Internal resistance prediction model training and fault prediction method, device and equipment

Through the internal resistance prediction model trained by support vector regression and gray wolf optimization algorithm, the problem of inaccurate and damage of battery internal resistance measurement in online systems is solved, and the internal resistance is accurately predicted without damaging the battery, ensuring the safe operation of the system .

CN119064784APending Publication Date: 2024-12-03EMERSON NETWORK POWER CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310645261.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-06-01
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The prior art is prone to interference when measuring the internal resistance of lead-acid batteries in an online system, and the DC discharge method causes damage to the battery, and it is impossible to accurately predict its internal resistance without damaging the battery, affecting the safe operation of the system.

Method used

Support vector regression (SVR) algorithm combined with gray wolf optimization algorithm is used to predict battery internal resistance through training models, and parameter adjustment is used using historical data, battery pack data and environmental data to generate an internal resistance prediction model that does not damage the battery.

Benefits of technology

It realizes accurate prediction of battery internal resistance without damaging the battery, timely detection of faults, ensure safe operation of the system, and avoids economic losses caused by battery failure.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119064784A_ABST
    Figure CN119064784A_ABST
Patent Text Reader

Abstract

The invention provides an internal resistance prediction model training and fault prediction method, device and equipment, and the internal resistance prediction model training method comprises the steps: obtaining a training set comprising a plurality of samples, wherein the samples comprise historically measured battery monomer data, data of a battery pack where a battery monomer is located, environment data, and labeled battery monomer internal resistance; inputting the battery monomer data, the data of the battery pack where the battery monomer is located and the environmental data into a to-be-trained internal resistance prediction model, and performing parameter adjustment on the current internal resistance prediction model by taking the output of the labeled internal resistance of the battery monomer as a target; when a training ending condition is met, obtaining an internal resistance prediction model for predicting the internal resistance of the battery; according to the internal resistance prediction model training and fault prediction method, device and equipment provided by the invention, the internal resistance of the battery can be predicted online under the condition of not damaging the battery, and whether the battery has a fault can be detected according to the predicted internal resistance, so that the safe operation of the system is ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of lead-acid batteries, and in particular to an internal resistance prediction model training and fault prediction method, device, and equipment. Background Art

[0002] Valve-regulated lead-acid batteries are widely used in transportation, communications, electricity, computers, scientific research and other fields due to their maintenance-free, low pollution, high power density and other advantages. The health status of lead-acid batteries as backup power sources plays an important role in the safe and stable operation of the system. Therefore, accurate prediction of battery internal resistance to remind maintenance personnel to deal with failed batteries in time is a prerequisite for ensuring the safe and stable operation of the system. If the lead-acid battery fails and cannot supply power to the operating equipment in time, it will cause system downtime and lead to huge economic losses.

[0003] In the computer room of a data center, lead-acid batteries are generally grouped together, and the power supply capacity of the battery group is closely related to the available capacity of the outdated battery. Therefore, it is very important to detect and deal with the failed battery in time. According to relevant information, internal resistance is an important parameter to characterize whether the battery is failed. According to the provisions of the IEEE1188 standard, when the internal resistance of the battery is 1.3 times its initial value, the capacity of the battery may be less than 80%. There are currently many ways to measure internal resistance. Among them, the AC discharge method is to use an AC source to inject tens of milliamperes of current into the battery under test (generally less than 1Khz), divide the AC into dozens of points, use phase-locked technology, and then measure the voltage dynamic response of the corresponding point on the battery pole, and then calculate the internal resistance of the battery according to Ohm's law. However, the online system is not applicable and is easily disturbed, making the measurement result inaccurate; most data centers use DC discharge, which requires real-time large current instantaneous discharge of the battery group, but this method has certain damage to the battery. Based on the above problems, there is an urgent need for a method that can predict the internal resistance of the battery without damaging the battery to ensure the safe operation of the system. Summary of the invention

[0004] The present application provides an internal resistance prediction model training and fault prediction method, device, and equipment, which can predict the internal resistance of a battery online without damaging the battery. Based on the predicted internal resistance, it is possible to detect whether the battery has a fault, thereby ensuring the safe operation of the system.

[0005] In a first aspect, the present application provides a method for training an internal resistance prediction model, comprising:

[0006] Acquire a training set including a plurality of samples, wherein the samples include historically measured battery cell data, battery pack data where the battery cell is located, environmental data, and annotated battery cell internal resistance;

[0007] Inputting the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model to be trained, and adjusting the parameters of the current internal resistance prediction model with the goal of outputting the marked battery cell internal resistance;

[0008] When the training end condition is met, an internal resistance prediction model for predicting the internal resistance of the battery is obtained.

[0009] Optionally, the internal resistance prediction model is a model using a support vector regression SVR algorithm, and parameter adjustment is performed on the current internal resistance prediction model, including:

[0010] Adjust the first parameter in the radial basis kernel function and the second parameter in the loss function in the SVR algorithm;

[0011] The first parameter is a bandwidth parameter γ, the second parameter is a penalty coefficient C, and the loss function is a function related to the difference between the internal resistance predicted by the prediction model and the marked internal resistance of the battery cell.

[0012] Optionally, adjust the parameters of the current prediction model, including:

[0013] According to the gray wolf GWO algorithm, a preset number of gray wolves are randomly generated and an iterative update is triggered; wherein the position of each gray wolf represents a set of parameters in the current internal resistance prediction model, the set of parameters including the bandwidth parameter γ and the penalty coefficient C;

[0014] When triggering an iterative update, the fitness of each gray wolf is calculated according to the position of each gray wolf, and the α wolf, β wolf, δ wolf and w wolf are determined according to the fitness, and the position of the prey is determined according to the positions of the α wolf, β wolf and δ wolf; wherein the α wolf is the leader, the β wolf and the δ wolf are the individuals with the second best fitness in the wolf pack, the w wolf is the other individuals in the wolf pack, and the position of the prey is the value of the bandwidth parameter γ and the penalty coefficient C;

[0015] Under the leadership of the alpha wolf, the position of each gray wolf is updated according to the location of the prey, and the next iterative update is triggered when the number of iterations does not reach the maximum value;

[0016] When the number of iterations reaches a maximum value, the current position of the prey is determined according to the current positions of the α wolf, the β wolf, and the δ wolf, and the current position of the prey is determined to be the optimal solution of the bandwidth parameter γ and the penalty coefficient C.

[0017] Optionally, the samples in the training set are generated in the following manner:

[0018] At every preset time interval, the internal resistance of the battery cell is calculated according to the DC detection method, and the battery cell data of the battery, the battery pack data where the battery cell is located, and the environmental data are collected accordingly;

[0019] The battery cell data, battery pack data where the battery cell is located, environmental data and battery cell internal resistance collected at different times are used as different samples.

[0020] Optionally, the battery cell data includes a cell voltage Cell V and monomer current I;

[0021] The battery pack data of the battery cell includes the battery pack voltage Bank v ;

[0022] The environmental data includes the environmental temperature T.

[0023] Optionally, the training end conditions are met, including:

[0024] Acquire multiple test samples including measured battery cell data, battery pack data where the battery cell is located, environmental data, and annotated battery cell internal resistance;

[0025] Inputting the battery cell data, the battery pack data where the battery cell is located, and the environmental data in the test sample into the prediction model, and outputting the predicted internal resistance;

[0026] The predicted internal resistance is compared with the marked internal resistance of the battery cell, and if it is determined that the error is within a preset range, the training of the prediction model is terminated.

[0027] In a second aspect, the present application provides a fault prediction method, comprising:

[0028] In response to a battery internal resistance prediction request, obtaining battery cell data, battery pack data where the battery cell is located, and environmental data;

[0029] The battery cell data, the battery pack data where the battery cell is located, and the environmental data are input into the internal resistance prediction model trained by the method described in any one of the first aspects to obtain the predicted internal resistance.

[0030] Optionally, also include:

[0031] comparing the predicted internal resistance with a standard threshold internal resistance of the battery;

[0032] When it is determined that the predicted internal resistance is greater than the standard threshold internal resistance of the battery, an alarm is issued; when it is determined that the predicted internal resistance is not greater than the standard threshold internal resistance of the battery, no alarm is issued.

[0033] In a third aspect, the present application provides an internal resistance prediction model training device, comprising:

[0034] A training set acquisition module, used to acquire a training set including a plurality of samples, wherein the samples include historically measured battery cell data, battery pack data where the battery cell is located, environmental data, and annotated battery cell internal resistance;

[0035] A parameter adjustment module, used for inputting the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model to be trained, and adjusting the parameters of the current internal resistance prediction model with the goal of outputting the marked battery cell internal resistance;

[0036] The training end module is used to obtain an internal resistance prediction model for predicting the internal resistance of the battery when the training end condition is met.

[0037] In a fourth aspect, the present application provides a fault prediction device, comprising:

[0038] A response module, used to respond to a battery internal resistance prediction request and obtain battery cell data, battery pack data where the battery cell is located, and environmental data;

[0039] The acquisition module is used to input the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model trained by any method described in the first aspect to obtain the predicted internal resistance.

[0040] In a fifth aspect, the present application provides an internal resistance prediction model training device, comprising:

[0041] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the method for training an internal resistance prediction model as described in any one of the first aspects.

[0042] In a sixth aspect, the present application provides a fault prediction device, comprising:

[0043] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform a fault prediction method as described in any one of the second aspects.

[0044] According to an internal resistance prediction model training and fault prediction method, device, and equipment provided in the present application, the internal resistance of a battery can be predicted online without damaging the battery. Based on the predicted internal resistance, whether the battery has a fault can be detected to ensure the safe operation of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application.

[0046] Figure 1 A flow chart is provided according to an embodiment of the present application;

[0047] Figure 2 A flow chart is provided according to an embodiment of the present application;

[0048] Figure 3 A flow chart is provided according to an embodiment of the present application;

[0049] Figure 4 A flow chart is provided according to an embodiment of the present application;

[0050] Figure 5 A schematic diagram of a device provided according to an embodiment of the present application;

[0051] Figure 6 A schematic diagram of a device provided according to an embodiment of the present application;

[0052] Figure 7 A schematic diagram of a device provided according to an embodiment of the present application;

[0053] Figure 8 A schematic diagram of a device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the technical solution of the present application, rather than all of the embodiments. Based on the embodiments recorded in the application documents, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the technical solution of the present application.

[0055] Batteries are very important sources of energy in data center rooms. The health of lead-acid batteries as backup power sources plays an important role in the safe and stable operation of the system. If a lead-acid battery fails and cannot supply power to the operating equipment in time, the system will be shut down, resulting in huge economic losses. In data center rooms, lead-acid batteries exist in groups, and the power supply capacity of the battery group is closely related to the available capacity of the outdated battery. Therefore, it is very important to detect and handle failed batteries in a timely manner. According to relevant information, internal resistance is an important parameter to characterize whether a battery has failed. According to the IEEE1188 standard, when the internal resistance of a battery is 1.3 times its initial value, the capacity of the battery may be less than 80%. Therefore, it is possible to determine whether the battery needs to be replaced based on the internal resistance of the battery.

[0056] In the related art, there are a variety of internal resistance measurement methods, such as AC discharge method and DC discharge method. The online system is not suitable for measuring the internal resistance of the battery using the AC discharge method, which is easily interfered with, resulting in inaccurate measurement results. The online system generally uses the DC discharge method to measure the internal resistance of the battery, but this method causes certain damage to the battery and reduces the battery life. Therefore, the present application provides an internal resistance prediction model training and fault prediction method, device, and equipment, which can predict the internal resistance of the battery online without damaging the battery. According to the predicted internal resistance, it is possible to detect whether the battery has a fault, so as to ensure the safe operation of the system.

[0057] Example 1

[0058] This application provides a method for training an internal resistance prediction model, such as Figure 1 As shown, including:

[0059] Step 101, obtaining a training set including a plurality of samples, wherein the samples include historically measured battery cell data, battery pack data where the battery cell is located, environmental data, and annotated battery cell internal resistance;

[0060] Optionally, the above-mentioned marked battery cell internal resistance is obtained by using the DC detection method in the existing related technology; at every preset time interval, the battery cell data is calculated according to the above-mentioned DC detection method, and the battery cell data of the battery, the battery pack data where the battery cell is located, and the environmental data are collected accordingly, and the battery cell data, the battery pack data where the battery cell is located, the environmental data and the battery cell internal resistance collected at different times are used as different samples, and the obtained multiple different samples are used as training sets; wherein, the battery cell data includes the cell voltage Cell V and single cell current I; the battery pack data where the battery cell is located includes the battery pack voltage Bank v ; The environmental data includes ambient temperature T.

[0061] Step 102, inputting the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model to be trained, and adjusting the parameters of the current internal resistance prediction model with the goal of outputting the marked battery cell internal resistance;

[0062] Optionally, since the amount of data of single cells is small, the above internal resistance prediction model is constructed using a support vector regression SVR (support vector regression) algorithm. The SVR algorithm can solve machine learning problems under small samples, handle the interaction of nonlinear features, and has strong generalization ability. Therefore, the single cell voltage CellV, single cell current I, battery pack voltage Bankv, and ambient temperature T are input into the internal resistance prediction model to be trained constructed using the SVR algorithm, and the parameters of the current internal resistance prediction model are adjusted with the goal of outputting the labeled battery cell internal resistance. When dealing with nonlinear problems, the radial basis kernel function has excellent performance, so the following kernel function is used to construct the above internal resistance prediction model:

[0063] K(x i , x) = exp(-γ‖x i -x‖ 2 )

[0064] Among them, γ is the bandwidth parameter, which is used to control the radial range of the function; x is the input feature vector, including the single cell voltage CellV, single cell current I, battery pack voltage Bankv, and ambient temperature T, so x = (Cell V ,I,Bank v , T); x i is the center of the kernel function.

[0065] According to the above kernel function K(x i , x) construct the discriminant function f(x):

[0066]

[0067] Among them, ω is the normal vector of the classification hyperplane in the high-dimensional space, and b is a constant term.

[0068] Alternatively, the SVR algorithm is simply understood as creating a "gap" on both sides of the linear function, and assuming that SVR can tolerate a deviation of ∈ between f(x) and y, and only calculates the loss when the deviation between f(x) and y is greater than ∈; construct the final function F(x) based on the above discriminant function f(x) and the loss function:

[0069]

[0070] Among them, C is the penalty coefficient, is an insensitive loss function. Specifically,

[0071]

[0072] f(x i ) is the corresponding battery internal resistance output according to the single cell voltage CellV, single cell current I, battery pack voltage Bankv and ambient temperature T in a sample, y i It is the internal resistance of the battery cell marked in this sample.

[0073] Step 103, when the training end condition is met, an internal resistance prediction model for predicting the internal resistance of the battery is obtained.

[0074] Optionally, meet the training end conditions, such as Figure 2 As shown, including:

[0075] Step 201, obtaining a plurality of test samples including measured battery cell data, battery pack data where the battery cell is located, environmental data and annotated battery cell internal resistance;

[0076] Step 202, inputting the battery cell data in the test sample, the battery pack data where the battery cell is located, and the environmental data into the prediction model, and outputting the predicted internal resistance;

[0077] Step 203 , comparing the predicted internal resistance with the marked battery cell internal resistance, and determining that the error is within a preset range, then the prediction model training ends.

[0078] Optionally, the internal resistance prediction model is a model that uses a support vector regression SVR algorithm, and parameters of the current internal resistance prediction model are adjusted, including: adjusting a first parameter in a radial basis kernel function in the SVR algorithm and a second parameter in a loss function; the first parameter is a bandwidth parameter γ, the second parameter is a penalty coefficient C, and the loss function is a function related to the difference between the internal resistance predicted by the prediction model and the marked internal resistance of the battery cell.

[0079] Optionally, adjust the parameters of the current prediction model, such as Figure 3 As shown, specifically including:

[0080] Step 301, randomly generating a preset number of gray wolves according to the gray wolf GWO algorithm, and triggering an iterative update; wherein the position of each gray wolf represents a set of parameters in the current internal resistance prediction model, the set of parameters including the bandwidth parameter γ and the penalty coefficient C;

[0081] Optionally, according to the gray wolf GWO algorithm, a preset number of gray wolves are randomly generated, and the position of each gray wolf represents a set of parameters in the current internal resistance prediction model, wherein the set of parameters includes the bandwidth parameter γ and the penalty coefficient C. Therefore, the objective function can be expressed as f(C, γ).

[0082] Step 302, when triggering iterative update, the fitness of each gray wolf is calculated according to the position of each gray wolf, the α wolf, the β wolf, the δ wolf and the w wolf are determined according to the fitness, and the position of the prey is determined according to the positions of the α wolf, the β wolf and the δ wolf; wherein the α wolf is the leader, the β wolf and the δ wolf are the individuals with the second best fitness in the wolf pack, the w wolf is the other individuals in the wolf pack, and the position of the prey is the value of the bandwidth parameter γ and the penalty coefficient C;

[0083] Optionally, in the SVR algorithm, the objective function is usually a measure of the prediction error, such as the mean absolute error or the mean square error; the objective function of the SVR algorithm in the present application is f(C, γ), and the fitness function value of each gray wolf in the gray wolf group can be expressed as:

[0084]

[0085] According to the calculated fitness of each gray wolf, the gray wolf with the best fitness is determined to be α wolf, the gray wolves with the second and third best fitness are determined to be β wolf and δ wolf respectively, and then the remaining gray wolves in the wolf pack are classified as w wolves; the α wolf is the leader, plays the role of decision-making and management of the wolf pack, and will lead the β wolf and δ wolf to hunt prey. The β wolf and δ wolf are the individuals with the second best fitness in the wolf pack. The w wolves are other individuals in the wolf pack. According to the current positions of the α wolf, β wolf, and δ wolf, the position of the prey is determined, that is, the current optimal solution of the bandwidth parameter γ and the penalty coefficient C is found.

[0086] Step 303, under the leadership of the alpha wolf, according to the location X of the prey P Update the position of each gray wolf, and trigger the next iterative update when the number of iterations does not reach the maximum value;

[0087] Optionally, under the leadership of the α wolf, the position of the prey will change continuously, and according to the change of the prey, the position of the gray wolf will also change; first, determine the position of the prey after the t-th iteration calculation as X P (t), the position of the gray wolf after the tth iteration is X(t); the distance between the gray wolf and the prey is calculated as:

[0088] D=|E*X P (t)-X(t)|

[0089] And according to the distance D between the gray wolf and the prey, the position of the gray wolf after the t+1th iteration is calculated as:

[0090] X(t+1)=X P (t)-A*D

[0091] Among them, the above E and A are coefficient adjustment factors, A = 2a*r1-a, E = 2*r2, where r1 and r2 are random numbers in the interval [0, 1], and a is a convergence factor. As the iteration proceeds, the value of a will decrease linearly from 2 to 0; specifically, the position coordinates of the prey are determined according to the coordinate updates of α wolf, β wolf, and δ wolf, and the process strategy is:

[0092]

[0093]

[0094]

[0095] Among them, X α (t), X β (t), X δ (t) represents the current positions of wolf α, wolf β, and wolf δ, respectively; X1, X2, and X3 represent the updated positions of wolf α, wolf β, and wolf δ, respectively; the new position of the prey is determined according to the updated positions of wolf α, wolf β, and wolf δ and is expressed as It is determined to be the current optimal solution of the bandwidth parameter γ and the penalty coefficient C.

[0096] Step 304, when the number of iterations reaches a maximum value, the current position of the prey is determined according to the current positions of the α wolf, the β wolf, and the δ wolf, and the current position of the prey is determined to be the optimal solution of the bandwidth parameter γ and the penalty coefficient C.

[0097] Optionally, when it is determined that the number of iterations reaches a maximum value, the iteration is stopped, and the current position of the prey is determined based on the current positions of the α wolf, β wolf, and δ wolf. The position of the prey is the optimal solution of the bandwidth parameter γ and the penalty coefficient C obtained by optimizing the gray wolf algorithm, which is the best hyperparameter combination.

[0098] According to an internal resistance prediction model training method provided in the present application, the internal resistance of a battery can be predicted online without damaging the battery, and the predicted internal resistance can be obtained.

[0099] Optionally, the samples in the training set are generated in the following manner: at every preset time interval, the internal resistance of the battery cell is calculated according to the DC detection method, and the battery cell data of the battery, the battery pack data where the battery cell is located, and the environmental data are collected accordingly; for example, the battery internal resistance is collected once every 7 days, and the collection time is 00:00:00 every Monday. The following Table 1 is the result of specific data collection:

[0100] Table 1

[0101] Collection time <![CDATA[Cell v (in)]]> R(uΩ) <![CDATA[Bank V (in)]]> I(A) T(℃) 11 / 20 0:00:00 2.24 874 536.61 -2.22 23.90 11 / 26 0:00:00 2.24 863 536.47 -2.30 24.10 12 / 3 0:00:00 2.23 868 536.35 -2.12 23.98 12 / 10 0:00:00 2.24 866 536.61 -2.12 23.99 12 / 17 0:00:00 2.24 861 536.59 -2.16 23.75

[0102] The battery cell data, battery pack data where the battery cell is located, environmental data and battery cell internal resistance collected at different times are used as different samples; and as many samples as possible are collected. The more samples there are, the more accurate the predicted internal resistance value will be.

[0103] Example 2

[0104] The present application also provides a fault prediction method, such as Figure 4 As shown, including:

[0105] Step 401, in response to a battery internal resistance prediction request, obtaining battery cell data, battery pack data where the battery cell is located, and environmental data;

[0106] Optionally, the above-mentioned battery cell data, battery pack data where the battery cell is located, and environmental data can be acquired by real-time monitoring by an existing dynamic environment detection system.

[0107] Step 402 , inputting the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model trained by the method described in any one of Embodiment 1 to obtain the predicted internal resistance.

[0108] Step 403, comparing the predicted internal resistance with a standard threshold internal resistance of the battery;

[0109] Step 404: an alarm is issued when it is determined that the predicted internal resistance is greater than a standard threshold internal resistance of the battery, and no alarm is issued when it is determined that the predicted internal resistance is not greater than the standard threshold internal resistance of the battery.

[0110] Optionally, the predicted internal resistance obtained according to the internal resistance prediction model needs to be compared with the threshold internal resistance of the same model specified by the national standard. When it is determined that the predicted internal resistance is greater than the standard threshold internal resistance of the battery, an alarm is issued. When it is determined that the predicted internal resistance is not greater than the standard threshold internal resistance of the battery, no alarm is issued. For details, please refer to Table 2:

[0111] Table 2

[0112]

[0113] Example 3

[0114] The present application also provides a device for training an internal resistance prediction model. Figure 5 As shown, including:

[0115] A training set acquisition module 501 is used to acquire a training set including a plurality of samples, wherein the samples include historically measured battery cell data, battery pack data where the battery cell is located, environmental data, and annotated battery cell internal resistance;

[0116] A parameter adjustment module 502 is used to input the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model to be trained, and adjust the parameters of the current internal resistance prediction model with the goal of outputting the marked battery cell internal resistance;

[0117] The training end module 503 is used to obtain an internal resistance prediction model for predicting the internal resistance of the battery when the training end condition is met.

[0118] Example 4

[0119] The present application also provides a fault prediction device, such as Figure 6 As shown, including:

[0120] A response module 601 is used to obtain battery cell data, battery pack data where the battery cell is located, and environmental data in response to a battery internal resistance prediction request;

[0121] The acquisition module 602 is used to input the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model trained by the method described in any one of the embodiments 1 to obtain the predicted internal resistance.

[0122] Example 5

[0123] The present application also provides an internal resistance prediction model training device, comprising:

[0124] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for training an internal resistance prediction model as described in any one of Example 1.

[0125] like Figure 7 As shown, the device includes a processor 701, a memory 702, a communication interface 703 and a bus 704. The processor 701, the memory 702 and the communication interface 703 are connected to each other via the bus 704.

[0126] The processor 701 is used to read and execute instructions in the memory 702, so that at least one processor can execute the method for training the internal resistance prediction model provided in the above-mentioned embodiment 1.

[0127] The memory 702 is used to store various instructions and programs of the method for training the internal resistance prediction model provided in the above-mentioned embodiment 1.

[0128] The bus 704 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0129] The processor 701 may be a central processing unit (CPU), a network processor (NP), a graphic processing unit (GPU), or any combination of CPU, NP, and GPU. It may also be a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0130] Example 6

[0131] The present application also provides a fault prediction device, comprising:

[0132] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the fault prediction method as described in any one of Example 2.

[0133] like Figure 8 As shown, the device includes a processor 801, a memory 802, a communication interface 803 and a bus 804. The processor 801, the memory 802 and the communication interface 803 are connected to each other via the bus 804.

[0134] The processor 801 is used to read and execute instructions in the memory 802, so that at least one processor can execute the fault prediction method provided in the above-mentioned embodiment 2.

[0135] The memory 802 is used to store various instructions and programs of the fault prediction method provided in the second embodiment.

[0136] The bus 804 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0137] The processor 801 may be a central processing unit (CPU), a network processor (NP), a graphic processing unit (GPU), or any combination of CPU, NP, and GPU. It may also be a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0138] The present application provides an internal resistance prediction model training and fault prediction method, device, and equipment, which can predict the internal resistance of a battery online without damaging the battery. Based on the predicted internal resistance, it is possible to detect whether the battery has a fault, thereby ensuring the safe operation of the system.

[0139] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0140] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0141] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0142] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0143] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for training an internal resistance prediction model, characterized in that: include: Acquire a training set including a plurality of samples, wherein the samples include historically measured battery cell data, battery pack data where the battery cell is located, environmental data, and annotated battery cell internal resistance; Inputting the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model to be trained, and adjusting the parameters of the current internal resistance prediction model with the goal of outputting the marked battery cell internal resistance; When the training end condition is met, an internal resistance prediction model for predicting the internal resistance of the battery is obtained.

2. The method according to claim 1, characterized in that The internal resistance prediction model is a model using a support vector regression SVR algorithm, and the parameters of the current internal resistance prediction model are adjusted, including: Adjust the first parameter in the radial basis kernel function and the second parameter in the loss function in the SVR algorithm; The first parameter is a bandwidth parameter γ, the second parameter is a penalty coefficient C, and the loss function is a function related to the difference between the internal resistance predicted by the prediction model and the marked internal resistance of the battery cell.

3. The method according to claim 2, characterized in that Adjust the parameters of the current prediction model, including: According to the gray wolf GWO algorithm, a preset number of gray wolves are randomly generated and an iterative update is triggered; wherein the position of each gray wolf represents a set of parameters in the current internal resistance prediction model, the set of parameters including the bandwidth parameter γ and the penalty coefficient C; When triggering an iterative update, the fitness of each gray wolf is calculated according to the position of each gray wolf, and the α wolf, β wolf, δ wolf and w wolf are determined according to the fitness, and the position of the prey is determined according to the positions of the α wolf, β wolf and δ wolf; wherein the α wolf is the leader, the β wolf and the δ wolf are the individuals with the second best fitness in the wolf pack, the w wolf is the other individuals in the wolf pack, and the position of the prey is the value of the bandwidth parameter γ and the penalty coefficient C; Under the leadership of the alpha wolf, the position of each gray wolf is updated according to the location of the prey, and the next iterative update is triggered when the number of iterations does not reach the maximum value; When the number of iterations reaches a maximum value, the current position of the prey is determined according to the current positions of the α wolf, the β wolf, and the δ wolf, and the current position of the prey is determined to be the optimal solution of the bandwidth parameter γ and the penalty coefficient C.

4. The method according to claim 1, characterized in that: The samples in the training set are generated in the following way: At every preset time interval, the internal resistance of the battery cell is calculated according to the DC detection method, and the battery cell data of the battery, the battery pack data where the battery cell is located, and the environmental data are collected accordingly; The battery cell data, battery pack data where the battery cell is located, environmental data and battery cell internal resistance collected at different times are used as different samples.

5. The method according to claim 1, characterized in that The battery cell data includes the cell voltage Cell V and monomer current I; The battery pack data of the battery cell includes the battery pack voltage Bank v ; The environmental data includes the environmental temperature T.

6. The method according to claim 1, characterized in that The training end conditions are met, including: Acquire multiple test samples including measured battery cell data, battery pack data where the battery cell is located, environmental data, and annotated battery cell internal resistance; Inputting the battery cell data, the battery pack data where the battery cell is located, and the environmental data in the test sample into the prediction model, and outputting the predicted internal resistance; The predicted internal resistance is compared with the marked internal resistance of the battery cell, and if it is determined that the error is within a preset range, the training of the prediction model is terminated.

7. A fault prediction method, characterized in that: include: In response to a battery internal resistance prediction request, obtaining battery cell data, battery pack data where the battery cell is located, and environmental data; The battery cell data, the battery pack data where the battery cell is located, and the environmental data are input into the internal resistance prediction model trained by the method described in any one of claims 1 to 6 to obtain the predicted internal resistance.

8. The method according to claim 7, characterized in that Also includes: comparing the predicted internal resistance with a standard threshold internal resistance of the battery; When it is determined that the predicted internal resistance is greater than the standard threshold internal resistance of the battery, an alarm is issued; when it is determined that the predicted internal resistance is not greater than the standard threshold internal resistance of the battery, no alarm is issued.

9. An internal resistance prediction model training device, characterized in that: include: A training set acquisition module, used to acquire a training set including a plurality of samples, wherein the samples include historically measured battery cell data, battery pack data where the battery cell is located, environmental data, and annotated battery cell internal resistance; A parameter adjustment module, used for inputting the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model to be trained, and adjusting the parameters of the current internal resistance prediction model with the goal of outputting the marked battery cell internal resistance; The training end module is used to obtain an internal resistance prediction model for predicting the internal resistance of the battery when the training end condition is met.

10. A fault prediction device, characterized in that: include: A response module, used to respond to a battery internal resistance prediction request and obtain battery cell data, battery pack data where the battery cell is located, and environmental data; The acquisition module is used to input the battery cell data, the battery pack data where the battery cell is located, and the environmental data into the internal resistance prediction model trained by the method described in any one of claims 1 to 6 to obtain the predicted internal resistance.

11. An internal resistance prediction model training device, characterized in that: include: at least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for training an internal resistance prediction model as described in any one of claims 1-6.

12. A fault prediction device, characterized in that: include: at least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the fault prediction method as described in any one of claims 7-8.