Early warning method and device for thermal runaway state of battery
Through the combination of pooling processing and long-term memory models, the battery charge status is quickly and accurately monitored, which solves the problem of low accuracy of battery thermal runaway warning in the existing technology, and realizes timely warning of battery thermal runaway, reducing fire or Explosion risk.
Patent Information
- Application Number
- CN202411469644.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-21
- Publication Date
- 2025-01-03
AI Technical Summary
The existing battery thermal runaway warning technology relies on the experience of operation and maintenance personnel, and has low accuracy, and the prediction accuracy of deep learning models is not high, so it cannot quickly identify the battery status, resulting in timely warning, which increases battery fire or explosion. risk.
By obtaining the characteristic data set of the battery during the charging process, pooling is performed to extract higher-order features, input it into the long and short-term memory model, predict the state of charge, and control the execution of the battery management system based on the highest charging voltage and state of charge level. Early warning operation.
It realizes fast and accurate battery charge status monitoring, reduces the processing volume of long and short-term memory models, can promptly warn, avoid the risk of thermal runaway, and improves the safety of the battery charging process.
Smart Images
Figure CN119247158A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of battery charging warning, and in particular to a battery thermal runaway state warning method and device. Background Art
[0002] With the development of large-scale energy storage devices, renewable energy sources such as wind and hydroelectric power can now be stored in the form of electricity. Energy storage batteries are widely used due to their stable electrochemical characteristics. During the charging of energy storage batteries, the performance of their internal energy storage units may undergo irreversible degradation due to changes in electrochemical composition, leading to thermal runaway of the energy storage battery and then internal short circuit of the battery. The whole process generates heat inside the battery, accelerating the thermal runaway process, making the thermal runaway faster and causing accidents such as fire or explosion.
[0003] Among existing technologies, traditional methods rely more on the experience of on-site maintenance personnel and manufacturer professionals to make judgments, and depend on the operator's understanding of the system and its current status. Its accuracy cannot be guaranteed and is only applicable to more obvious faults. Existing technologies also use deep learning models for monitoring, but this method has a low prediction accuracy and requires a large amount of data to be processed, so it is impossible to quickly and accurately identify the battery status to issue a timely warning. Summary of the invention
[0004] The present application provides a battery thermal runaway state warning method and device thereof to solve one or more technical problems existing in the prior art and at least provide a beneficial option or create conditions.
[0005] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by the practice of the present application.
[0006] According to one aspect of an embodiment of the present application, a battery thermal runaway state warning method is provided, the method comprising: Acquire a characteristic data set of the battery during the charging process, wherein the characteristic data set includes at least one time series data associated with the battery characteristics; Performing pooling processing on each of the time series data respectively to obtain each target output data corresponding to each of the time series data; Using each of the target output data as pooled sequence data, and obtaining a battery parameter corresponding to a first target moment in the pooled sequence data; Inputting the battery parameters into a pre-established long short-term memory model to obtain a predicted state of charge value of the battery at a second target time; Acquire the maximum charging voltage of the battery in each target time period during the charging process based on the characteristic data set; The thermal runaway state level of the battery is determined according to the highest charging voltage in the target time period where the first target moment is located and the state of charge prediction value, and the battery management system is controlled to perform a target warning operation based on the thermal runaway state level.
[0007] In one embodiment of the present application, based on the above solution, a single time series data is pooled, including the following steps: Dividing the time series data into a preset number of data regions, inputting all data of the time series data into each of the data regions, respectively, to obtain a pooled data set corresponding to each of the data regions; Average pooling processing, maximum pooling processing and minimum pooling processing are performed on each of the pooled data sets.
[0008] In one embodiment of the present application, based on the aforementioned scheme, the average pooling processing is: averaging all the data in the pooled data value to obtain the average value of the pooled data value, the maximum pooling processing is: finding the maximum data value among all the data in the pooled data value, and the minimum pooling processing is: finding the minimum data value among all the data in the pooled data value.
[0009] In one embodiment of the present application, based on the above scheme, the target output data can be obtained by the following method: If the difference between the maximum data value and the minimum data value is greater than the average value, the difference between the maximum data value and the minimum data value is used as the target output data; If the difference between the maximum data value and the minimum data value is less than or equal to the average value, half of the sum of the maximum data value and the average value is used as the target output data.
[0010] In one embodiment of the present application, based on the above scheme, the pre-established long short-term memory model includes a first layer LSTM model and a second layer LSTM model, and the battery parameters are input into the pre-established long short-term memory model to obtain the predicted value of the state of charge of the battery, including: Inputting the battery parameter into the first-layer LSTM model, performing a first logic operation on the battery parameter according to a first operation logic corresponding to the first-layer LSTM model, and obtaining a first logic output value; Inputting the first logic output value into the second-layer LSTM model, performing a second logic operation on the first logic output value according to a second operation logic corresponding to the second-layer LSTM model, to obtain a second logic output value; The second logic output value is used as the state of charge prediction value.
[0011] In one embodiment of the present application, based on the above solution, the step of obtaining a characteristic data set of the battery during the charging process includes: Acquiring current sequence data, voltage sequence data, temperature sequence data and power sequence data of the battery during charging; Extracting features from the current sequence data according to a characteristic relationship between the current of the battery and the state of charge of the battery to obtain first characteristic sequence data; Extracting features from the voltage sequence data according to a characteristic relationship between the voltage of the battery and the state of charge of the battery to obtain second characteristic sequence data; Extracting features from the temperature series data according to a characteristic relationship between the temperature of the battery and the state of charge of the battery to obtain third characteristic series data; Extracting features from the power sequence data according to a characteristic relationship between the power of the battery and the state of charge of the battery to obtain fourth characteristic sequence data; Generate the feature data set based on the first feature sequence data, the second feature sequence data, the third feature sequence data and the fourth feature sequence data; Among them, the first feature sequence data, the second feature sequence data, the third feature sequence data and the fourth feature sequence data are respectively time series data related to current characteristics, time series data related to voltage characteristics, time series data related to temperature characteristics, and time series data related to power characteristics.
[0012] In one embodiment of the present application, based on the above solution, the method further includes: Determine a first loss value of the first layer LSTM model after performing a first logic operation based on the first logic output value, and determine a second loss value of the second layer LSTM model after performing a second logic operation based on the second logic output value; The long short-term memory model is trained according to the first loss value and the second loss value.
[0013] In one embodiment of the present application, based on the above solution, determining the thermal runaway state level of the battery according to the highest charging voltage in the target time period where the first target time is located and the state of charge prediction value includes: If the predicted state of charge value is lower than a first preset threshold value, and the maximum charging voltage is greater than a preset voltage threshold value, determining that the thermal runaway state level is level one; If the state of charge prediction value is greater than or equal to a first preset threshold value and less than a second preset threshold value, and the maximum charging voltage is greater than a preset voltage threshold value, determining that the thermal runaway state level is level 2; If the state of charge prediction value is greater than or equal to a second preset threshold and the maximum charging voltage is greater than a preset voltage threshold, it is determined that the thermal runaway state level is level three.
[0014] In one embodiment of the present application, based on the above solution, the controlling the battery management system to perform a target warning operation based on the thermal runaway state level includes: If the thermal runaway state level is level one, controlling the battery management system to send a prompt signal indicating that the battery state is abnormal to the terminal; If the thermal runaway state level is level 2, controlling the battery management system to send an alarm signal indicating that the battery state is at risk to the terminal; If the thermal runaway state level is level three, the battery management system is controlled to cut off the power supply of the battery and send an alarm signal that the battery is in a hot state to the terminal.
[0015] According to one aspect of an embodiment of the present application, a battery thermal runaway state warning device is provided, the device comprising: A first acquisition unit, configured to acquire a characteristic data set of the battery during the charging process, wherein the characteristic data set includes at least one time series data associated with a battery characteristic; A pooling unit, used to perform pooling processing on each of the time series data respectively to obtain each target output data corresponding to each of the time series data; a second acquisition unit, configured to use each of the target output data as pooled sequence data, and acquire a battery parameter corresponding to a first target moment from the pooled sequence data; A prediction unit, used for inputting the battery parameters into a pre-established long short-term memory model to obtain a predicted value of the state of charge of the battery at a second target time; A third acquisition unit, configured to acquire, based on the characteristic data set, a maximum charging voltage of the battery in each target time period during a charging process; An early warning unit is used to determine the thermal runaway state level of the battery according to the highest charging voltage in the target time period where the first target moment is located and the charge state prediction value, and perform a target early warning operation based on the thermal runaway state level.
[0016] The beneficial effects of this application are: By pooling the time series data in the feature data set, that is, by extracting the high-order features of each time series data to form each target output data, the amount of data that needs to be processed by the subsequent long-short-term memory model is reduced. By obtaining the battery parameters corresponding to the first target moment in the pooled sequence data and inputting them into the pre-established long-short-term memory model, the state of charge prediction value of the battery at the second target moment can be obtained, so that the thermal runaway state level of the battery can be determined according to the highest charging voltage of the target time period where the first target moment is located and the state of charge prediction value, and the battery management system is controlled to perform the target warning operation based on the thermal runaway state level.
[0017] Therefore, pooling processing can greatly reduce the processing volume of the long short-term memory model. At the same time, the long short-term memory model can be used to accurately predict the battery's state of charge. The present application can quickly and accurately predict the battery's state of charge at future times based on the real-time battery data during the current charging process. It can accurately monitor the battery's charging operation status in real time and effectively avoid the risk of thermal runaway of the battery during the charging process. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following is a brief description of the drawings required for use in the description of the embodiments. Obviously, the drawings described are only part of the embodiments of the present application, not all of the embodiments, and those skilled in the art can also obtain other design solutions and drawings based on these drawings without creative work.
[0019] Figure 1 Flowchart of a battery thermal runaway state warning method according to an embodiment of the present application; Figure 2 A graph showing the relationship between the state of charge and voltage (a), current (b), temperature (c), and power (d) according to an embodiment of the present application; Figure 3 is a block diagram of a battery thermal runaway state warning device according to an embodiment of the present application; Figure 4 It is a logic diagram of the internal operation of the pooling layer according to an embodiment of the present application. DETAILED DESCRIPTION
[0020] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more comprehensive and complete and fully convey the concept of the example embodiments to those skilled in the art.
[0021] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present application. However, those skilled in the art will appreciate that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, known methods, devices, realizations or operations are not shown or described in detail to avoid blurring the various aspects of the application.
[0022] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or micro-control node devices.
[0023] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.
[0024] It should be noted that the "multiple" mentioned in this article refers to two or more. "And / or" describes the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.
[0025] The following is a further introduction to the background technology of this application: With the development of large-scale energy storage devices, renewable energy sources such as wind and hydropower can now be stored in the form of electricity. Energy storage batteries are widely used due to their stable electrochemical characteristics. During the charging of energy storage batteries, the performance of their internal energy storage units may undergo irreversible degradation due to changes in electrochemical composition, leading to thermal runaway of the energy storage battery and then internal short circuit of the battery. The whole process generates heat inside the battery, which accelerates the thermal runaway process, speeds up the thermal runaway process, and causes accidents such as fire or explosion. In order to ensure the reliable operation of energy storage devices, the development of an early warning method for the thermal runaway state of energy storage batteries has become the key to current research.
[0026] Among the existing technologies, traditional methods rely more on the experience of on-site maintenance personnel and manufacturer professionals to make judgments, and depend on the operator's understanding of the system and its current status. Its accuracy cannot be guaranteed and is only applicable to more obvious faults. At present, advanced deep learning methods are used for charging fault warning of energy storage batteries. They mainly analyze the historical charging operation data of energy storage batteries to construct a mapping relationship between feature quantities and battery operating status. For example: using neural network algorithms to achieve fault and fire warnings for energy storage batteries, the key to the method is to determine appropriate input features to reflect the charging and operation attenuation characteristics of energy storage batteries as comprehensively as possible. However, the existing technologies, monitoring and estimation accuracy are not high, and these methods are time-consuming and memory-consuming when used, and require optimization of algorithms and data processing.
[0027] The purpose of this application is to provide a battery thermal runaway state warning method. First, the corresponding relationship between the state of charge of the energy storage battery and the battery temperature, voltage, current, and system power is analyzed to obtain relevant features to provide a basis for subsequent model training and prediction. Then, a CNN (Convolutional Neural Networks)-LSTM (long and short-term memory) model is constructed, and an improved pooling layer is proposed to further improve the rapid and accurate identification and analysis of the battery state of charge.
[0028] The present application can reduce the amount of data processing through pooling processing, so that the long short-term memory model can accurately and quickly output the battery's state of charge prediction value, monitor the battery status in real time, and provide timely warnings when thermal runaway occurs and precursors to thermal runaway appear, thereby avoiding safety risks caused by thermal runaway.
[0029] The implementation details of the technical solution of the embodiment of the present application are described in detail below: First of all, it should be noted that the solution of the embodiment of the present application is based on the CNN (Convolutional Neural Networks)-LSTM (long and short-term memory) model architecture, in which the features of the battery data are extracted from CNN to form a feature data set (formed by multiple feature sequences), and the feature data set is sent to the pooling layer for downsampling processing. Although the pooling processing of the pooling layer is relatively common in this field, if the pooling processing is performed through the pooling layer to accurately obtain the data we need, that is, how to extract high-order features on the basis of retaining the original main features to reduce the data processing required by the LSTM model, it is an important highlight of the embodiment of the present application.
[0030] According to one aspect of an embodiment of the present application, a battery thermal runaway state warning method is provided. Figure 1 This is a flow chart of a battery thermal runaway state warning method according to an embodiment of the present application. The method includes at least steps 110 to 160, which are described in detail as follows: In step 110, a characteristic data set of a battery during charging is acquired, wherein the characteristic data set includes at least one time series data associated with a battery characteristic.
[0031] Specifically, the state of charge can be represented by the letter S. As a key parameter of the battery management system, the state of charge S can reflect the state of thermal balance during battery operation. Accurate prediction of the state of charge S can reflect and warn of possible thermal runaway of the battery. Therefore, this application carries out battery status warning work based on the prediction of the state of charge of the energy storage battery.
[0032] When predicting the state of charge S of an energy storage battery, the relationship between each input variable and S must be determined. The correlation between S and various parameter variables is as follows: Figure 2 The characteristic data set proposed in the embodiment of the present application refers to a collection of various parameter variables, that is, time series data associated with battery characteristics, specifically four time series data of voltage, current, temperature and output power.
[0033] The method of obtaining a characteristic data set of a battery during the charging process includes: obtaining current sequence data, voltage sequence data, temperature sequence data and power sequence data of the battery during the charging process; performing feature extraction on the current sequence data according to the characteristic relationship between the current of the battery and the state of charge of the battery to obtain first characteristic sequence data; performing feature extraction on the voltage sequence data according to the characteristic relationship between the voltage of the battery and the state of charge of the battery to obtain second characteristic sequence data; performing feature extraction on the temperature sequence data according to the characteristic relationship between the temperature of the battery and the state of charge of the battery to obtain third characteristic sequence data; performing feature extraction on the power sequence data according to the characteristic relationship between the power of the battery and the state of charge of the battery to obtain fourth characteristic sequence data; generating the characteristic data set based on the first characteristic sequence data, the second characteristic sequence data, the third characteristic sequence data and the fourth characteristic sequence data; wherein the first characteristic sequence data, the second characteristic sequence data, the third characteristic sequence data and the fourth characteristic sequence data are respectively time series data related to current characteristics, time series data related to voltage characteristics, time series data related to temperature characteristics and time series data related to power characteristics.
[0034] from Figure 2It can be seen that in addition to the commonly used current, voltage and temperature monitoring data, system power also has a direct impact on the change trend of S. Accordingly, the prediction model input signal is divided into four dimensions, namely the voltage, current, temperature and output power of the energy storage battery system, and the model output is set to S (state of charge). The data shown in the figure is a curve chart of the changes in various parameters of the energy storage battery within 24 hours a day.
[0035] In step 120, pooling processing is performed on each of the time series data to obtain each target output data corresponding to each of the time series data.
[0036] Specifically, in the CNN (Convolutional Neural Networks)-LSTM (long and short-term memory) model architecture, the feature data set is extracted through the CNN layer, and the feature data set is pooled through the pooling layer, so that the pooled feature data set can be calculated and analyzed more quickly after being sent to the LSTM, thereby quickly predicting the state of charge S.
[0037] Therefore, the embodiments of the present application perform pooling processing on each of the time series data respectively to obtain each target output data corresponding to each of the time series data, and the target output data is the time series data after the pooling processing.
[0038] In one embodiment of the present application, pooling is performed on a single time series data, comprising the following steps: Dividing the time series data into a preset number of data regions, inputting all data of the time series data into each of the data regions, respectively, to obtain a pooled data set corresponding to each of the data regions; Average pooling processing, maximum pooling processing and minimum pooling processing are performed on each of the pooled data sets.
[0039] Specifically, the preset number of data regions can be set according to actual needs, and the preset number can be any value such as 3, 5, 10, etc., which is not limited here. For example, if a single time series data includes 1,000 data, then the 1,000 data can be divided into 10 data regions, each data region corresponds to a spatiotemporal feature, and the 100 data in each data region are pooled data sets, then the 1,000 data can be divided into data regions with 10 different spatiotemporal features, and then the 10 regions are respectively subjected to average pooling, maximum pooling, and minimum pooling.
[0040] In one embodiment of the present application, the average pooling processing is: averaging all data in the pooled data value to obtain the average value of the pooled data value, the maximum pooling processing is: finding the maximum data value among all data in the pooled data value, and the minimum pooling processing is: finding the minimum data value among all data in the pooled data value.
[0041] Specifically, in addition to the convolution layer, CNN also includes a pooling layer. The pooling layer is a downsampling operation after the convolution layer feature extraction. This application proposes a new pooling method to improve the training efficiency of the convolutional neural network, which is called maximum average minimum pooling. The proposed pooling method improves the performance of the neural network by calculating the optimal value during the training process of the convolutional neural network.
[0042] The first is average pooling. The average pooling method generates an output value by using a filter to obtain the average value of part of the data. The average pooling method smoothes the data sequence. The specific method of average pooling is to average all the data in the data area, while the maximum pooling process is to find the maximum value of the data in the current data area, and the minimum pooling process is to find the minimum value of the data in the current data area.
[0043] In one embodiment of the present application, the target output data can be obtained by the following method: If the difference between the maximum data value and the minimum data value is greater than the average value, the difference between the maximum data value and the minimum data value is used as the target output data; If the difference between the maximum data value and the minimum data value is less than or equal to the average value, half of the sum of the maximum data value and the average value is used as the target output data.
[0044] Specifically, the above average pooling, maximum pooling and minimum pooling cannot obtain the final pooling layer output value, that is, the target output data cannot be obtained, and the final target output data can only be obtained after the operation logic. The specific operation logic can be referred to Figure 4 As shown, the following is combined Figure 4 The calculation logic of the target output data is further explained: If the difference between the maximum data value and the minimum data value is greater than the average value, the difference between the maximum data value and the minimum data value is used as the target output data. If the difference between the maximum data value and the minimum data value is less than or equal to the average value, half of the sum of the maximum data value and the average value is used as the target output data. By defining the above-mentioned operation judgment logic as a new function, that is, the target output function, this function can be called in the pooling layer of the CNN architecture, and it can be easily integrated into the CNN to quickly output the target output data.
[0045] In step 130, each of the target output data is used as pooled sequence data, and the battery parameter corresponding to the first target moment is obtained from the pooled sequence data.
[0046] Specifically, the first target time may be any time selected, that is, the battery parameters corresponding to the time are obtained, namely, voltage, current, temperature and power.
[0047] In step 140, the battery parameters are input into a pre-established long short-term memory model to obtain a predicted state of charge value of the battery at a second target time.
[0048] Specifically, the pre-established long short-term memory model includes a two-layer LSTM structure, that is, the pooled sequence data output from the pooling layer needs to pass through two LSTMs before the final predicted state of charge S can be output. The structure predicted by the two-layer LSTM structure is more accurate. At the same time, the amount of pooled sequence data of the pooling layer is small, which enables the pre-established long short-term memory model to quickly and accurately output the predicted state of charge S.
[0049] In one embodiment of the present application, the pre-established long short-term memory model includes a first-layer LSTM model and a second-layer LSTM model, and the inputting of the battery parameters into the pre-established long short-term memory model to obtain the predicted value of the state of charge of the battery includes: Inputting the battery parameter into the first-layer LSTM model, performing a first logic operation on the battery parameter according to a first operation logic corresponding to the first-layer LSTM model, and obtaining a first logic output value; Inputting the first logic output value into the second-layer LSTM model, performing a second logic operation on the first logic output value according to a second operation logic corresponding to the second-layer LSTM model, to obtain a second logic output value; The second logic output value is used as the state of charge prediction value.
[0050] The first operation logic and the second operation logic can be the same operation logic, and the second target moment can be any moment after the current moment. Then the pooled sequence data output by the pooling layer is output to the long short-term memory model of the embodiment of the present application. The two-layer LSTM model included in the long short-term memory model performs the final output value through its unique operation logic. Then the first layer operates on the pooled sequence data to obtain the first logical output value, and inputs the preliminary prediction value into the second layer for operation. The second logical output value of the second layer is the charge state prediction value.
[0051] At the same time, experiments using LSTM with different numbers of layers show that the state of charge predicted by the double-layer LSTM structure is the most accurate. Further comparative experiments are set up for verification. The results show that the proposed model can accurately predict the state of charge of the energy storage battery, thereby reflecting the possible battery thermal runaway phenomenon. This algorithm can overcome the problem of easy overfitting and low accuracy of the model in existing methods.
[0052] The first operation logic or the second operation logic is the internal operation logic of the LSTM model, and the operation process is as follows: 1) Determine the value of the forget gate f t ; Where: f t is the value of the forget gate; W f is the weight matrix; h t-1 , x t ] is the output of LSTM at the previous moment h t and the current input x t The matrix composed of b f is the bias term; σ is the sigmoid function.
[0053] 2) Determine the value of the input gate i t ; Where: i t is the value of the input gate; W i is the weight matrix of the input gate; b i is the bias term of the input gate.
[0054] 3) Determine the value of the output gate ot ; Where: o t is the value of the output gate; W o is the weight matrix of the output gate; b o is the bias term of the output gate.
[0055] 4) Calculate the unit used to describe the current input
[0056] ; Where: tanh(•) is the hyperbolic tangent function; W c is the weight matrix of the unit state; b c is the bias term of the input unit.
[0057] 5) Calculate the value of the memory cell at the current moment ; 6) Calculate the final output of LSTM at the current moment ; In order to improve the prediction accuracy of the LSTM model and reduce its error, the present invention uses the error-correlated linear analysis method to make improvements and proposes a relationship as shown below to associate the input variables with the historical prediction errors.
[0058] ; in: e h Represents the LSTM prediction error; f ( x 1, ..., x n ) is the main function on the input, ( x 1, ..., x n ) indicates input.
[0059] The prediction model after error correction is shown as follows: ; in: h ( x 1, ..., x n ) is the established LSTM prediction model; f ( x 1, ..., x n ) is the error linear correction function,h A-LSTM is the final output of the LSTM algorithm.
[0060] In step 150, the highest charging voltage of the battery in each target time period during the charging process is obtained based on the characteristic data set.
[0061] The target time period can be set as needed. For example, if the current time is 15:00, the target time period can be set to the time period between 14:00 and 14:30. The closer the time period is to the current time, the more accurate the predicted state of charge. By obtaining the highest charging voltage in the target time period, it can be compared with the preset voltage threshold. If it is greater than the preset voltage threshold, it indicates that there is a risk of thermal runaway of the battery. The specific judgment can be made in combination with the current state of charge.
[0062] In step 160, the thermal runaway state level of the battery is determined according to the highest charging voltage in the target time period where the first target moment is located and the charge state prediction value, and the battery management system is controlled to perform a target warning operation based on the thermal runaway state level.
[0063] Specifically, the thermal runaway state level of the battery can be determined by the maximum charging voltage in the above-mentioned target time period and the charge state prediction value. The battery management system is controlled to perform different target warning operations according to different thermal runaway state levels, so that when the battery is at risk of thermal runaway, corresponding warning operations can be obtained to prevent the occurrence of battery thermal runaway.
[0064] In one embodiment of the present application, determining the thermal runaway state level of the battery according to the highest charging voltage in the target time period where the first target time is located and the state of charge prediction value includes: If the predicted state of charge value is lower than a first preset threshold value, and the maximum charging voltage is greater than a preset voltage threshold value, determining that the thermal runaway state level is level one; If the state of charge prediction value is greater than or equal to a first preset threshold value and less than a second preset threshold value, and the maximum charging voltage is greater than a preset voltage threshold value, determining that the thermal runaway state level is level 2; If the state of charge prediction value is greater than or equal to a second preset threshold and the maximum charging voltage is greater than a preset voltage threshold, it is determined that the thermal runaway state level is level three.
[0065] Specifically, the preset voltage threshold can be set according to actual needs, and can be set to a voltage of 120V, 240V, 300V, etc., which is not limited here.
[0066] The first preset threshold value may be specifically 60%. When the state of charge S is lower than 60% and the maximum charging voltage is greater than the preset voltage threshold, it indicates that the charging state of the battery is abnormal, and the thermal runaway state may be defined as level one.
[0067] The second preset threshold value may be specifically 80%. When the state of charge S is between 60% and 80% and the maximum charging voltage is greater than the preset voltage threshold, it indicates that there may be risks in the charging state of the battery, and the thermal runaway state may be defined as level 2.
[0068] If the state of charge S is greater than 80% and the maximum charging voltage is greater than the preset voltage threshold, it means that the battery is likely to suffer thermal runaway and cause a fire. The thermal runaway state can be defined as level three.
[0069] In one embodiment of the present application, controlling the battery management system to perform a target warning operation based on the thermal runaway state level includes: If the thermal runaway state level is level one, controlling the battery management system to send a prompt signal indicating that the battery state is abnormal to the terminal; If the thermal runaway state level is level 2, controlling the battery management system to send an alarm signal indicating that the battery state is at risk to the terminal; If the thermal runaway state level is level three, the battery management system is controlled to cut off the power supply of the battery and send an alarm signal that the battery is in a hot state to the terminal.
[0070] Specifically, the terminal can be a mobile phone, computer or other device that can receive signals. If the thermal runaway state level is level 1, a prompt signal indicating that the battery state is abnormal is sent to the terminal. Protective measures should be taken in time to make it converge to normal.
[0071] If the thermal runaway state level is level 2, an alarm signal indicating that the battery state is at risk will be sent to the terminal, and timely alarms will be issued and measures will be taken to prevent the battery from being in danger.
[0072] If the thermal runaway state level is level three, the battery management system is controlled to cut off the power supply of the battery and send an alarm signal that the battery is in a hot state to the terminal, so as to prompt the staff or users to take emergency measures quickly and take protective measures in time.
[0073] In one embodiment of the present application, the method further includes: Determine a first loss value of the first layer LSTM model after performing a first logic operation based on the first logic output value, and determine a second loss value of the second layer LSTM model after performing a second logic operation based on the second logic output value; The long short-term memory model is trained according to the first loss value and the second loss value.
[0074] Specifically, the first loss value of the first layer LSTM model after performing the first logical operation is obtained by comparing the first logic output value with the preset reference output value, and the second loss value of the second layer LSTM model after performing the second logical operation is obtained by comparing the second logic output value with the preset reference output value, wherein the preset reference output value can be a state of charge value under an ideal state, which is obtained from a large number of experiments.
[0075] The long short-term memory model is trained by using the first loss value and the second loss value, that is, the first layer LSTM and the second layer LSTM are trained respectively, so as to update the first layer LSTM and the second layer LSTM, so that the output prediction of the updated first layer LSTM and the second layer LSTM is more accurate.
[0076] In summary, the method proposed in the present application can quickly and accurately predict the state of charge of the battery, and can also perform analysis and early warning based on the predicted state of charge, thereby further avoiding the risk of thermal runaway of the battery during the charging process.
[0077] Figure 3 : This is a block diagram of a battery thermal runaway state warning device 300 according to an embodiment of the present application. According to a battery thermal runaway state warning device 300 according to an embodiment of the present application, the device 300 includes: a first acquisition unit 301, a pooling unit 302, a second acquisition unit 303, a prediction unit 304, a third acquisition unit 305, and a warning unit 306.
[0078] A first acquisition unit 301 is used to acquire a characteristic data set of the battery during the charging process, wherein the characteristic data set includes at least one time series data associated with the battery characteristics; A pooling unit 302 is used to perform pooling processing on each of the time series data to obtain each target output data corresponding to each of the time series data; A second acquisition unit 303, configured to use each of the target output data as pooled sequence data, and acquire a battery parameter corresponding to a first target moment from the pooled sequence data; A prediction unit 304, configured to input the battery parameters into a pre-established long short-term memory model to obtain a predicted value of the state of charge of the battery at a second target time; A third acquisition unit 305, configured to acquire, based on the characteristic data set, a maximum charging voltage of the battery in each target time period during a charging process; The early warning unit 306 is used to determine the thermal runaway state level of the battery according to the highest charging voltage in the target time period where the first target moment is located and the charge state prediction value, and perform a target early warning operation based on the thermal runaway state level.
[0079] As another aspect, the present application further provides a computer-readable storage medium on which a program product capable of implementing the method provided above in this specification is stored. In some possible implementations, various aspects of the present application may also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary implementations of the present application described in the above "Embodiment Method" section of this specification.
[0080] According to the program product for implementing the above method in the embodiment of the present application, it can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present application is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.
[0081] The program product may be any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.
[0082] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0083] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.
[0084] Program code for performing the operations of the present application may be written in any combination of one or more programming languages, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).
[0085] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present application, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.
[0086] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be performed without departing from the scope thereof. The scope of the present application is limited only by the appended claims.
Claims
1. A battery thermal runaway state early warning method, characterized in that: The method comprises: Acquire a characteristic data set of the battery during the charging process, wherein the characteristic data set includes at least one time series data associated with the battery characteristics; Performing pooling processing on each of the time series data respectively to obtain each target output data corresponding to each of the time series data; Using each of the target output data as pooled sequence data, and obtaining a battery parameter corresponding to a first target moment in the pooled sequence data; Inputting the battery parameters into a pre-established long short-term memory model to obtain a predicted state of charge value of the battery at a second target time; Acquire the maximum charging voltage of the battery in each target time period during the charging process based on the characteristic data set; The thermal runaway state level of the battery is determined according to the highest charging voltage in the target time period where the first target moment is located and the state of charge prediction value, and the battery management system is controlled to perform a target warning operation based on the thermal runaway state level.
2. The method according to claim 1, characterized in that Pooling is performed on a single time series data, comprising the following steps: Dividing the time series data into a preset number of data regions, inputting all data of the time series data into each of the data regions, respectively, to obtain a pooled data set corresponding to each of the data regions; Average pooling processing, maximum pooling processing and minimum pooling processing are performed on each of the pooled data sets.
3. The method according to claim 2, characterized in that The average pooling process is: averaging all the data in the pooled data value to obtain the average value of the pooled data value, the maximum pooling process is: finding the maximum value of all the data in the pooled data value, and the minimum pooling process is: finding the minimum value of all the data in the pooled data value.
4. The method according to claim 3, characterized in that The target output data can be obtained by the following method: If the difference between the maximum data value and the minimum data value is greater than the average value, the difference between the maximum data value and the minimum data value is used as the target output data; If the difference between the maximum data value and the minimum data value is less than or equal to the average value, half of the sum of the maximum data value and the average value is used as the target output data.
5. The method according to claim 1, characterized in that The pre-established long short-term memory model includes a first-layer LSTM model and a second-layer LSTM model, and the battery parameters are input into the pre-established long short-term memory model to obtain the predicted value of the state of charge of the battery, including: Inputting the battery parameter into the first-layer LSTM model, performing a first logic operation on the battery parameter according to a first operation logic corresponding to the first-layer LSTM model, and obtaining a first logic output value; Inputting the first logic output value into the second-layer LSTM model, performing a second logic operation on the first logic output value according to a second operation logic corresponding to the second-layer LSTM model, to obtain a second logic output value; The second logic output value is used as the state of charge prediction value.
6. The method according to claim 1, characterized in that The step of obtaining a characteristic data set of the battery during the charging process includes: Acquiring current sequence data, voltage sequence data, temperature sequence data and power sequence data of the battery during charging; Extracting features from the current sequence data according to a characteristic relationship between the current of the battery and the state of charge of the battery to obtain first characteristic sequence data; Extracting features from the voltage sequence data according to a characteristic relationship between the voltage of the battery and the state of charge of the battery to obtain second characteristic sequence data; Extracting features from the temperature series data according to a characteristic relationship between the temperature of the battery and the state of charge of the battery to obtain third characteristic series data; Extracting features from the power sequence data according to a characteristic relationship between the power of the battery and the state of charge of the battery to obtain fourth characteristic sequence data; Generate the feature data set based on the first feature sequence data, the second feature sequence data, the third feature sequence data and the fourth feature sequence data; Among them, the first feature sequence data, the second feature sequence data, the third feature sequence data and the fourth feature sequence data are respectively time series data related to current characteristics, time series data related to voltage characteristics, time series data related to temperature characteristics, and time series data related to power characteristics.
7. The method according to claim 5, characterized in that The method further comprises: Determine a first loss value of the first layer LSTM model after performing a first logic operation based on the first logic output value, and determine a second loss value of the second layer LSTM model after performing a second logic operation based on the second logic output value; The long short-term memory model is trained according to the first loss value and the second loss value.
8. The method according to claim 1, characterized in that The determining the thermal runaway state level of the battery according to the highest charging voltage in the target time period where the first target moment is located and the state of charge prediction value includes: If the predicted state of charge value is lower than a first preset threshold value, and the maximum charging voltage is greater than a preset voltage threshold value, determining that the thermal runaway state level is level one; If the state of charge prediction value is greater than or equal to a first preset threshold value and less than a second preset threshold value, and the maximum charging voltage is greater than a preset voltage threshold value, determining that the thermal runaway state level is level 2; If the state of charge prediction value is greater than or equal to a second preset threshold and the maximum charging voltage is greater than a preset voltage threshold, it is determined that the thermal runaway state level is level three.
9. The method according to claim 8, characterized in that The controlling the battery management system to perform a target warning operation based on the thermal runaway state level includes: If the thermal runaway state level is level one, controlling the battery management system to send a prompt signal indicating that the battery state is abnormal to the terminal; If the thermal runaway state level is level 2, controlling the battery management system to send an alarm signal indicating that the battery state is at risk to the terminal; If the thermal runaway state level is level three, the battery management system is controlled to cut off the power supply of the battery and send an alarm signal that the battery is in a hot state to the terminal.
10. A battery thermal runaway state warning device, characterized in that: The device comprises: A first acquisition unit, configured to acquire a characteristic data set of the battery during the charging process, wherein the characteristic data set includes at least one time series data associated with a battery characteristic; A pooling unit, used to perform pooling processing on each of the time series data respectively to obtain each target output data corresponding to each of the time series data; a second acquisition unit, configured to use each of the target output data as pooled sequence data, and acquire a battery parameter corresponding to a first target moment from the pooled sequence data; A prediction unit, used for inputting the battery parameters into a pre-established long short-term memory model to obtain a predicted value of the state of charge of the battery at a second target time; A third acquisition unit, configured to acquire, based on the characteristic data set, a maximum charging voltage of the battery in each target time period during a charging process; An early warning unit is used to determine the thermal runaway state level of the battery according to the highest charging voltage in the target time period where the first target moment is located and the charge state prediction value, and perform a target early warning operation based on the thermal runaway state level.