average probability of thermal runaway of the ternary lithium battery under different sample environmental characteristic parameters is collected, and the sample thermal runaway probability set is marked; the minimum time for the battery characteristic parameters of the ternary lithium battery to undergo a preset mutation amplitude under different local environmental characteristic parameters is collected, and the sample predicted mutation time set is marked; machine learning is used to construct a thermal runaway probability predictor and a battery parameter mutation predictor, wherein the input features of the thermal runaway probability predictor and the battery parameter mutation predictor are environmental characteristic parameters, and the output features are thermal runaway probability and predicted mutation time, respectively; the sample environmental characteristic parameter set is used, and the sample thermal runaway probability set and the sample predicted mutation time set are respectively combined to perform supervised training on the thermal runaway probability predictor and the battery parameter mutation predictor until convergence.

[0082] Furthermore, the time calculation module 13 is further configured to perform the following steps:

[0083] According to the battery characteristic parameter sequence, the minimum time for the battery characteristic parameter to undergo a preset mutation amplitude is calculated to obtain the actual mutation time; and according to the predicted mutation time and the actual mutation time, the unit mutation time is calculated to obtain.

[0084] Furthermore, the identification and warning module 14 is further configured to perform the following steps:

[0085] Construct a thermal runaway warning identifier array; randomly select and configure a first number of thermal runaway warning identifiers; obtain the time for the thermal runaway warning identifiers under the first number to perform thermal runaway identification and warning calculations on the battery characteristic parameter segments to obtain a first warning time; calculate and obtain a first thermal runaway warning score based on the first warning time, the first number, the unit mutation time and the thermal runaway probability; continue to randomly configure a random number of thermal runaway warning identifiers, optimize the number, and obtain an optimal number with the largest warning score; randomly select the optimal number of thermal runaway warning identifiers, input each battery characteristic parameter segment, and output the optimal number of thermal runaway identification results, screen the thermal runaway identification results with the highest occurrence ratio, and obtain a thermal runaway warning result, wherein the thermal runaway identification result includes yes or no, and warning processing is performed when the thermal runaway warning result is yes.

[0086] Furthermore, the identification and warning module 14 is further configured to perform the following steps:

[0087] According to the thermal management data of the ternary lithium battery in the historical period, a set of sample battery characteristic parameter segments is collected, and according to whether the ternary lithium battery has thermal runaway under different sample battery characteristic parameter segments, the sample thermal runaway identification results are marked to obtain a set of sample thermal runaway identification results; the sample battery characteristic parameter segment set and the sample thermal runaway identification result set are randomly divided to obtain multiple groups of thermal runaway warning training data, wherein there is a data intersection between every two groups of thermal runaway warning training data; machine learning is used to construct multiple thermal runaway warning identifiers, wherein the input feature of each thermal runaway warning identifier is the battery characteristic parameter segment, and the output feature is the thermal runaway identification result; the multiple groups of thermal runaway warning training data are respectively used to supervise the multiple thermal runaway warning identifiers until convergence to obtain a thermal runaway warning identifier array.

[0088] Furthermore, the identification and warning module 14 is further configured to perform the following steps:

[0089] Calculate the ratio of the unit mutation time to the first warning time to obtain a first time warning score; calculate the ratio of the first quantity to the thermal runaway probability to obtain a first accurate warning score; and calculate a first thermal runaway warning score based on the first time warning score and the first accurate warning score.

[0090] Through the above detailed description of a time-series prediction method for thermal runaway warning of a ternary lithium battery in this specification, those skilled in the art can clearly understand a time-series prediction system for thermal runaway warning of a ternary lithium battery in this embodiment. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0091] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A time series prediction method for thermal runaway warning of ternary lithium batteries, characterized in that: The method comprises: Monitor the battery characteristic parameters of the ternary lithium battery to obtain a battery characteristic parameter sequence, and divide it into segments according to the time sequence to obtain a battery characteristic parameter segment sequence; Collect the environmental characteristic parameters of the ternary lithium battery operation, predict the probability of thermal runaway of the lithium battery, obtain the thermal runaway probability, and predict the mutation of the battery characteristic parameters to obtain the predicted mutation time; Calculate the actual mutation time of the battery characteristic parameter mutation according to the battery characteristic parameter sequence, and combine it with the predicted mutation time to obtain the unit mutation time; According to the unit mutation time and thermal runaway probability, thermal runaway warning identification resources are optimized to obtain optimal thermal runaway warning discrimination resources, and thermal runaway identification warning is performed on each battery characteristic parameter segment to obtain a thermal runaway warning result.

2. The method for early warning of thermal runaway of a ternary lithium battery based on time series prediction according to claim 1, characterized in that: Monitor the battery characteristic parameters of the ternary lithium battery to obtain the battery characteristic parameter sequence, and divide it into time series to obtain the battery characteristic parameter segment sequence, including: Monitoring battery characteristic parameters of a ternary lithium battery to obtain a battery characteristic parameter sequence, wherein the battery characteristic parameters include battery temperature; The battery characteristic parameter sequence is divided into time series according to a preset time window to obtain multiple battery characteristic parameter segments, which are arranged to obtain a battery characteristic parameter segment sequence.

3. The method for early warning of thermal runaway of a ternary lithium battery based on time series prediction according to claim 1, characterized in that: Collect the environmental characteristic parameters of the ternary lithium battery operation, predict the probability of thermal runaway of the lithium battery, obtain the thermal runaway probability, and predict the mutation of the battery characteristic parameters to obtain the predicted mutation time, including: Collect environmental characteristic parameters in the operating environment of the ternary lithium battery, wherein the environmental characteristic parameters include ambient temperature and pressure; Call the thermal runaway probability predictor and battery parameter mutation predictor; The environmental characteristic parameters are input into the thermal runaway probability predictor and the battery parameter mutation predictor respectively, and the prediction output obtains the thermal runaway probability and the predicted mutation time.

4. The method for early warning of thermal runaway of a ternary lithium battery based on time series prediction according to claim 3, characterized in that: The training steps of the thermal runaway probability predictor and the battery parameter mutation predictor include: Based on the historical operating data of ternary lithium batteries, a set of sample environmental characteristic parameters is collected; Collect the average probability of thermal runaway of ternary lithium batteries under different sample environmental characteristic parameters, and mark the obtained sample thermal runaway probability set; Collect the minimum time for the battery characteristic parameters of the ternary lithium battery to undergo a preset mutation amplitude under different environmental characteristic parameters, and mark the sample predicted mutation time set; Using machine learning, a thermal runaway probability predictor and a battery parameter mutation predictor are constructed, wherein the input features of the thermal runaway probability predictor and the battery parameter mutation predictor are environmental characteristic parameters, and the output features are the thermal runaway probability and the predicted mutation time, respectively; The sample environment characteristic parameter set is used, and the sample thermal runaway probability set and the sample predicted mutation time set are respectively combined to perform supervised training on the thermal runaway probability predictor and the battery parameter mutation predictor until convergence.

5. The method for early warning of thermal runaway of a ternary lithium battery based on time series prediction according to claim 1, characterized in that: Calculating the actual mutation time of the battery characteristic parameter mutation according to the battery characteristic parameter sequence and combining it with the predicted mutation time to obtain the unit mutation time includes: Calculating the minimum time for a battery characteristic parameter to undergo a preset mutation amplitude based on the battery characteristic parameter sequence to obtain an actual mutation time; The unit mutation time is calculated based on the predicted mutation time and the actual mutation time.

6. The method for early warning of thermal runaway of a ternary lithium battery based on time series prediction according to claim 1, characterized in that: Based on the unit mutation time and thermal runaway probability, thermal runaway warning identification resources are optimized to obtain the optimal thermal runaway warning discrimination resources. Thermal runaway identification and warning are performed for each battery characteristic parameter segment, including: Build a thermal runaway warning identifier array; Randomly selecting and configuring a first number of thermal runaway warning identifiers; Obtaining the time for the thermal runaway warning identifier under the first quantity to perform thermal runaway identification and warning calculation on the battery characteristic parameter segments to obtain a first warning time; Calculate a first thermal runaway warning score according to the first warning time, the first quantity, the unit mutation time, and the thermal runaway probability; Continue to randomly configure a random number of thermal runaway warning identifiers and optimize the number to obtain the optimal number with the highest warning score; The optimal number of thermal runaway warning identifiers is randomly selected, each battery characteristic parameter segment is input, and the optimal number of thermal runaway identification results is output. The thermal runaway identification results with the largest occurrence ratio are screened to obtain thermal runaway warning results, wherein the thermal runaway identification results include yes or no, and warning processing is performed when the thermal runaway warning result is yes.

7. The method for early warning of thermal runaway of a ternary lithium battery based on time series prediction according to claim 6, characterized in that: Build a thermal runaway warning identifier array, including: Based on the thermal management data of the ternary lithium battery over a long period of time, a set of characteristic parameter segments of the sample batteries is collected, and based on whether the ternary lithium battery has thermal runaway under different characteristic parameter segments of the sample batteries, the sample thermal runaway identification results are marked to obtain a set of sample thermal runaway identification results; Randomly dividing the sample battery characteristic parameter segment set and the sample thermal runaway identification result set to obtain multiple groups of thermal runaway warning training data, wherein there is a data intersection between every two groups of thermal runaway warning training data; Using machine learning, we build multiple thermal runaway warning identifiers. The input features of each thermal runaway warning identifier are battery characteristic parameter segments, and the output features are thermal runaway identification results. The plurality of groups of thermal runaway warning training data are respectively used to perform supervised training on the plurality of thermal runaway warning identifiers until convergence, thereby obtaining a thermal runaway warning identifier array.

8. The method for early warning of thermal runaway of a ternary lithium battery based on time series prediction according to claim 6, characterized in that: Calculating a first thermal runaway warning score according to the first warning time, the first quantity, the unit mutation time, and the thermal runaway probability includes: Calculating the ratio of the unit mutation time to the first warning time to obtain a first time warning score; calculating a ratio of the first number to the thermal runaway probability to obtain a first accurate warning score; A first thermal runaway warning score is calculated based on the first time warning score and the first accuracy warning score.

9. A time series prediction thermal runaway warning system for ternary lithium batteries, characterized in that: A method for early warning of thermal runaway of a ternary lithium battery for implementing a time series prediction according to any one of claims 1 to 8, the system comprising: The data monitoring module is used to monitor the battery characteristic parameters of the ternary lithium battery, obtain the battery characteristic parameter sequence, and divide it into battery characteristic parameter segment sequences according to the time sequence; The parameter acquisition module is used to collect the environmental characteristic parameters of the ternary lithium battery operation, predict the probability of thermal runaway of the lithium battery, obtain the thermal runaway probability, and predict the sudden change of the battery characteristic parameters to obtain the predicted mutation time; A time calculation module, configured to calculate the actual mutation time of the battery characteristic parameter mutation according to the battery characteristic parameter sequence, and obtain the unit mutation time by combining the predicted mutation time; The identification and warning module is used to optimize the thermal runaway warning identification resources based on the unit mutation time and the thermal runaway probability, obtain the optimal thermal runaway warning discrimination resources, perform thermal runaway identification and warning for each battery characteristic parameter segment, and obtain a thermal runaway warning result.

Images