Overcharge internal pressure diagnostic methods, diagnostic systems, and battery management systems for hard-cased batteries

Abstract

This application belongs to the field of battery safety technology, and specifically discloses a hard-shell battery overcharge internal pressure diagnosis method, diagnosis system and battery management system. Through this application, by detecting the non-electrical quantity on the surface of the battery explosion-proof valve - strain parameters, approximate physical models and PINN models of strain parameters, measured values ​​of battery surface temperature and state of charge SOC are constructed respectively, and each physical parameter in the approximate physical model is used as a parameter group to be trained, and the parameters are fitted through loss function and gradient descent method, thereby improving the diagnostic accuracy. In addition, the constructed total loss function comprehensively considers the deviations between the physical information of the PINN model predicted data, measured data, and approximate physical model, and combines parameter identification and model training to dynamically combine the parameter identification process of physical information and the learning process of the neural network to improve the model learning efficiency.

Description

Technical Field

[0001] The present application belongs to the field of battery safety technology, and more specifically, to a hard-shell battery overcharge internal pressure diagnosis method, a diagnosis system, and a battery management system. Background Art

[0002] The rapid development of electrochemical energy storage has directly promoted the widespread application of lithium batteries. Traditional lithium battery overcharge safety warning solutions only diagnose battery abnormalities through electrical quantities such as voltage and state of charge (SOC). However, the complex electrical environment in the energy storage system often leads to inaccurate detection of electrical quantities such as battery voltage and SOC, which makes it impossible to fundamentally solve the safety problems of energy storage batteries. Such problems may also occur in other types of hard-shell batteries.

[0003] Research has shown that many non-electrical quantities can diagnose abnormal conditions inside batteries earlier, such as stress and strain, temperature, and internal pressure. However, in existing energy storage engineering applications, there are few collection methods and early warning measures for non-electrical quantities. Battery diagnosis and maintenance are mostly based on large-scale shutdowns for maintenance and fire extinguishing as safety measures, which directly affects the operating efficiency of energy storage power stations. At the same time, current early warning strategies mostly monitor anomalies through fixed thresholds. The disadvantage of this method is that it cannot adapt to the complex operating environment of energy storage batteries. Different thresholds need to be set for different environments. The workload is often huge and the effect is poor, which seriously affects the operating efficiency of the energy storage battery system. Therefore, it is very important to diagnose internal abnormalities of batteries by adding new perception parameters and diagnostic algorithms, which is also an important research direction in the field of battery safety.

[0004] In response to this problem, patent CN117148186A proposed to detect the non-electrical quantity on the surface of the battery explosion-proof valve - the strain of the strain gauge, and theoretically derive the physical model of the lithium battery explosion-proof valve strain and the battery surface temperature and battery charge state, and verified the effectiveness of the model through experiments. However, the model constructed in this patent is very rough and has low accuracy, and no clear early warning method is given. Summary of the invention

[0005] In view of the defects of the prior art, the purpose of this application is to provide a hard shell battery overcharge internal pressure diagnosis method, diagnosis system and battery management system, aiming to solve the problem of low accuracy and inability to clearly warn in existing non-embedded internal pressure diagnosis methods.

[0006] To achieve the above objectives, in a first aspect, the present application provides a method for diagnosing overcharge internal pressure of a hard-shell battery, wherein the hard-shell battery has an explosion-proof valve, and the diagnostic method comprises:

[0007] Obtain the measured value of the surface strain parameter of the explosion-proof valve of the battery to be tested, the measured value of the battery surface temperature and the state of charge SOC;

[0008] The measured value of the surface temperature of the battery to be tested and the SOC are input into the trained PINN model to obtain the predicted value of the surface strain parameter of the explosion-proof valve of the battery to be tested;

[0009] Calculate the difference between the predicted value of the strain parameter on the surface of the explosion-proof valve of the battery to be tested and the measured value of the strain parameter, and determine whether the difference exceeds the diagnostic threshold, and diagnose it as abnormal internal pressure; otherwise, diagnose it as normal internal pressure;

[0010] The PINN model is trained in the following way:

[0011] Taking (battery surface temperature, SOC) as training samples and the measured values ​​of the surface strain parameters of the battery explosion-proof valve as labels, the PINN model is input for training so that it can simultaneously learn the physical information of the approximate physical model of the battery explosion-proof valve strain parameters and the battery surface temperature and SOC and the data characteristics of the measured data. The training loss function is composed of a first loss, a second loss and a third loss. The first loss represents the loss value between the measured data and the physical information, which is used to measure the deviation between the approximate physical model and the measured data; the second loss represents the loss value between the predicted value of the PINN model and the physical information, which is used to measure the deviation between the approximate physical model and the predicted data of the PINN model; the third loss represents the loss value between the predicted data of the PINN model and the measured data, which is used to measure the deviation between the measured data and the predicted data of the PINN model.

[0012] Preferably, the surface strain parameter of the explosion-proof valve is the resistance, strain or potential difference of a strain gauge, and the strain gauge is closely attached to the outer surface of the explosion-proof valve.

[0013] Preferably, the diagnostic threshold is determined by:

[0014] The interquartile range (IQR) is determined by using the box plot method through the residual sequence of the first several points in the training sample;

[0015] 3 times IQR was used as the diagnostic threshold;

[0016] The residual sequence is the difference between the predicted value and the corresponding measured value obtained by inputting the battery SOC and the surface temperature into the trained PINN model respectively.

[0017] Preferably, in the training phase, each physical parameter in the approximate physical model is used as a parameter group to be trained, and parameter fitting is performed through a loss function and a gradient descent method.

[0018] Preferably, all training samples are from various charging mode operating condition tests of the same type of battery with normal internal pressure.

[0019] Preferably, the charging mode operating conditions include constant current and constant voltage charging and pulse current charging.

[0020] Preferably, the diagnostic method is applied to detect abnormal internal pressure of the battery at the initial stage of battery overcharging, so as to achieve early fault diagnosis.

[0021] Preferably, the method further comprises: when the internal pressure is diagnosed as abnormal, sending a signal to a host computer so that it immediately cuts off power and removes the faulty battery.

[0022] To achieve the above-mentioned objectives, in a second aspect, the present application provides a hard shell battery overcharge internal pressure diagnostic system, comprising: at least one memory for storing programs; at least one processor for executing the programs stored in the memory, when the programs stored in the memory are executed, the processor is used to execute the diagnostic method described in the first aspect.

[0023] To achieve the above-mentioned purpose, in the third aspect, the present application provides a hard shell battery management system, including: a measurement circuit for real-time acquisition of the measured values ​​of the surface strain parameters of the explosion-proof valve, the measured values ​​of the battery surface temperature and the state of charge SOC; and a hard shell battery overcharge internal pressure diagnostic system as described in the second aspect.

[0024] It can be understood that the beneficial effects of the second to third aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0025] In general, the above technical solutions conceived by this application have the following beneficial effects compared with the prior art:

[0026] The present application provides a hard-shell battery overcharge internal pressure diagnosis method, diagnosis system and battery management system, and proposes to construct an approximate physical model and PINN model of strain parameter-battery surface temperature measured value and state of charge SOC respectively by detecting the non-electrical quantity on the surface of the battery explosion-proof valve - strain parameter, and use each physical parameter in the approximate physical model as a parameter group to be trained, and perform parameter fitting through loss function and gradient descent method to improve the diagnostic accuracy. In addition, the constructed total loss function integrates the deviations between the physical information of the PINN model prediction data, measured data, and approximate physical model, and combines parameter identification and model training to dynamically combine the parameter identification process of physical information and the learning process of the neural network to improve the model learning efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A flow chart of a hard-shell battery overcharge internal pressure diagnosis method provided in this application.

[0028] Figure 2 This is a diagram of the strain mechanism model architecture of a lithium battery explosion-proof valve based on a physical information fusion neural network (PINN) provided in an embodiment of the present application.

[0029] Figure 3The predicted value and actual value fitting results and residual graph of the strain training set of lithium battery explosion-proof valve based on physical information fusion neural network (PINN) provided in the embodiment of the present application.

[0030] Figure 4 This is a residual distribution histogram of the physical information fusion-based neural network (PINN) training set provided in an embodiment of the present application.

[0031] Figure 5 A flow chart of the abnormal battery internal pressure diagnosis phase is provided for the embodiment of the present application.

[0032] Figure 6 The fitting results and residual diagram of the predicted values ​​and actual values ​​of the validation set under normal internal pressure conditions of the physical information fusion-based neural network (PINN) provided in the embodiment of the present application.

[0033] Figure 7 The residual sequence diagram of the predicted value and actual value of the explosion-proof valve strain gauge resistance of the lithium battery provided in the embodiment of the present application under the 0.5C constant current charging condition, as well as the upper and lower warning limits.

[0034] Figure 8 The lithium battery provided in the embodiment of the present application is provided with the fitting results and residual diagram of the predicted value and actual value of the validation set under the condition of abnormal internal pressure caused by overcharging.

[0035] Fig. 9 The embodiment of the present application provides a residual sequence of the predicted value and actual value of the resistance of the explosion-proof valve strain gauge under the condition of abnormal internal pressure caused by overcharging of the lithium battery, as well as a warning curve diagram and upper and lower limits of the warning.

[0036] Fig.10 A schematic diagram of the structure of an electronic device provided in this application. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0038] The term "and / or" in this article is a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The symbol " / " in this article indicates that the associated objects are in an or relationship, for example, A / B means A or B.

[0039] The terms "first" and "second" in the specification and claims herein are used to distinguish different objects rather than to describe a specific order of the objects. For example, a first response message and a second response message are used to distinguish different response messages rather than to describe a specific order of the response messages.

[0040] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0041] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc.; multiple elements refer to two or more elements, etc.

[0042] Next, the technical solutions provided in the embodiments of the present application are introduced.

[0043] The present application provides a method for diagnosing overcharge internal pressure of a hard-shell battery, wherein the hard-shell battery has an explosion-proof valve, and the diagnostic method comprises:

[0044] Obtain the measured value of the surface strain parameter of the explosion-proof valve of the battery to be tested, the measured value of the battery surface temperature and the state of charge SOC;

[0045] The measured value of the surface temperature of the battery to be tested and the SOC are input into the trained PINN model to obtain the predicted value of the surface strain parameter of the explosion-proof valve of the battery to be tested;

[0046] Calculate the difference between the predicted value of the strain parameter on the surface of the explosion-proof valve of the battery to be tested and the measured value of the strain parameter, and determine whether the difference exceeds the diagnostic threshold, and diagnose it as abnormal internal pressure; otherwise, diagnose it as normal internal pressure;

[0047] The PINN model is trained in the following way:

[0048] Taking (battery surface temperature, SOC) as training samples and the measured values ​​of the surface strain parameters of the battery explosion-proof valve as labels, the PINN model is input for training so that it can simultaneously learn the physical information of the approximate physical model of the battery explosion-proof valve strain parameters and the battery surface temperature and SOC and the data characteristics of the measured data. The training loss function is composed of a first loss, a second loss and a third loss. The first loss represents the loss value between the measured data and the physical information, which is used to measure the deviation between the approximate physical model and the measured data; the second loss represents the loss value between the predicted value of the PINN model and the physical information, which is used to measure the deviation between the approximate physical model and the predicted data of the PINN model; the third loss represents the loss value between the predicted data of the PINN model and the measured data, which is used to measure the deviation between the measured data and the predicted data of the PINN model.

[0049] Preferably, the surface strain parameter of the explosion-proof valve is the resistance, strain or potential difference of a strain gauge, and the strain gauge is closely attached to the outer surface of the explosion-proof valve.

[0050] The battery explosion-proof valve is a safety device located on the surface of the battery. It serves as a vent when the internal pressure of the battery is abnormal. When the pressure is abnormal, the explosion-proof valve bulges outward until it opens. The explosion-proof valve is usually composed of a valve body, a diaphragm, a spring, a valve seat and other components. The strain gauge is attached to the surface of the diaphragm using a room temperature hardening instant adhesive. The strain gauge should be a unidirectional strain gauge with a smaller surface area than the explosion-proof valve diaphragm. The Wheatstone bridge conditioning circuit and high-precision analog-to-digital conversion chip can be used to convert the resistance of the strain gauge into a digital signal that can be recognized and calculated by the computer. By monitoring the changes in the digital signal, the internal pressure of the battery can be monitored. The internal pressure of the battery changes due to the deintercalation of lithium and heat generation of the electrode material. Therefore, through experiments under different charging modes (such as constant current and constant voltage charging, pulse current charging, etc.), and by using a charge and discharge tester and a data acquisition system, the battery state of charge (SOC) and surface temperature are obtained. Finally, the strain on the surface of the explosion-proof valve and the approximate physical model of the battery surface temperature and SOC can be obtained.

[0051] Preferably, the diagnostic threshold is determined by:

[0052] The interquartile range (IQR) is determined by using the box plot method through the residual sequence of the first several points in the training sample;

[0053] 3 times IQR was used as the diagnostic threshold;

[0054] The residual sequence is the difference between the predicted value and the corresponding measured value obtained by inputting the battery SOC and the surface temperature into the trained PINN model respectively.

[0055] Preferably, in the training phase, each physical parameter in the approximate physical model is used as a parameter group to be trained, and parameter fitting is performed through a loss function and a gradient descent method.

[0056] Preferably, all training samples are from various charging mode operating condition tests of the same type of battery with normal internal pressure.

[0057] Preferably, the charging mode operating conditions include constant current and constant voltage charging and pulse current charging.

[0058] Preferably, the diagnostic method is applied to detect abnormal internal pressure of the battery at the initial stage of battery overcharging, so as to achieve early fault diagnosis.

[0059] Preferably, the method further comprises: when the internal pressure is diagnosed as abnormal, sending a signal to a host computer so that it immediately cuts off power and removes the faulty battery.

[0060] Correspondingly, the present application provides a hard shell battery overcharge internal pressure diagnosis system, including: at least one memory for storing programs; at least one processor for executing the programs stored in the memory, when the program stored in the memory is executed, the processor is used to execute the above-mentioned diagnostic method.

[0061] On this basis, the present application provides a hard shell battery management system, including: a measurement circuit for real-time collection of the measured values ​​of the surface strain parameters of the explosion-proof valve, the measured values ​​of the battery surface temperature and the state of charge SOC; and the above-mentioned hard shell battery overcharge internal pressure diagnostic system.

[0062] Example

[0063] This embodiment takes a square lithium battery as an example to illustrate the entire internal pressure diagnosis method process.

[0064] Lithium batteries are accompanied by temperature changes during use, which causes thermal stress. Hard-shell lithium batteries are layered winding structures. Due to the different thermal expansion coefficients between the winding layers, the thermal expansion of each layer of material will restrict each other and also produce thermal stress. Thermal stress is reflected in the battery cell as the change in internal pressure of the battery. At the same time, due to the gaps inside the square lithium battery, the gas in the gaps expands and contracts due to the temperature, and the extrusion caused by the expansion of the battery cell will cause the internal gas pressure of the battery to change regularly and cyclically.

[0065] Step S1. construct an approximate physical model of the resistance value of the strain gauge on the surface of the lithium battery explosion-proof valve and the battery temperature and state of charge (SOC) under the charging cycle state when the battery internal pressure is normal.

[0066] As a discharge port when the internal pressure of the battery is abnormal, the battery explosion-proof valve is much more sensitive to internal pressure than other parts of the battery surface. The change in resistance of the surface strain gauge can be used as a representation of the change in internal pressure of the battery. The internal pressure of the battery changes due to the deintercalation and heat generation of the electrode material, so through experiments under different working conditions, an approximate physical model of the resistance of the surface strain gauge of the explosion-proof valve and the battery temperature and SOC can be obtained.

[0067] The expansion and contraction of the battery's internal winding core caused by internal stress changes the volume of the internal voids of the battery, recorded as ΔV. Assuming that the change in electrode thickness caused by charging and discharging is Δh pole Assuming that the internal temperature of the battery remains unchanged, the expansion of the lithium battery is unconstrained, and the separator and current collector are not compressed and deformed, the electrode thickness changes by Δh pole It is linearly related to the change of SOC:

[0068]

[0069] Among them, h pole is the initial thickness of the electrode, ω - is the volume fraction of negative electrode material, ω + is the volume fraction of the positive electrode material, α - is the volume expansion coefficient of the negative electrode material, α + is the volume expansion coefficient of the positive electrode material, β is the excess coefficient of the negative electrode, both are constants. Assuming the initial SOC is 0, then ΔSOC = SOC.

[0070] Similarly, assuming that SOC remains unchanged, we have:

[0071]

[0072] In the formula, Δh temp is the change in electrode thickness during the static stage, ΔT is the temperature change of the battery surface collected, and it is approximately assumed that the entire battery is at the same temperature, so ΔT can represent the overall temperature change of the battery, and α is the thermal expansion coefficient, which is approximately considered to be a constant. Combining the above formula, we can get:

[0073]

[0074] Assuming that the volume change of the internal gap of the battery is ΔV, we have:

[0075] ΔV=Δh pole S pole

[0076] V′=V-ΔV

[0077] Where V′ is the changed internal gap of the battery, V is the original internal gap of the battery, and S pole is the surface area of ​​the electrode after expansion, which remains unchanged when the internal stress changes.

[0078] The explosion-proof valve pressure is the internal stress, which is generated by the internal gas of the battery acting on the explosion-proof valve. According to the ideal gas state equation:

[0079] P in V=nRT

[0080] as well as

[0081] F in =P in S valve

[0082]

[0083] and generalized Hooke's law

[0084]

[0085] Among them, P in is the internal pressure of the battery; n is the amount of gas; R is the molar gas constant. When the battery is in a safe state, it is approximately assumed that no obvious gas is generated, so the above three values ​​are all constants; σ valve F is the stress on the battery explosion-proof valve, and its direction is perpendicular to the surface of the explosion-proof valve; in F is the pressure inside the battery acting on the explosion-proof valve; out The pressure applied to the explosion-proof valve from the outside of the battery, usually F out =P0S valve , P0 is the standard atmospheric pressure; S valve is the explosion-proof valve area; E is the elastic modulus; μ is the Poisson's ratio; ε1 is the strain in the same direction as the metal strain gauge, σ1 is the stress in the same direction as the metal strain gauge; σ2 is the stress perpendicular to the surface of the explosion-proof valve, and it can be considered that σ2=σ valve ; σ3 is the stress in the direction perpendicular to both σ1 and σ2; in the safe state of the battery, it is approximately assumed that the explosion-proof valve has only one principal strain in one direction, that is, ε2=ε3=0.

[0086] Integrating the above formulas, we get:

[0087]

[0088] And because the change in resistance of the metal strain gauge and its stress and strain have the following relationship:

[0089]

[0090] And σ2=σ valve , simplify the above formula, specify the strain direction, and get:

[0091]

[0092]

[0093] Among them, K1, K2, K3, K4, K5, and K6 are all approximately constant coefficients, thereby obtaining an approximate physical model of the resistance of the battery explosion-proof valve strain gauge and the battery temperature and SOC. This approximate model reflects the variation law of the resistance of the strain gauge on the surface of the explosion-proof valve under normal battery internal pressure.

[0094] Step S2. Based on a physical information fusion neural network (PINN), an accurate model of the resistance value of the metal strain gauge on the surface of the explosion-proof valve and the battery temperature and charge state under normal internal pressure of the lithium battery is constructed.

[0095] Since the approximate physical model in S1 linearizes a large number of relationships, it will cause large residuals in the prediction of the approximate model. To solve the above problem, a neural network based on physical information fusion (PINN) is used to fuse the approximate physical model proposed in S1 with the measured data.

[0096] The experimental data comes from lithium battery experiments under specific working conditions. Two sets of different experimental conditions are designed for neural network model training.

[0097] The first group of working conditions: at room temperature of 25°C, the battery was subjected to intermittent constant current charging at 0.5C, charging for 0.4h, standing for 1h, for a total of 5 cycles, charging the battery from a state of charge of 0% to a state of charge of 100%.

[0098] The second group of working conditions: at room temperature of 25°C, the battery is charged with a 0.5C uninterrupted constant current charge from a state of charge of 0% to a state of charge of 100%.

[0099] After obtaining the experimental data, the data needs to be filtered to reduce the impact of noise interference during the acquisition process on subsequent model training. Median filtering is used here, which replaces the value with the median within a data point window to remove data noise. The window size selected here is 5.

[0100] After obtaining the filtered data and the approximate model, it is necessary to construct PINN so that it can simultaneously learn the physical characteristics of the approximate model and the data characteristics of the experimental data. At the same time, the gradient descent principle of the neural network can be used to fit the parameters of K1, K2, K3, K4, K5, and K6 in the approximate model. The specific implementation process is: first, customize K1~K6 as variable parameters, and use the customized loss function loss1 to identify a set of K1~K6 parameters through the gradient descent method in the neural network so that it can fit the experimental data to the maximum extent. Then, the data characteristics of the experimental data and the physical characteristics of the physical model are learned through the customized loss functions loss2 and loss3.

[0101] To achieve the above functions, a custom neural network architecture is required, such as Figure 2 As shown. Define the network architecture as a fully connected neural network with two input features and one output feature, a total of 3 hidden layers, 32 neurons in each hidden layer, and use an exponentially variable learning rate to more accurately learn data and physical features. The custom loss function is as follows:

[0102]

[0103] Among them, K1~K6 are physical parameters that need to be solved by neural network, R(ΔT,SOC) is the strain gauge resistance value predicted by neural network; loss1 represents the loss value between the actual value and physical information, which is used to measure the deviation between the approximate physical equation and the measured data; loss2 represents the loss value between the predicted value of the neural network and the physical information, which is used to measure the deviation between the approximate physical equation and the predicted data of the neural network; loss3 represents the loss value between the predicted data of the neural network and the measured data, which is used to measure the deviation between the measured data and the predicted data of the neural network; the three are added together to obtain the loss function loss for neural network training, and finally a set of fitted constant coefficients K1~K6 and the PINN model for prediction are obtained. The predicted data and actual data in the test set data are as follows Figure 3 As shown, it has a high degree of fit.

[0104] The obtained PINN model has stronger generalization ability than the ordinary fully connected neural network (FCNN) and better accuracy than the approximate physical model.

[0105] This application adopts the above method to construct the total loss function, combines parameter identification and model training, and dynamically combines the parameter identification process of physical information and the learning process of the neural network, which can improve the efficiency of model learning.

[0106] Step S3. Based on the accurate resistance value model and residual distribution of the strain gauge of the lithium battery explosion-proof valve, abnormal internal pressure of the lithium battery is diagnosed.

[0107] The diagnostic method is as follows Figure 4 As shown, this method uses the PINN model constructed and trained in S2, and uses the residual between the predicted value and the actual value in the training process to determine its distribution function as shown in Figure 5 As shown in the figure, through the residual distribution histogram, it can be seen that the residual distribution does not conform to the normal distribution, so the 3σ principle in statistical principles cannot be used to judge outliers. Therefore, the box plot method is used to identify outliers by using quartiles (Q1, Q2, Q3) and interquartile range (InterquartileRange, IQR), which can better adapt to data of various distributions.

[0108] Quartiles divide the data set into four equal parts, each containing 25% of the data points. There are three main quartiles:

[0109] First quartile (Q1): 25% of the data in the data set are less than or equal to this value.

[0110] Second quartile (Q2): Also known as the median, 50% of the data in the data set is less than or equal to this value.

[0111] Third quartile (Q3): 75% of the data in the data set is less than or equal to this value.

[0112] The interquartile range is the difference between Q3 and Q1, that is, IQR = Q3-Q1. It is an important indicator to measure the degree of dispersion of the median value in a data set. By excluding the highest and lowest 25% of data points, IQR focuses on the middle 50% of the data set, so it is insensitive to the existence of extreme values. Therefore, 3IQR is used as the threshold to diagnose severe outliers and reduce the false alarm rate. This threshold is not affected by the initial value and battery consistency differences and has the characteristics of self-adaptation. When the residual threshold is exceeded, it can be considered that the internal pressure of the battery deviates from the normal value, and a large amount of gas may be produced inside. The power should be cut off immediately to remove the faulty battery cell, thereby realizing the rapid online diagnosis of abnormal internal pressure of the lithium battery.

[0113] It should be noted that during the training phase, a certain type of battery is used, and the training conditions need to include constant current and constant voltage charging and pulse current charging. It is not affected by battery differences, which means that it is applicable to all batteries of the same type and the generalization ability is applicable. For other types of batteries with the same structure (hard shell batteries with explosion-proof valves), you only need to train a new accurate model and continue to use the method in S3.

[0114] In the example of normal internal pressure, the data of constant current charging under 0.2C condition is used as the verification set data, and the predicted data and actual data in the verification set are as follows: Figure 6 shown.

[0115] The theoretical resistance value of the strain gauge on the surface of the explosion-proof valve under normal internal pressure of the lithium battery is calculated based on the above precise model. pred Compared with the actual resistance value R of the strain gauge measured real The residual sequence is obtained by subtraction, and then the threshold is obtained by calculating the first 500 points of the data set according to the box plot method. The warning threshold and residual sequence are as follows: Figure 7 As shown in the figure, it can be seen that under normal internal pressure, the residual data points are strictly within the diagnostic upper and lower thresholds. It should be noted that if the number of selected points is less than 500, the diagnostic accuracy is low; if it is more than 500, the calculation time is too long.

[0116] The diagnostic threshold obtained by this method is not affected by the initial value of the strain gauge resistance and the battery difference, and has the characteristics of self-adaptation and strong generalization ability. When the residual threshold is exceeded, it can be considered that the internal pressure of the battery deviates from the normal value, and a large amount of gas may be produced inside. The power should be turned off immediately to remove the faulty battery, thereby realizing the rapid online diagnosis of abnormal internal pressure of lithium batteries. In the operation and maintenance stage, the faulty battery can be replaced to provide more targeted operation and maintenance services.

[0117] In the process of battery overcharging causing abnormal internal pressure, the predicted data and actual data are as follows Figure 8 As shown, the corresponding working condition is divided into two stages:

[0118] Stage 1: 0.5C constant current and constant voltage charging to SOC = 100%;

[0119] Stage 2: Overcharge at a constant current of 0.5C until the battery voltage reaches 5V and then stop charging. At this point, the battery is slightly swollen, the battery SOC is 105%, and the internal pressure is abnormal.

[0120] Use the method in S3 to determine the upper and lower limits of the warning, and draw the residual points and the upper and lower limits of the warning as shown in the figure Fig. 9 Before the battery is charged to SOC = 100%, all points are within the upper and lower warning limits; about 3 minutes after the battery is overcharged (the SOC is about 102%), a large amount of gas is produced inside the battery, and the internal pressure increases, resulting in a large deviation between the theoretical model prediction results and the actual measurement results, resulting in a large number of abnormal values ​​in the residual, exceeding the warning limit.

[0121] Based on the method in the above embodiment, the embodiment of the present application provides an electronic device, such as Fig.10 As shown, the electronic device may include: a processor, a communications interface, a memory, and a communication bus, wherein the processor, the communications interface, and the memory communicate with each other via the communication bus. The processor may call the logic instructions in the memory to execute the method in the above embodiment.

[0122] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application.

[0123] Based on the method in the above embodiment, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method in the above embodiment.

[0124] Based on the method in the above embodiment, an embodiment of the present application provides a computer program product. When the computer program product runs on a processor, the processor executes the method in the above embodiment.

[0125] It is understandable that the processor in the embodiment of the present application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The general-purpose processor can be a microprocessor or any conventional processor.

[0126] The method steps in the embodiments of the present application can be implemented by hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC.

[0127] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions may be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state disk (SSD)), etc.

[0128] It should be understood that the various numerical numbers involved in the embodiments of the present application are only used for the convenience of description and are not used to limit the scope of the embodiments of the present application.

[0129] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for diagnosing overcharged internal pressure of a hard-shell battery, characterized in that: The hard shell battery has an explosion-proof valve, and the diagnostic method includes: Obtain the measured value of the surface strain parameter of the explosion-proof valve of the battery to be tested, the measured value of the battery surface temperature and the state of charge SOC; The measured value of the surface temperature of the battery to be tested and the SOC are input into the trained PINN model to obtain the predicted value of the surface strain parameter of the explosion-proof valve of the battery to be tested; Calculate the difference between the predicted value of the strain parameter on the surface of the explosion-proof valve of the battery to be tested and the measured value of the strain parameter, and determine whether the difference exceeds the diagnostic threshold, and diagnose it as abnormal internal pressure; otherwise, diagnose it as normal internal pressure; The PINN model is trained in the following way: Taking (battery surface temperature, SOC) as training samples and the measured values ​​of the surface strain parameters of the battery explosion-proof valve as labels, the PINN model is input for training so that it can simultaneously learn the physical information of the approximate physical model of the battery explosion-proof valve strain parameters and the battery surface temperature and SOC and the data characteristics of the measured data. The training loss function is composed of a first loss, a second loss and a third loss. The first loss represents the loss value between the measured data and the physical information, which is used to measure the deviation between the approximate physical model and the measured data; the second loss represents the loss value between the predicted value of the PINN model and the physical information, which is used to measure the deviation between the approximate physical model and the predicted data of the PINN model; the third loss represents the loss value between the predicted data of the PINN model and the measured data, which is used to measure the deviation between the measured data and the predicted data of the PINN model.

2. The diagnostic method according to claim 1, characterized in that The surface strain parameter of the explosion-proof valve is the resistance, strain or potential difference of the strain gauge, and the strain gauge is closely attached to the outer surface of the explosion-proof valve.

3. The diagnostic method according to claim 1, characterized in that The diagnostic threshold is determined by: The interquartile range (IQR) is determined by using the box plot method through the residual sequence of the first several points in the training sample; 3 times IQR was used as the diagnostic threshold; The residual sequence is the difference between the predicted value and the corresponding measured value obtained by inputting the battery SOC and the surface temperature into the trained PINN model respectively.

4. The diagnostic method according to claim 1, characterized in that In the training phase, the physical parameters in the approximate physical model are used as the parameter group to be trained, and the parameters are fitted through the loss function and gradient descent method.

5. The diagnostic method according to claim 1, characterized in that All training samples come from various charging mode operating condition tests with normal internal pressure of the same model battery.

6. The diagnostic method according to claim 5, characterized in that The charging mode operating conditions include constant current and constant voltage charging and pulse current charging.

7. The diagnostic method according to claim 1, characterized in that The diagnostic method is applied to detect abnormal internal pressure of a battery at the initial stage of battery overcharging, so as to realize early fault diagnosis.

8. The diagnostic method according to any one of claims 1 to 7, characterized in that: Also includes: When the internal pressure is diagnosed as abnormal, a signal is sent to the host computer to immediately cut off the power and remove the faulty battery.

9. A hard case battery overcharge internal pressure diagnosis system, characterized in that: include: at least one memory for storing a program; At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the diagnostic method according to any one of claims 1 to 8.

10. A hard shell battery management system, characterized in that: include: The measurement circuit is used to collect the measured values ​​of the surface strain parameters of the explosion-proof valve, the measured values ​​of the battery surface temperature and the state of charge SOC in real time. ； The hard case battery overcharge internal pressure diagnostic system as claimed in claim 9.

Images